File content as of revision 234:ad68d7f8eff1:

04/30/2017 02:45 - added CHANGELOG.txt
04/30/2017 04:51 - non-saturating tables, 8KW out to 600 rad/s. Extremely high top speed (1500 rad/s). code changes: removed a bunch of debug variables and commented-out code
05/01/2017 00:04 - added some decoupling terms, which may or may not work.
05/01/2017 01:21 - driving code, 0.7g, achieved 40mph. Has throttle filtering
05/01/2017 03:17 - changed throttle limits to match 56k pulldown; relaxed THROTTLE_CRAZY bounds a bit
05/01/2017 06:40 - more tweaks to throttle limits
05/01/2017 07:21 - rising edge from disabled state now resets usec count in PwmIn
05/01/2017 07:58 - W_CRAZY limit bumped up, never an issue on the road but may behave strangely on the bench
05/01/2017 08:13 - enabled->disabled now resets usec count in PwmIn
05/01/2017 09:00 - bumped up THROTTLE_LOW from 1020 to 1050
05/02/2017 06:54 - fast logging is now back to logging vd, vq instead of decoupling terms
05/03/2017 08:24 - added circular buffer code
05/03/2017 08:27 - CircularBuffer moved to its own .cpp file
05/03/2017 08:36 - added median and moving average filters
05/03/2017 09:26 - added use_median option to circular buffers (since average filters may not want to recalculate the median every sample)
05/03/2017 09:39 - indexing bugfix in median code
05/04/2017 08:53 - added fast path for median-of-3
05/04/2017 10:33 - removed throttle_crazy checks as these lead to dangerous behavior; beginnings of new filter code in main.cpp
05/04/2017 11:14 - removed w_crazy checks, switched velocity to median filtering
05/04/2017 12:44 - moved median filter objects to ControlStruct, is_driving() now checks % torque cmd instead of raw throttle value
05/04/2017 13:34 - changed logic in PwmIn::handle_fall, which no longer resets timer on invalid edges
05/04/2017 17:16 - encoder class now sets TIM->ARR to CPR - 1
05/07/2017 13:36 - added overmodulation option to config_loop
05/07/2017 13:42 - bugfix in constrain_norm(), which previously stepped from max to 1 magnitude when limiting kicked in
07/01/2017 18:32 - added AutoMapper, AutoThrottleMapper classes for automated stall testing; parameters in this commit are currently connfigured for driving
09/04/2017 03:08 - added delay between startup and PwmIn creation in BREMSConfig.cpp to avoid power switch bounce induced lockups during logic power-on
09/05/2017 04:57 - formatting fix
02/07/2018 20:59 - new InvertingThrottleMapper, InterpolatingLutMapper classes. blended, smaller tables. BufferedLogger logging class
02/08/2018 22:03 - integrated BufferedLogger into motor control code, logger and commutation both run (but validity of logged data is untested)
02/08/2018 22:48 - integrated CommandProcessor and PreferenceWriter (but code does not use dynamic config variables or set error flags yet)
02/09/2018 14:28 - check for valid config in BREMSConfig.cpp is now 'if (PREFS_VALID != 1)', as flash bytes default to 0xFF
02/09/2018 15:22 - added functionality to disable logging during long UI operations (flash write)
02/09/2018 16:01 - moved pinouts to layout.h
02/09/2018 18:22 - moved hard-coded config values to defaults.h, started filling out errors.cpp and callbacks.cpp
02/09/2018 19:12 - changed object initialization order in BREMSConfig.cpp to read values back from flash first
02/09/2018 21:45 - started switching code to run on flash-saved parameters instead of hardcoded ones; moved delay in BREMSConfig.cpp to avoid intermittent freeze after programming
02/09/2018 21:54 - removed extra delay in BREMSConfig as it does not seem to mitigate the freezing issue; migrated to dynamic loop gains
02/09/2018 22:41 - removed ROW, COLUMNS from configurable values, for now. removed a bunch of extra #include's
02/09/2018 22:58 - everything but main.cpp and friends moved to flash configuration
02/10/2018 00:35 - added fancy masked error system
02/10/2018 01:05 - everything configured through flash variables, started adding support for operating modes
02/10/2018 01:35 - can now switch command source to terminal, added 'clear' command (which clears the screen on POSIX-compliant terminals)
02/10/2018 17:00 - fixed a bug in BREMSConfig where the delay was in the wrong spot
02/11/2018 02:11 - added BREMS_mode. most modes are stubs right now (BREMS_CFG does a rudimentary encoder zero-ing, BREMS_CFG disables outputs)
02/11/2018 02:54 - long commands can no longer be executed while control.enabled == true, invalid BREMS_src now sets control.user_cmd to 0
02/11/2018 03:16 - blocked switching modes when motor is running
02/11/2018 20:13 - scrambled the mode and error handling code in commutate() to be cleaner
02/11/2018 21:11 - further improvements to mode handling, output overrides and output disabling now live in their own functions
02/11/2018 21:21 - cmd_exit() checks that motor is not running
02/13/2017 20:33 - set serial RX interrupt to be lower priority than commutation interrupt
02/13/2018 23:15 - cmd_flush() now works in all modes (otherwise the controller would only ever boot up in MODE_CFG)
02/20/2018 00:54 - moved modes.cpp to correct location
05/29/2018 21:33 - microcontroller now resets resolver decode IC after voltages stabilize, to prevent position sensor from being off by pi radians
06/26/2018 15:47 - GetUnlimitedElecPosition() is now a private member of PositionSensorEncoder class (it is only used to compute mechanical position)
06/26/2018 15:57 - added induction machine CurrentModel, which is probably wrong
10/02/2018 02:15 - PwmIn::get_throttle() returns 0, not -1, if throttle is disabled.
10/02/2018 02:21 - processCmd() now uses else-if statements, prints message if command is invalid
10/02/2018 03:10 - added overly-complicated LedBlinker class for blinking status codes over an LED, added blinker object to IOStruct, STATUS_LED to hardware.h
10/16/2018 22:59 - renamed CircularBuffer to fCircularBuffer to avoid conflict with new mbed libs
10/27/2018 19:12 - moved encoder object to use PA_1 and PA_15; removed some spurious analog registers in BREMSConfigRegisters which are no longer being used
10/28/2018 21:27 - added some explanatory comments to BREMSConfigRegisters
11/10/2018 00:50 - moved RESOLVER_RESET pin to PC_0 to match new hardware
11/10/2018 01:26 - renamed ia_supp_offset, ib_supp_offset to ad1, ad2_supp_offset to reflect the fact that they correspond to ADC channel offsets and not currents (which can be renamed in elsewhere)
11/10/2018 01:32 - new pin mapping for new hardware
11/10/2018 01:50 - swapped ia and ib (back to what they were before 11/10/2018)
11/10/2018 03:36 - restructured code to allow for proper encoder zeroing - MODE_ZERO and MODE_CHR execute run-once functions when they are entered, then drop to MODE_CFG
11/10/2018 04:14 - 'some modes override FOC outputs' removed from commutate(), as it is no longer needed (ZERO, CHR have their own routines, CFG disables output anyway)
11/10/2018 04:21 - cmd_setp() sets setpoint to zero if there are errors present (to avoid sudden start of motor if errors are cleared)