Previously the runtime page state was mixed with the page configuration.
This resolves this by:
a) making the configuration entries constant so they do not need to be
in RAM.
b) Maintaining page state independently.
Change to using an array of `uint8_t` for the `d_min_` axis parameters. Simplifies the code.
The CMS menu entries were incorrectly placed under the `FILTER PP` submenu. Moved to the `MISC PP` submenu.
Directly and easily set the minimum value for D on pitch and roll.
- Boost back to the primary D setting is generated from gyro or setpoint inputs.
- Setpoint input is stick derived, faster by 10ms approx, and does not respond to propwash.
- Gyro input is motor derived and slower, but responds to propwash
- timing value sets balance between gyro (100) and setpoint (0) boost factors
- gain value sets overall sensitivity
- default D mins are 20 roll 22 pitch
- default D is 35, 38; if undefined then normal 30, 32 values
TUNING
- D value to flip overshoot control
- D_min to noise, lower values mean cooler motors but perhaps more propwash.
- Advance, higher values bring the boost in earlier, and stronger overall, useful for very high flip rates, but dampen stick responsiveness slightly.
- Gain value adjust against logs, checking maximal boost with flips and some rise with propwash, should be edging to get up from the min value in normal flight.
This code cuts D by a specified percentage durning normal flight.
It lets D smoothly rise up to normal during rapid gyro moves like flips and rolls, and increase during prop wash events.
D should now be tuned to values the 'normal' 30-45 range.
If D is 40, a dterm_cut_percentage of 65 will cut D to 14 in normal flight, but the quad will still get full 40 of D to control bounce-back after flips and about 25 of D during strong prop wash.
The dterm_cut_percentage can be adjusted via the OSD, from the D filtering page.
Adding d_cut results in cooler motors, lower amounts of noise in motor traces and faster reactions to quick stick inputs.
Too high a dterm_cut_percentage may bring out P oscillation from lack of D. Values of 70% are generally OK.
Input is gyro differential (delta). Frequencies above 40hz (above propwash) are attenuated with a configurable (dterm_cut_range_hz) biquad filter. Lower values for range can be used if the quad is very noisy or gets low frequency D resonant oscillation. Up to 50 or 60hz may suit clean quads where prop wash control is the main priority. Too high a range value results in D being boosted from noise in normal flight.
The boost signal is 'integrated, smoothed and delayed' with a 7hz PT1 'dterm_cut_lowpass_hz' filter. The default of 7Hz gives about the right amount of smoothing and delay. Higher numbers cause the boost to come on faster, with less delay. Lower values delay the boost effect and cause it to last longer.
The dterm_cut_gain amount controls the strength of the boost effect by amplifying the input to the boosting effect. If the quad is flow gently, a higher gain value may be needed to gain full boost.
Logging with set debug_mode = D_CUT allows recording of realtime D values on roll and pitch into debug 2 and 3. The reatime D value should reach its set maximum during rapid turns, ideally at about the time D itself peaks. If it fails to reach the maximum, gain should be increased.
The D_Cut feature is not enabled on LUXV2RACE, OMNIBUS, SPRACINGF3NEO because there isn't enough flash space.
There have been a number of requests for this as it seems to be common for racers to tune these values and adjust the settings as they become more comfortable with a track. Previously the settings could only be adjusted in the CLI and required a computer. Adding them to the CMS makes it easy to adjust in the field.
Adds a race start assistance system that allows the pilot to pitch forward and then release the sticks with the quad holding position for the race start.
Restructures the PID controller to decouple feedforward from D.
Cleaned up the structure of the PID controller; moved some feature-based enhancements out of the main structure.
Feedforward becomes a separate component of the PID controller and there is now:
f_pitch
f_roll
f_yaw
The default values of 60 for pitch and roll matches the default setpoint weight used in BF3.4. Yaw previously had no setpoint weight capability so the default here needs to be discussed. Currently it's also set to 60 and flight testing seems positive. Feedforward on yaw adds a lot of value so I don't think we want to default to 0. Instead we need decide on the default.
All occurences of setpoint weight have been replaced by feedforward. "setpoint_relax_ratio" has been renamed to "feedforward_transition".
The pidSum now consists of P + I + D + F.
D has been added back for yaw (disabled by default with d_yaw = 0). We've found little need for D for normal quads but it may have value for other configurations - particularly tricopters.
Updated CMS menus to support adjusting the feedforward for each axis.
Changed the default for "rc_interp_ch" to be "RPYT". Need yaw to be smoothed to support feedforward.
Open issues:
Needs BFC support
- Need to add support for the axis "F" gains.
- Remove "setpoint weight" slider.
- Rename "D Setpoint transition" to "Feedforward transition"
Needs BBE support
- Header "setpoint_relaxation_ratio" has been renamed "feedforward_transition"
- Header "dterm_setpoint_weight" has been replaced with an array named "feed_forward_weight".
example: H feed_forward_weight:65,60,60 (R,P,Y)
- PID component "AXISF" has been added for all axes. Should be handled like P, I and D values.
- PidSum calculation needs to include F.
Needs LUA script support
- Support the renamed "setpoint_relax_ratio".
- Support for feedforward weight on all 3 axes.
Open code issues:
- rc_adjustments.c - support for adjusting feedforward weight for all axes. Currently only supporting roll - needs coordination with BFC.
* Dual-stage Gyro Filtering: PT1, FKF, and Biquad RC+FIR2
* Builds on the previous work of apocolipse.
* Fixes 'stage2'/'stage1' mis-naming to reflect where it is applied in the loop.
That is, the older Biquad, PT1, Denoise (FIR) filters are 'stage2' - applied
after dynamic and static notches (if enabled), and the controversial PT1,
'fast Kalman' filter, and Biquad RC+FIR2 filters, are 'stage1'. e.g. before
dynamic notch.
* FKF bruteforce Kalman gain removed. Calculate from half of PT1 RC constant,
automatically taking loop-time into account.
* New union type definition for stage1 filtering.
* New gyro sensor members for stage1 filter application function and states for
all three supported filter types
* New enum types for stage1 v. stage2. dterm lowpass type references 'stage2'.
* updates to CMS/MSP/FC to allow compilation (untested, probably breaks
MSP, Lua, and ~comms with BFC~).
* Refactors FKF initialization, update and associated structures to be faster by
not continuously calculating 'k'. Filter gain is calculated once during
initialization from RC constant as per PT1 and Biquad RC+FIR2. It was
discovered this converges to static value within 100 samples at 32kHz, so can
be removed. Remove related interface (CLI) settings.
* update dterm_lowpass_type to use new 'TABLE_LOWPASS2_TYPE' (biquad/pt1/FIR)
* Stage 1 defaults to PT1, 763Hz (equivalent to Q400 / R88 from quasi-kalman
filter) - suitable for 32kHz sampling modes. Can be switched to Biquad
RC+FIR2, and FKF.
* Update `#if defined(USE_GYRO_SLEW_LIMITER) to `#ifdef`.
* Includes optional Lagged Moving Average 'smoothing' pipeline step, applied (in
code) after the output of stage1.
* (diehertz): Removed redundant pointers from gyro filtering
* blackbox: fix indentation
* cms IMU menu: fix indentation
* filters: remove USE_GYRO_FIR_FILTER_DENOISE in filter type enum
* gyro sensors: go back to `if defined()` form. for slew limiter
* gyro sensors: increment parameter group version
* due to non-appending changes, the version must be bumped.
* Second PT1 on DTerm
This PR replaces the default biquad filter with a second PT1 set to
200Hz.
Basically allows the user to enable a second, set point configurable,
PT1 type first order low-pass filter on DTerm.
This is useful because most noise in most logs arises from D, not P.
The default is set to on, at twice the normal Dterm setpoint. This
provides greater Dterm cut than a single PT1, and twice the steepness
of cut above the second setpoint. Modelling shows significant
reductions in higher frequency Dterm noise with only minor additional
delay.
The improvement in noise performance will be less than for biquad, but
the delay is considerably less.
If with the default settings the overall noise improves a lot, it may
be possible bring D both filtering set points to higher numbers (e.g.
140/280), or alternatively remove other filters such as the notch
filters, while maintaining an adequate level of control over noise.
* Update names, old defaults, fix whitespace
Defaults restored to biquad with second PT1 off. ‘lpf’ retained as
abbreviation for values, otherwise generally remove ‘Filter’ where
redundant, replace ‘FilterLpf’ with ‘Lowpass’, etc, thanks Fujin and
DieHertz
* Remove underscore in lowpass_2, add hz to setpoint for lowpass
Thanks DieHertz
* completed replacing lpf with lowpass, added _hz to all lowpass set points in profile
Thanks DieHertz
* fix whitespace
fixed whitespace in settings.c
* whitespace attempt #57
* change lpf to lowpass where appropriate elsewhere
Note did not change OSD abbreviations, they are still LPF, and did not
change gyro_lpf anywhere.
* second attempt at a simple PT1 implementation
Basically copied from the DtermNotch implementation
* Second PT1 on DTerm
This PR replaces the default biquad filter with a second PT1 set to
200Hz.
Basically allows the user to enable a second, set point configurable,
PT1 type first order low-pass filter on DTerm.
This is useful because most noise in most logs arises from D, not P.
The default is set to on, at twice the normal Dterm setpoint. This
provides greater Dterm cut than a single PT1, and twice the steepness
of cut above the second setpoint. Modelling shows significant
reductions in higher frequency Dterm noise with only minor additional
delay.
The improvement in noise performance will be less than for biquad, but
the delay is considerably less.
If with the default settings the overall noise improves a lot, it may
be possible bring D both filtering set points to higher numbers (e.g.
140/280), or alternatively remove other filters such as the notch
filters, while maintaining an adequate level of control over noise.
* Rebase
* Remove underscore in lowpass_2, add hz to setpoint for lowpass
Thanks DieHertz
* completed replacing lpf with lowpass, added _hz to all lowpass set points in profile
Thanks DieHertz
* fix whitespace
fixed whitespace in settings.c
* whitespace attempt #57
* change lpf to lowpass where appropriate elsewhere
Note did not change OSD abbreviations, they are still LPF, and did not
change gyro_lpf anywhere.
* second attempt at a simple PT1 implementation
Basically copied from the DtermNotch implementation
* Whitespace fix - thanks, Ledvinap
* Fix PG issue
by moving added dterm_lowpass2_hz to bottom of struct
* Got rid of redundant indirection
* Fixed indentantion shifts
the flight controller code.
Likely the dependencies can be further improved. This is a
minimal-impact solution while there are a lot of other Betaflight PR's
still pending.
