- Added failsafe flightmode and rc control box.
To make failsafe procedure a separate flight mode and make it possible
to trigger failsafe with an AUX switch.
- Failsafe mode is activated when failsafe is active.
RC link lost is simulated with the failsafe AUX switch.
When NOT armed: failsafe switch to failsafe mode is shown in GUI (mode
tab).
- Activate failsafe mode with AUX switch.
- Prevent arming when failsafe via AUX switch is active (safety issue).
- Make failsafe disarm if motors armed and throttle was LOW (2D & 3D)
for `failsafe_throttle_low_delay` time (__JustDisarmEvent__).
Applied code changes to effectively add pull request: Make failsafe
disarm if motors armed and throttle low #717.
- Use failsafeIsMonitoring() to actually start monitoring.
- Added `failsafe_kill_switch` to code.
When set to 1 (0 is default), the failsafe switch will instantly disarm
(__KillswitchEvent__) instead of executing the landings procedure.
Arming is NOT locked after
this, so the craft could be re-armed if needed.
This is intended for racing quads where damage and danger must be
minimized in case of a pilot error.
- Added `failsafe_throttle_low_delay`, adapted documentation.
Used to adjust the time throttle level must have been LOW
to _only disarm_ instead of _full failsafe procedure_
(__JustDisarmEvent__).
- Updated the failsafe documentation.
- Re-enable arming at end of failsafe procedure.
At the end of a handled failsafe event, that means: auto-landing,
__JustDisarmEvent__ or __KillswitchEvent__, the RX link is monitored for
valid data.
Monitoring is a part of the failsafe handling, which means the craft is
still in failsafe mode while this is done.
Arming is re-enabled (allowed) when there is a valid RX link for more
then XX seconds, where XX depends on the handled event like this:
1. XX = 30 seconds after auto landing.
2. XX = 3 seconds after __JustDisarmEvent__.
3. XX = 0 seconds after __KillswitchEvent__.
NOTE: When armed via an AUX switch, you will have to switch to the
disarmed position at the very end to be able to re-arm.
The failsafe mode will not end until you do.
- __KillswitchEvent__ has now priority over __JustDisarmEvent__
- Apply rxfail values instantly when failsafe switch is ON
- Added missing cases to display.c
Show M when failsafe is monitoring for RX recovery (AND disarming when
armed with a switch).
===
Reworked the code from counter-based to time-based.
- AUX failsafe switch now has identical behavior to RX loss.
- Added RX failure and RX recovery timing.
- __KillswitchEvent__ skips RX failure detection delay (direct disarm).
===
[UNIT TESTS]
Adapted failsafe related unittests from counter-based to time-based
- Added failsafeOnValidDataFailed() to some tests
- Removed duplicate test setup from rc_controls_unittest.cc
- Removed magic numbers from rx_ranges_unittest.cc and rx_rx_unittest.cc
- Reworked all test-cases for flight_failsafe_unittest.cc
The problem is caused by hardware counters (timekeeping and PPM/PWM
measurement) that keep running while the firmware is frozen. The result
is misinterpretation of received data.
EEPROM read & write now call a suspend and resume function to make RX
ignore incoming wrong data during reads/writes (and flush the wrong data
on resume).
Fixes issue #1257
(+1 squashed commit)
- Moved the check for skipSamples to the right place.
As commented by hydra
Stick channels only have AUTO and HOLD mode.
AUX channels only have SET and HOLD mode.
Added check to parameter in CLI.
Modified rx.md to reflect changes (and more).
+1 squashed commit:
- A cleaner approach for the same functionality
Basically addressing all comments from Hydra
Note: since it didn't actually calibrate anything it has been renamed to
rxrange.
Added ability to reset rxranges using `rxrange reset` - this follows the
same pattern as other cli commands.
On bad (out-of-range) pulses; ROLL, PITCH, YAW will go to `mid_rc` and
THROTTLE will go to `rx_min_usec` (to `mid_rc` for 3D mode). So these
channels will no longer be set by the user directly.
Fallback values for the aux switches (0 .. max) can be set with this
version. Since these switches may trigger all kind of things, the user
needs control over them in case of a RX failsafe event.
A single flight control channel failure (first 4) when using parallel
PWM is interpreted as a failure for all flight control channels (first
4), since the craft may be uncontrollable when one channel is down. (+4
squashed commit)
Squashed commit:
[dbfea9e] Apply fallback values also when serial_rx init failed and/or
RX
disconnected and/or no signal received.
[b5a2ecd] Added get/set MSP commands for RXFAIL config
Bumped API minor version up.
[c0e31ce] minor change for coding standard
[322705f] Added programmable RX channel defaults on rx lost Update #2
Main rule logic and MSP commands ported from baseflight.
Gimbal mixer updated to use rules. This allows us to remove more
conditional logic. Operation of gimbal servos is now different.
pterm_cut_hz added
Let's play with this as well to get more noise filtered
Code Cleanup
Make filter more flexible for reuse
rewrite correction pterm
Define static delta in filter
Fix array count
ident
return function for filter
Filter Function enhanced
Full software filtering (DTerm, PTerm, Gyro, Acc)
Normalize Variables
Revert Back gyro settings
Bugfix gyro/acc filter // (MPU60xx equalize lpf settings)
Moved filtering to mw.c
This has been done to prevent reusing old cycletime for filter function.
acc_cut_hz removed (not needed)
Harakiri zero fix
1) the 2 extra servos were not used for anything, yet.
2) the MSP packet to set ALL the servo configurations in one go is
larger than the MSP input buffer size.
Likely the MSP_SET_SERVO_CONF should be updated with command that takes
an index of a servo to configure, as per mode ranges/colors/leds/etc.
Fixes#1002
This primarily is to avaoid stalling the main loop when beeping and
flashing.
This is needed because oneshot ESCs do not receive updates when the main
loop is stalled.
Additionally the beeper code for sticks held in disarm position is
changed since it also clashed with profile selection. Now profile
selections can be seen and heard clearly.
Other subsections of the system that changed the LED0 state while the
main loop is running have been updated to use the status indicator API
instead of blindly hitting the hardware which previously caused lots of
odd LED flashing behaviour - now it is consistent.
