538 lines
13 KiB
C++
Raw Normal View History

2015-08-01 13:42:27 -07:00
/*
* Copyright (C) 2014 Jared Boone, ShareBrained Technology, Inc.
*
* This file is part of PortaPack.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2, or (at your option)
* any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; see the file COPYING. If not, write to
* the Free Software Foundation, Inc., 51 Franklin Street,
* Boston, MA 02110-1301, USA.
*/
#include "portapack.hpp"
#include "portapack_hal.hpp"
#include "portapack_dma.hpp"
#include "portapack_cpld_data.hpp"
#include "portapack_persistent_memory.hpp"
2015-08-01 13:42:27 -07:00
#include "hackrf_hal.hpp"
#include "hackrf_gpio.hpp"
using namespace hackrf::one;
2022-04-06 08:16:57 +12:00
2015-08-01 13:42:27 -07:00
#include "clock_manager.hpp"
#include "event_m0.hpp"
2015-08-01 13:42:27 -07:00
#include "backlight.hpp"
2015-08-01 13:42:27 -07:00
#include "touch_adc.hpp"
#include "audio.hpp"
2015-08-01 13:42:27 -07:00
2017-05-31 11:42:12 -07:00
#include "wm8731.hpp"
using wolfson::wm8731::WM8731;
#include "ak4951.hpp"
using asahi_kasei::ak4951::AK4951;
#include "cpld_update.hpp"
#include "optional.hpp"
2022-04-06 08:16:57 +12:00
#include "irq_controls.hpp"
#include "file.hpp"
#include "sd_card.hpp"
#include "string_format.hpp"
2015-08-01 13:42:27 -07:00
namespace portapack {
portapack::IO io {
portapack::gpio_dir,
portapack::gpio_lcd_rdx,
portapack::gpio_lcd_wrx,
2015-08-01 13:42:27 -07:00
portapack::gpio_io_stbx,
portapack::gpio_addr,
portapack::gpio_lcd_te,
2023-04-16 18:23:54 +02:00
portapack::gpio_dfu,
2015-08-01 13:42:27 -07:00
};
portapack::BacklightCAT4004 backlight_cat4004;
portapack::BacklightOnOff backlight_on_off;
lcd::ILI9341 display;
2015-08-01 13:42:27 -07:00
I2C i2c0(&I2CD0);
SPI ssp1(&SPID2);
si5351::Si5351 clock_generator {
i2c0, hackrf::one::si5351_i2c_address
};
ClockManager clock_manager {
i2c0, clock_generator
};
2017-05-31 11:42:12 -07:00
WM8731 audio_codec_wm8731 { i2c0, 0x1a };
AK4951 audio_codec_ak4951 { i2c0, 0x12 };
2017-05-31 11:42:12 -07:00
ReceiverModel receiver_model;
TransmitterModel transmitter_model;
2015-08-01 13:42:27 -07:00
TemperatureLogger temperature_logger;
2015-08-01 13:42:27 -07:00
bool antenna_bias { false };
uint32_t bl_tick_counter { 0 };
void set_antenna_bias(const bool v) {
antenna_bias = v;
}
2015-08-23 05:08:38 +02:00
bool get_antenna_bias() {
return antenna_bias;
}
2020-06-08 01:23:44 +02:00
bool speaker_mode { false };
void set_speaker_mode(const bool v) {
speaker_mode = v;
if (speaker_mode)
audio::output::speaker_unmute();
else
audio::output::speaker_mute();
}
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
static constexpr uint32_t systick_count(const uint32_t clock_source_f) {
return clock_source_f / CH_FREQUENCY;
}
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
static constexpr uint32_t systick_load(const uint32_t clock_source_f) {
return systick_count(clock_source_f) - 1;
}
Upstream merge to make new revision of PortaPack work (#206) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Fix merge conflicts
2019-01-11 07:56:21 +01:00
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
constexpr uint32_t i2c0_bus_f = 400000;
constexpr uint32_t i2c0_high_period_ns = 900;
typedef struct {
uint32_t clock_f;
uint32_t systick_count;
uint32_t idivb;
uint32_t idivc;
} clock_config_t;
2015-08-01 13:42:27 -07:00
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
static constexpr uint32_t idiv_config(const cgu::CLK_SEL clk_sel, const uint32_t idiv) {
return cgu::IDIV_CTRL { 0, idiv-1, 1, clk_sel };
Upstream merge to make new revision of PortaPack work (#206) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Fix merge conflicts
2019-01-11 07:56:21 +01:00
}
2015-08-01 13:42:27 -07:00
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
constexpr clock_config_t clock_config_irc {
12000000, systick_load(12000000),
idiv_config(cgu::CLK_SEL::IRC, 1),
idiv_config(cgu::CLK_SEL::IRC, 1),
};
constexpr clock_config_t clock_config_pll1_boot {
96000000, systick_load(96000000),
idiv_config(cgu::CLK_SEL::PLL1, 9),
idiv_config(cgu::CLK_SEL::PLL1, 3),
};
constexpr clock_config_t clock_config_pll1_step {
100000000, systick_load(100000000),
idiv_config(cgu::CLK_SEL::PLL1, 1),
idiv_config(cgu::CLK_SEL::PLL1, 1),
};
constexpr clock_config_t clock_config_pll1 {
200000000, systick_load(200000000),
idiv_config(cgu::CLK_SEL::PLL1, 2),
idiv_config(cgu::CLK_SEL::PLL1, 1),
};
constexpr I2CClockConfig i2c_clock_config_400k_boot_clock {
.clock_source_f = clock_config_pll1_boot.clock_f,
.bus_f = i2c0_bus_f,
.high_period_ns = i2c0_high_period_ns,
};
constexpr I2CClockConfig i2c_clock_config_400k_fast_clock {
.clock_source_f = clock_config_pll1.clock_f,
.bus_f = i2c0_bus_f,
.high_period_ns = i2c0_high_period_ns,
};
constexpr I2CConfig i2c_config_boot_clock {
.high_count = i2c_clock_config_400k_boot_clock.i2c_high_count(),
.low_count = i2c_clock_config_400k_boot_clock.i2c_low_count(),
};
constexpr I2CConfig i2c_config_fast_clock {
.high_count = i2c_clock_config_400k_fast_clock.i2c_high_count(),
.low_count = i2c_clock_config_400k_fast_clock.i2c_low_count(),
};
2015-08-01 13:42:27 -07:00
enum class PortaPackModel {
R1_20150901,
R2_20170522,
};
static bool save_config(int8_t value){
persistent_memory::set_config_cpld(value);
if(sd_card::status() == sd_card::Status::Mounted){
make_new_directory("/hardware");
File file;
auto sucess = file.create("/hardware/settings.txt");
if(!sucess.is_valid()) {
file.write_line(to_string_dec_uint(value));
}
}
return true;
}
int read_file(std::string name) {
std::string return_string = "";
File file;
auto success = file.open(name);
if(!success.is_valid()) {
char one_char[1];
for(size_t pointer = 0; pointer < file.size() ; pointer++) {
file.seek(pointer);
file.read(one_char, 1);
return_string += one_char[0];
}
return std::stoi(return_string);
}
return -1;
}
static int load_config(){
static Optional<int> config_value;
if(!config_value.is_valid()){
int8_t value = portapack::persistent_memory::config_cpld();
if((value <= 0 || value >= 5) && sd_card::status() == sd_card::Status::Mounted){
int data = read_file("/hardware/settings.txt");
if(data != -1) {
config_value = data;
}
} else {
config_value = value;
}
}
return config_value.value();
}
static PortaPackModel portapack_model() {
static Optional<PortaPackModel> model;
if( !model.is_valid() ) {
const auto switches_state = get_switches_state();
if (switches_state[(size_t)ui::KeyEvent::Up]){
save_config(1);
// model = PortaPackModel::R2_20170522; // Commented these out as they should be set down below anyway
}
else if (switches_state[(size_t)ui::KeyEvent::Down]){
save_config(2);
// model = PortaPackModel::R1_20150901;
}
else if (switches_state[(size_t)ui::KeyEvent::Left]){
save_config(3);
// model = PortaPackModel::R1_20150901;
}
else if (switches_state[(size_t)ui::KeyEvent::Right]){
save_config(4);
// model = PortaPackModel::R2_20170522;
}
else if (switches_state[(size_t)ui::KeyEvent::Select]){
save_config(0);
}
if (load_config() == 1) {
model = PortaPackModel::R2_20170522;
} else if (load_config() == 2) {
model = PortaPackModel::R1_20150901;
} else if (load_config() == 3) {
model = PortaPackModel::R1_20150901;
} else if (load_config() == 4) {
model = PortaPackModel::R2_20170522;
2022-04-07 21:30:37 +12:00
} else {
if( audio_codec_wm8731.detected() ) {
model = PortaPackModel::R1_20150901; // H1R1
} else {
model = PortaPackModel::R2_20170522; // H1R2, H2, H2+
}
2022-04-07 21:30:37 +12:00
}
}
return model.value();
}
2022-04-05 14:44:32 +12:00
//audio_codec_wm8731 = H1R1 & H2+
//audio_codec_ak4951 = H1R2
2022-04-05 14:41:12 +12:00
static audio::Codec* portapack_audio_codec() {
/* I2C ready OK, Automatic recognition of audio chip */
return (audio_codec_wm8731.detected())
? static_cast<audio::Codec*>(&audio_codec_wm8731)
: static_cast<audio::Codec*>(&audio_codec_ak4951)
;
}
static const portapack::cpld::Config& portapack_cpld_config() {
return (portapack_model() == PortaPackModel::R2_20170522)
2022-04-06 13:28:10 +12:00
? portapack::cpld::rev_20170522::config
: portapack::cpld::rev_20150901::config;
}
Backlight* backlight() {
return (portapack_model() == PortaPackModel::R2_20170522)
? static_cast<portapack::Backlight*>(&backlight_cat4004) // R2_20170522
: static_cast<portapack::Backlight*>(&backlight_on_off); // R1_20150901
}
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
#define ARRAY_SIZE(a) (sizeof(a)/sizeof(a[0]))
Upstream merge to make new revision of PortaPack work (#206) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Fix merge conflicts
2019-01-11 07:56:21 +01:00
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
static LPC_CGU_BASE_CLK_Type* const base_clocks_idivc[] = {
&LPC_CGU->BASE_PERIPH_CLK,
&LPC_CGU->BASE_M4_CLK,
&LPC_CGU->BASE_APB1_CLK,
&LPC_CGU->BASE_APB3_CLK,
&LPC_CGU->BASE_SDIO_CLK,
&LPC_CGU->BASE_SSP1_CLK,
};
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
static void set_idivc_base_clocks(const cgu::CLK_SEL clock_source) {
for(uint32_t i=0; i<ARRAY_SIZE(base_clocks_idivc); i++) {
base_clocks_idivc[i]->AUTOBLOCK = 1;
base_clocks_idivc[i]->CLK_SEL = toUType(clock_source);
Upstream merge to make new revision of PortaPack work (#206) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Fix merge conflicts
2019-01-11 07:56:21 +01:00
}
}
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
static void set_clock_config(const clock_config_t& config) {
LPC_CGU->IDIVB_CTRL.word = config.idivb;
LPC_CGU->IDIVC_CTRL.word = config.idivc;
systick_adjust_period(config.systick_count);
halLPCSetSystemClock(config.clock_f);
}
static void shutdown_base() {
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
i2c0.stop();
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
set_clock_config(clock_config_irc);
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
cgu::pll1::disable();
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
set_idivc_base_clocks(cgu::CLK_SEL::IRC);
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
cgu::pll1::ctrl({
.pd = 1,
.bypass = 0,
.fbsel = 0,
.direct = 1,
.psel = 0,
.autoblock = 1,
.nsel = 0,
.msel = 23,
.clk_sel = cgu::CLK_SEL::IRC,
});
cgu::pll1::enable();
while( !cgu::pll1::is_locked() );
set_clock_config(clock_config_pll1_boot);
i2c0.start(i2c_config_boot_clock);
clock_manager.shutdown();
}
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
/* Clock scheme after exiting bootloader in SPIFI mode:
*
* XTAL_OSC = powered down
*
* PLL0USB = powered down
* PLL0AUDIO = powered down
* PLL1 = IRC * 24 = 288 MHz
*
* IDIVA = IRC / 1 = 12 MHz
* IDIVB = PLL1 / 9 = 32 MHz
* IDIVC = PLL1 / 3 = 96 MHz
* IDIVD = IRC / 1 = 12 MHz
* IDIVE = IRC / 1 = 12 MHz
*
* BASE_USB0_CLK = PLL0USB
* BASE_PERIPH_CLK = IRC
* BASE_M4_CLK = IDIVC (96 MHz)
* BASE_SPIFI_CLK = IDIVB (32 MHZ)
*
* everything else = IRC
*/
/* Clock scheme during PortaPack operation:
*
* XTAL_OSC = powered down
*
* PLL0USB = powered down
* PLL0AUDIO = GP_CLKIN, Fcco=491.52 MHz, Fout=12.288 MHz
* PLL1 =
* OG: GP_CLKIN * 10 = 200 MHz
* r9: GP_CLKIN * 20 = 200 MHz
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
* IDIVA = IRC / 1 = 12 MHz
* IDIVB = PLL1 / 2 = 100 MHz
* IDIVC = PLL1 / 1 = 200 MHz
* IDIVD = PLL0AUDIO / N (where N is varied depending on decimation factor)
* IDIVE = IRC / 1 = 12 MHz
*
* BASE_USB0_CLK = PLL0USB
* BASE_PERIPH_CLK = IRC
* BASE_M4_CLK = IDIVC (200 MHz)
* BASE_SPIFI_CLK = IDIVB (100 MHZ)
* BASE_AUDIO_CLK = IDIVD
*
* everything else = IRC
*/
bool init() {
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
set_idivc_base_clocks(cgu::CLK_SEL::IDIVC);
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
i2c0.start(i2c_config_boot_clock);
2015-08-01 13:42:27 -07:00
// Keeping this here for now incase we need to revert
// if( !portapack::cpld::update_if_necessary(portapack_cpld_config()) ) {
// shutdown_base();
// return false;
// }
// if( !hackrf::cpld::load_sram() ) {
// chSysHalt();
// }
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(100);
Sync up recent portapack-hackrf changes. (#229) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Setup: Format clock reference frequency in MHz, not Hz. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere. * Pin config: VREGMODE=1, add other pins for completeness, comment detail * Pin setup: More useful comments. * Pin setup: Change some defaults, only set up PortaPack pins if detected. * Pin setup: Disable LPC pull-ups on PP CPLD data bus, as CPLD is pulling up. * Baseband: Allow larger HackRF firmware image. * HackRF: Remove USER_INTERFACE CMake variable. * CPLD: Make use of HackRF CPLD tool to generate code.
2019-03-11 22:24:18 -07:00
configure_pins_portapack();
2017-05-02 06:45:23 +01:00
portapack::io.init();
Upstream merge to make new revision of PortaPack work (#206) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Fix merge conflicts
2019-01-11 07:56:21 +01:00
clock_manager.init_clock_generator();
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
i2c0.stop();
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(10);
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
set_clock_config(clock_config_irc);
cgu::pll1::disable();
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
/* Incantation from LPC43xx UM10503 section 12.2.1.1, to bring the M4
* core clock speed to the 110 - 204MHz range.
*/
/* Step into the 90-110MHz M4 clock range */
/* OG:
* Fclkin = 40M
* /N=2 = 20M = PFDin
* Fcco = PFDin * (M=10) = 200M
* r9:
* Fclkin = 10M
* /N=1 = 10M = PFDin
* Fcco = PFDin * (M=20) = 200M
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
* Fclk = Fcco / (2*(P=1)) = 100M
*/
cgu::pll1::ctrl({
.pd = 1,
.bypass = 0,
.fbsel = 0,
.direct = 0,
.psel = 0,
.autoblock = 1,
.nsel = hackrf_r9 ? 0UL : 1UL,
.msel = hackrf_r9 ? 19UL : 9UL,
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
.clk_sel = cgu::CLK_SEL::GP_CLKIN,
});
cgu::pll1::enable();
while( !cgu::pll1::is_locked() );
set_clock_config(clock_config_pll1_step);
/* Delay >50us at 90-110MHz clock speed */
volatile uint32_t delay = 1400;
while(delay--);
set_clock_config(clock_config_pll1);
/* Remove /2P divider from PLL1 output to achieve full speed */
cgu::pll1::direct();
i2c0.start(i2c_config_fast_clock);
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(10);
PortaPack Sync, take 2 (#215) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * PCB: Change PCB stackup, Tg, clarify solder mask color, use more metric. * PCB: Move HackRF header P9 to B.CrtYd layer. * PCB: Change a Tg reference I missed. * PCB: Update footprints for parts with mismatched CAD->tape rotation. Adjust a few layer choice and line thickness bits. * PCB: Got cold feet, switched back to rectangular pads. * PCB: Add Eco layers to be visible and Gerber output. * PCB: Use aux origin for plotting, for tidier coordinates. * PCB: Output Gerber job file, because why not? * Schematic: Correct footprints for two reference-related components. * Schematic: Remove manfuacturer and part number for DNP component. * Schematic: Specify resistor value, manufacturer, part number for reference oscillator series termination. * PCB: Update netlist and footprints from schematic. * Netlist: Updated component values, footprints. * PCB: Nudge some components and traces to address DRC clearance violations. * PCB: Allow KiCad to update zone timestamps (again?!). * PCB: Generate *all* Gerber layers. * Schematic, PCB: Update revision to 20181025. * PCB: Adjust fab layer annotations orientation and font size. * PCB: Hide mounting hole reference designators on silk layer. * PCB: Shrink U1, U3 pads to get 0.2mm space between pads. * PCB: Set pad-to-mask clearance to zero, leave up to fab. Set minimum mask web to 0.2mm for non-black options. * PCB: Revise U1 pad shape, mask, paste, thermal drills. Clearance is improved at corner pads. * PCB: Tweak U3 for better thermal pad/drill/mask/paste design. * PCB: Change solder mask color to blue. * Schematic, PCB: Update revision to 20181029. * PCB: Bump minimum mask web down a tiny bit because KiCad is having trouble with math. * Update schematic * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Clock Manager: Actually store chosen clock reference Similarly-named local was covering a member and discarding the value. * Clock Manager: Reference type which contains source, frequency. * Setup: Display reference source, frequency in frequency correction screen. * LPC43xx API: Add extern "C" for use from C++. * Use LPC43xx API for SGPIO, GPDMA, I2S initialization. * I2S: Add BASE_AUDIO_CLK management. * Add MOTOCON_PWM clock/reset structure. * Serial: Fix dumb typos. * Serial: Remove extra reference operator. * Serial: Cut-and-paste error in structure type name. * Move SCU structure from PAL to LPC43xx API. It'd be nice if I gave some thought to where code should live before I commit it. * VAA power: Move code to HackRF board file It doesn't belong in PAL. * MAX5 CPLD: Add SAMPLE and EXTEST methods. * Flash image: Change packing scheme to use flash more efficiently. Application is now a single image for both M4 bootstrap and M0. Baseband images come immediately after application binary. No need to align to large blocks (and waste lots of flash). * Clock Manager: Remove PLL1 power down function. * Move and rename peripherals reset function to board module. * Remove unused peripheral/clock management. * Clock Manager: Extract switch to IRC into separate function. * Clock Manager: More explicit shutdown of clocks, clock generator. * Move initialization to board module. * ChibiOS: Rename "application" board, add "baseband" board. There are now two ChibiOS "boards", one which runs the application and does the hardware setup. The other board, "baseband", does very little setup. * Clock Manager: Remove unused crystal enable/disable code. * Clock Manager: Restore clock configuration to SPIFI bootloader state before app shutdown. * Reset peripherals on app shutdown. Be careful not to reset M0APP (the core we're running on) or GPIO (which is holding the hardware in a stable state). * M4/baseband hal_lld_init: use IDIVA, which is configured earlier by M0. This was causing problems during restart into HackRF mode. Baseband hal_lld_init changed M4 clock from IDIVA (set by M0) to PLL1, which was unceremoniously turned off during shutdown. * Audio app: Stop audio PLL on shutdown. * M4 HAL: Make LPC43XX_M4_CLK_SRC optional. This was changing the BASE_M4_CLK when a baseband was run. * LPC43xx C++ layer: Fix IDIVx constructor IDIV narrow field width. * Application board: hide the peripherals_reset function, as it isn't useful except during hardware init. * Consolidate hardware init code to some degree. ClockManager is super-overloaded and murky in its purpose. Migrate audio from IDIVC to IDIVD, to more closely resemble initial clock scheme, so it's simpler to get back to it during shutdown. * Migrate some startup code to application board. * Si5351: Use correct methods for reset(). update_output_enable_control() doesn't reset the enabled outputs to the reset state, unless the object is freshly initialized, which it isn't when performing firmware shutdown. For similar reasons, use set_clock_control() instead of setting internal state and then using the update function. * GPIO: Set SPIFI CS pin to match input buffer state coming out of bootloader. * Change application board.c to .cpp, with required dependent changes * Board: Clean up SCU configuration code/data. * I2S: Add shutdown code and use it. * LPC43xx: Consolidate a bunch of structures that had been scattered all over. ...because I'm an undisciplined coder. * I2S: Fix ordering of branch and base clock disable. Core was hanging, presumably because the register interface on the branch/peripheral was unresponsive after the base clock was disabled. * Controls: Save and expose raw navigation wheel switch state I need to do some work on debouncing and ignoring simultaneous key presses. * Controls: Add debug view for switches state. * Controls: Ignore all key presses until all keys are released. This should address some mechanical quirks of the navigation wheel used on the PortaPack. * Clock Manager: Wait for only the necessary PLL to lock. Wasn't working on PortaPacks without a built-in clock reference, as that uses the other PLL. TODO: Switching PLLs may be kind of pointless now... * CMake: Pull HackRF project from GitHub and build. * CMake: Remove commented code. * CMake: Clone HackRF via HTTPS, not SSH. * CMake: Extra pause for slow post-DFU firmware boot-up. * CMake: TODO to fix SVF/XSVF file source. * CMake: Ask HackRF hackrf_usb to make DFU binary. * Travis-CI: Add dfu-util, now that HackRF firmware is being built for inclusion. * Travis-CI: Update build environment to Ubuntu xenial Previously Trusty. * Travis-CI: Incorrectly structured my request for dfu-util package. I'm soooo talented. * ldscript: Mark flash, ram with correct R/W/X flags. * ldscript: Enlarge M0 flash region to 1Mbyte, the size of the HackRF SPI flash. * Receiver: Hide PPM adjustment if clock source is not HackRF crystal. * Documentation: Update product photos and README. * Documentation: Add TCXO feature to README description. * Application: Rearrange files to match HAVOC directory structure. * Map view in AIS (#213) * Added GeoMapView to AISRecentEntryDetailView * Added autoupdate in AIS map * Revert "Map view in AIS (#213)" This reverts commit 262c030224b9ea3e56ff1c8a66246e7ecf30e41f. This commit will be cherry-picked onto a clean branch, then re-committed after a troublesome pull request is reverted. * Revert "Upstream merge to make new revision of PortaPack work (#206)" This reverts commit 920b98f7c9a30371b643c42949066fb7d2441daf. This pull request was missing some changes and was preventing firmware from functioning on older PortaPacks. * CPLD: Pull bitstream from HackRF project. * SGPIO: Identify pins on CPLD by their new functions. Pull down HOST_SYNC_EN. * CPLD: Don't load HackRF CPLD bitstream into RAM. Trying to converge CPLD implementations, so this shouldn't be necesssary. HOWEVER, it would be good to *check* the CPLD contents and provide a way to update, if necessary. * CPLD: Tweak clock generator config to match CPLD timing changes in HackRF. * PinConfig: Drive CPLD pins correctly. * CMake: Use jboone/hackrf master branch, now that CPLD fixes are there. * CMake: Fix HackRF CPLD SVF dependency. Build would break on the first pass, but work if you restarted make. * CMake: Fix my misuse of the HackRF CMake configuration -- was building from too deep in the directory tree * CMake: Work-around for CMake 3.5 not supporting ExternalProject_Add SOURCE_SUBDIR. * CMake: Choose a CMP0005 policy to quiet CMake warnings. * Settings: Show active clock reference. Only show PPM adjustment for HackRF source. * Radio Settings: Change reference clock text color. Make consistent color with other un-editable text. TODO: This is a bit of a hack to get ui::Text objects to support custom colors, like the Label structures used elsewhere.
2019-02-03 19:25:11 +01:00
/* Cache some configuration data from persistent memory. */
persistent_memory::cache::init();
2022-04-06 08:16:57 +12:00
touch::adc::init();
controls_init();
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(10);
Upstream merge to make new revision of PortaPack work (#206) * Power: Turn off additional peripheral clock branches. * Update schematic with new symbol table and KiCad standard symbols. Fix up wires. * Schematic: Update power net labels. * Schematic: Update footprint names to match library changes. * Schematic: Update header vendor and part numbers. * Schematic: Specify (arbitrary) value for PDN# net. * Schematic: Remove fourth fiducial. Not standard practice, and was taking up valuable board space. * Schematic: Add reference oscillator -- options for clipped sine or HCMOS output. * Schematic: Update copyright year. * Schematic: Remove CLKOUT to CPLD. It was a half-baked idea. * Schematic: Add (experimental) GPS circuit. Add note about charging circuit. Update date and revision to match PCB. * PCB: Update from schematic change: now revision 20180819. Diff was extensive due to net renumbering... * PCB: Fix GPS courtyard to accommodate crazy solder paste recommendation in integration manual. PCB: Address DRC clearance violation between via and oscillator pad. * PCB: Update copyright on drawing. * Update schematic and PCB date and revision. * gitignore: Sublime Text editor project/workspace files * Power: Power up or power down peripheral clock at appropriate times, so firmware doesn't freeze... * Clocking: Fix incorrect shift for CGU IDIVx_CTRL.PD field. * LPC43xx: Add CGU IDIVx struct/union type. * Power: Switch off unused IDIV dividers. Make note of active IDIVs and their use. * HackRF Mode: Upgrade firmware to 2018.01.1 (API 1.02) * MAX V CPLD: Refactor class to look more like Xilinx CoolRunner II CPLD class. * MAX V CPLD: Add BYPASS, SAMPLE support. Rename enter_isp -> enable, exit_isp -> disable. Use SAMPLE at start of flash process, which somehow addresses the problem where CFM wouldn't load into SRAM (and become the active bitstream) after flashing. * MAX V CPLD: Reverse verify data checking logic to make it a little faster. * CPLD: After reprogramming flash, immediately clamp I/O signals, load to SRAM, and "execute" the new bitstream. * Si5351: Refactor code, make one of the registers more type-safe. Clock Manager: Track selected reference clock source for later use in user interface. * Clock Manager: Add note about PPM only affecting Si5351C PLLA, which always runs from the HackRF 25MHz crystal. It is assumed an external clock does not need adjustment, though I am open to being convinced otherwise... * PPM UI: Show "EXT" when showing PPM adjustment and reference clock is external. * CPLD: Add pins and logic for new PortaPack hardware feature(s). * CPLD: Bitstream to support new hardware features. * Clock Generator: Add a couple more setter methods for ClockControl registers. * Clock Manager: Use shared MCU CLKIN clock control configuration constant. * Clock Manager: Reduce MCU CLKIN driver current. 2mA should be plenty. * Clock Manager: Remove redundant clock generator output enable. * Bootstrap: Remove unnecessary ldscript hack to locate SPIFI mode change code in RAM. * Bootstrap: Get CPU operating at max frequency as soon as possible. Update SPIFI speed comment. Make some more LPC43xx types into unions with uint32_t. * Bootstrap: Explicitly configure IDIVB for SPIFI, despite LPC43xx bootloader setting it. * Clock Manager: Init peripherals before CPLD reconfig. Do the clock generator setup after, so we can check presence of PortaPack reference clock with the help of the latest CPLD bitstream. * Clock Manager: Reverse sense of conditional that determines crystal or non-crystal reference source. This is for an expected upcoming change where multiple external options can be differentiated. * Bootstrap: Consolidate clock configuration, update SPIFI rate comment. * Clock Manager: Use IDIVA for clock source for all peripherals, instead of PLL1. Should make switching easier going forward. Don't use IRC as clock during initial clock manager configuration. Until we switch to GP_CLKIN, we should go flat out... * ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution. * PortaPack IO: Expose method to set reference oscillator enable pin. * Pin configuration: Do SPIFI pin config with other pins, in preparation for eliminating separate bootloader. * Pin configuration: Disable input buffers on pins that are never read. * Revert "ChibiOS M0: Change default clock speed to 204MHz, since bootstrap now maxes out clock speed before starting M0 execution." This reverts commit c0e2bb6cc4cc656769323bdbb8ee5a16d2d5bb03. * Remove unused board files. * Add LPC43xx functions. * chibios: Replace code with per-peripheral structs defining clocks, interrupts, and reset bits. * LPC43xx: Add MCPWM peripheral struct. * clock generator: Use recommended PLL reset register value. Datasheet recommends a value. AN619 is quiet on the topic, claims the low nibble is default 0b0000. * GPIO: Tweak masking of SCU function. I don't remember why I thought this was necessary... * HAL: Explicitly turn on timer peripheral clocks used as systicks, during init. * SCU: Add struct to hold pin configuration. * PAL: Add functions to address The Glitch. https://greatscottgadgets.com/2018/02-28-we-fixed-the-glitch/ * PAL/board: New IO initialization code Declare initial state for SCU pin config, GPIOs. Apply initial state during PAL init. Perform VAA slow turn-on to address The Glitch. * Merge M0 and M4 to eliminate need for bootstrap firmware During _early_init, detect if we're running on the M4 or M0. If M4: do M4-specific core initialization, reset peripherals, speed up SPIFI clock, start M0, go to sleep. If M0: do all the other things. * Pins: Miscellaneous SCU configuration tweaks. * Little code clarity improvement. * bootstrap: Remove, not necessary. * Clock Manager: Large re-working to support external references. * Fix merge conflicts
2019-01-11 07:56:21 +01:00
clock_manager.set_reference_ppb(persistent_memory::correction_ppb());
clock_manager.enable_if_clocks();
2015-08-01 13:42:27 -07:00
clock_manager.enable_codec_clocks();
radio::init();
2022-04-06 08:16:57 +12:00
sdcStart(&SDCD1, nullptr);
sd_card::poll_inserted();
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(10);
2023-05-05 12:58:14 +02:00
portapack::cpld::CpldUpdateStatus result = portapack::cpld::update_if_necessary(portapack_cpld_config());
if ( result == portapack::cpld::CpldUpdateStatus::Program_failed ) {
2023-05-05 12:46:59 +02:00
chThdSleepMilliseconds(10);
// Mode left (R1) and right (R2,H2,H2+) bypass going into hackrf mode after failing CPLD update
2023-05-05 13:39:04 +02:00
// Mode center (autodetect), up (R1) and down (R2,H2,H2+) will go into hackrf mode after failing CPLD update
2023-05-05 12:46:59 +02:00
if (load_config() != 3 /* left */ && load_config() != 4 /* right */){
shutdown_base();
return false;
2022-04-07 08:41:46 +12:00
}
2022-04-06 08:16:57 +12:00
}
if( !hackrf::cpld::load_sram() ) {
chSysHalt();
}
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(10); // This delay seems to solve white noise audio issues
2022-04-07 21:22:47 +12:00
LPC_CREG->DMAMUX = portapack::gpdma_mux;
gpdma::controller.enable();
2022-05-05 08:31:54 +12:00
chThdSleepMilliseconds(10);
2022-04-07 21:22:47 +12:00
audio::init(portapack_audio_codec());
2022-04-07 21:22:47 +12:00
return true;
2015-08-01 13:42:27 -07:00
}
2023-04-01 18:36:57 +02:00
void shutdown(const bool leave_screen_on) {
gpdma::controller.disable();
2023-04-01 19:16:01 +02:00
if (!leave_screen_on) {
2023-04-01 18:36:57 +02:00
backlight()->off();
display.shutdown();
}
2015-08-01 13:42:27 -07:00
radio::disable();
audio::shutdown();
hackrf::cpld::init_from_eeprom();
shutdown_base();
2015-08-01 13:42:27 -07:00
}
} /* namespace portapack */