BetaFlight 3.2

skraiman edited this page Nov 9, 2017 · 79 revisions

IMPORTANT ANNOUNCEMENT

Betaflight v3.2 will be the last version to include STM32F1 based flight controllers. From v3.3 onwards support for those flight controllers will no longer be provided, this includes the NAZE, CC3D (original), ALIENFLIGHTF1 and MICROSCISKY and their clones.

If you have any questions or concerns please advise in Boris' BetaFlight thread. Providing notice here so that there is plenty of time to prepare. v3.2 will be released in October, whilst v3.3 won't be until next year.

You can see some information here : https://github.com/betaflight/betaflight/projects/2 however it doesn't look complete.

From Boris (30 June 2017):
Well release candidate is around the corner. Basically a lot of fundamental changes. Not much of those will affect the user, but there are also quite some new features as well.
The major complaint about Betaflight was too frequent updates in the past. So we slowed that down. Not having the pressure for quick releases gives some time for thinking and researching new things.

Current V3.2 files are here: https://ci.betaflight.tech/job/Betaflight/lastSuccessfulBuild/artifact/obj/ or https://betaflight.qmd.cl/

Note: When discussing this Version in the Forums (Boris' BetaFlight thread)

Please State: Version and Build Number.

If having a problem check ISSUES to see if it have been reported and a possible work around.

From Boris:

Hi Guys. I hope everyone is fine.

We finally have a release candidate on schedule like promised!
https://github.com/betaflight/betaflight/releases

Betaflight 3.2.1

Maintenance Release

Please read the Instructions for Upgrading.

This release contains bugfixes and target changes only. For a full list of new features see 3.2.0.

Fixes:

  • Configuration does not reset to defaults after an upgrade, and is corrupted instead (#4280);
  • MavLink port sharing with MSP does not work (#4287);
  • CRSF Telemetry gets stuck when ACC is turned off (#4279);
  • Wrong timer assignment for PA14 (#4297);
  • GD32F350x6 ESCs not supported by 4-way interface (#4329).
  • When a battery is connected, the ESCs are initialised twice in a row (#4322);
  • Camera control menu is overlaid with the OSD statistics page if camera control is invoked from the statistics page (#4293);
  • VTX band colours for LED_STRIP are wrong (#4362);
  • Dshot beacon sounds when beeper is turning off (#4360);
  • Dshot ESC info is inconsistent in CLI (#4308);
  • Dshot1200 timing is wrong for F4 and F7 (#4367);
  • Stick commands for camera control do not repeat when holding stick (#4368);
  • Dshot on F7 does not work (#4165);
  • Toggling WiFi on a RunCam Split will turn it off in flight (#4369);
  • Arming flag names are not used in CLI in some cases, even if compiled in (#4370).

Target updates:

  • KIWIF4V2 / PLUMF4: Moved LED_STRIP pin back to VTX.DTA (#4285);
  • New target: XRACERF4 (#4255).
  • ALIENFLIGHT: Fixed instable hardware detection issue. (#4300).
  • BEEBRAIN_V2: Separated into V2D and V2F (#4318).

Betaflight 3.2.0 Release

New:

  • Full F7 support (@sambas @blckmn )
  • SITL simulator support (@cs8425)
  • Added crosshairs to CMS (@lostcontrol)
  • Improved ITerm windup handling for tricopter (@martinbudden)
  • Added SP RACING F3 OSD/PDB suport (@hydra)
  • Added horizon_tilt_effect command (@ethomas997)
  • Improved scheduler efficiency (@lilcw)
  • Increased motor output resolution (@borisbstyle)
  • Improved pwm timer precision (@blckmn)
  • Added new motor protocol Proshot1000 (@TonyBazz)
  • Improved configuration architecture (PG implementation @martinbudden , @ledvinap)
  • OSD improvements
  • New target support
  • Added Automatic Notch filter based on noise frequency (@rav-rav , @martinbudden)
  • Fix/enable disabling of rc smoothing in level modes
  • Flip inverted quad on ground (anti-turtle mode) (@brycedjohnson)
  • Improved blackbox storage to be more compact. Allows recording at higher rates and/or longer logs on flash storage (@martinbudden)
  • Camera control
  • TBS compatible LED frequency indicator
  • beeper / OSD / CLI indication of reason for not arming;
  • Added Iterm limit to prevent strong accumulations

Issues

Experimental - use with caution

  • Added experimental crash detection and recovery (@martinbudden)
  • Added experimental CPU overclock options for F4

Fixes:

  • Fix GPS serial overflow (@mikeller)
  • Fixed destabilisation on full throttle when Airmode disabled
  • Less chance on yaw spins with Iterm limit

Known issues:

  • The configuration after an upgrade from a previous version of the firmware is not properly reset, and is likely to be corrupted. To fix this, always do a CLI backup, before the upgrade, then do a 'Reset to defaults' / CLI defaults after upgrading, and restore from the backup (#4280);
  • MavLink port sharing with MSP does not work (#4287);
  • Camera control menu is overlaid with the OSD statistics page if camera control is invoked from the statistics page (#4293);
  • When a battery is connected, the ESCs are initialised twice in a row (i.e. the initialisation beeps sound twice) (#4257).

RC2 Changes:

  • Fix for new blackbox denominators
  • Improved ACC scaling in some cases

RC3 Changes

  • MATEKF405 Enable follow me spektrum binding
  • Fix for blackbox p_denom
  • ROM savings
  • Add missing AHI sidebar in cli
  • Check for notch 0 in filter init
  • rework on pin timer mapping
  • Fix buzzer on alienwhoop v2.0
  • Spektrum Telemtry fix for FlightPack Capacity
  • Fix Acc reporting for some gyros in configurator

RC4 Changes

  • Added Experimental Slew filter
  • Fixed lockup on MSP for dterm filter change
  • Improved gyro debug logging
  • Validate features when not supported
  • Experimental Yaw overflow handling
  • Added OSD warning for battery not full
  • Fix cli settings for OSD timers
  • Added Iterm limit to prevent strong accumulations
  • Reset iterm after crash for crash detection

RC4 Bugs (see the Issues and Pull requests in Github for details)

  • gyro debug and notch debug doesnt work. No data is displaying in analyzer. Fixed in Build #179.
  • various Issues with DSHOT1200
  • NAZE target not able to log data to Black Box
  • Gyro Cal beeps issue 4107
  • effectively made the setpoint weight behave as if it was set to zero (no matter what you set it to) if the setpoint transition was at the default 1.0. Fixed in RC5.

RC5 Changes

  • Default setpoint weight has been changed to 0. This should result in flight behaviour identical to previous 3.2 RC's if the user left the relax ratio at the default of 1, because relax ratio of 1 was disabling setpoint weight in those RC's. This has been fixed in RC5, so that D weight is active even if relax is 1. Users wanting D weight behaviour like 3.1 should set weight to 0.6.
  • Removed experimental Slew filter
  • Disabled beeping on gyro calibration during boot
  • Added gyro overflow protection for some newer gyros (ICMxxx gyro series is known to show inversion due to overflow above +/-2000deg/sec)
  • Added internal resistance adjustment for camera control
  • Show MSP version in cli

RC5 Bugs (see the Issues and Pull requests in Github for details)

RC6 Changes

  • Disable small angle during crashflip
  • Fix mixer for dshot 3D
  • Improved Mixer for crashflip mode
  • CCD fix for parallel PWM
  • Remove pidsumlimit from crashflip mode

Other Features:

Black Box Viewer

https://github.com/betaflight/blackbox-log-viewer/commits/master

Instructions for Upgrading -

DO NOT use Copy/Paste of CLI commands since many have changed- either names or new ones. Use the latest Configurator to setup then the CLI Tab and manually set.

Joshua Bardwell's Betaflight 3.2 Ultimate Setup Guide

New features and changes of how old feature work

Switching from 'tlm_inversion' to 'tlm_inverted'

With the change of the configuration parameter tlm_inversion to the parameter tlm_inverted, the scope and function of the parameter have changed:

  • tlm_inverted applies to all telemetry protocols;
  • tlm_inverted = on means that telemetry is expected to be inverted compared to what it is for the selected protocol (i.e. when using SmartPort, tlm_inverted = off means that the flight controller expects the telemetry signal to be inverted serial, since this is the default for SmartPort).

This means that for all protocols, if unmodified hardware is used tlm_inverted = off is most likely the correct setting.

(Also, note that this only works for F3 / F7 based boards, or for F4 boards on ports with switchable external inverters.)

Dynamic Filters

TO DO - fill in CLI commands and what they do.

NOTE: F1 boards can't handle dynamic filters.. no space, no power for it.

This is the current signal path in ascii art:

gyro -> dynamicNotch -> notch1 -> notch2 -> lpf -> P term
-> motors
-> D term -> notchD -> lpfD -> setpointRelax&Weight /

Dynamic filter runs in the gyro loop, so PID loop freq doesn't matter. Don't run 8k on F3 processors, use 4/4 on those. F4 / F7 should run fine on higher freq.

mjbudden posted on Boris' thread regarding the gyro/PID loop speed and Dynamic Filtering: Official recommendation from myself and r.a.v. (who between us wrote the code) is not to use above 4k/4k on F3 processors if using dynamic filtering. When running 8k gyro you have 125 microseconds in each loop and it is simply not enough time to run the calculations. So the PID loop will run slower than 8k anyway. There may be no outward signs of problems, but there will be jitter in the PID loops and that can potentially cause problems.

Crash Recovery

TO DO - fill in CLI commands and what they do.

Briefly the mode works like this:

  1. A crash is detected if crash_dthreshold and crash_gthreshold are exceded.
  2. Once a crash is detected, the craft tries to level itself, ignoring RC input on roll and pitch axes.
  3. Craft continues to try and level itself until either:
  4. crash_time milliseconds after the crash, or
  5. the craft angle is less than crash_recovery_angle degrees on both the roll and pitch axes and the craft gyro rate is less that crash_recovery_rate degrees/second on both roll and pitch axes.
  6. For testing purposes crash recovery can be enabled by turning the beeper on (this will be removed/changed if crash recovery proves successful).

TCHTHSKY Posted in Boris' thread- Here's the code for it. It has comments:
And here is the feature request:
And here's this:
Here's another from earlier in this thread:

Important note from mjbudden:

The crash recovery routine does not currently attempt to limit yaw spin - it just tried to level the quad. I did not anticipate the "yaw spin to the moon" problem when I wrote the crash recovery routine. I'm now working on a solution. See PR github.com/betaflight/betaflight/pull/3909 in github for the current discussion on the subject. Feel free to contribute to the discussion. Also black box logs of "yaw spins to the moon" are very welcome and will help analyze and solve the problem.
From Yamaford:
Just to clear this up for others that may read this wrong, I don't think your crash recovery code is the culprit as the crazy "YSTTM" (Yaw spin to the moon) is happening with or without acc or crash recovery enabled right?
From mjbudden:
Correct. Yaw spin to the moon happens with or without crash recovery enabled and is not cause by crash recovery code.
There is a new PR that tries to fix this, see PR 3909. If you could test this and provide blackbox logs it would be most useful.

Crash Recovery while failsafe is active

In the event that a crash is detected while the vehicle is in failsafe mode, it will immediately disarm. This is to prevent damage or injury should the vehicle strike an object or the ground during a failsafe event (i.e. loss of RX signal). This should also disarm the vehicle after failsafe LANDING has put the vehicle on the ground.

Once the failsafe condition has cleared (i.e. RX signal is restored for a minimum of 30 seconds), the vehicle will once again be ready to arm.

"flip after crash mode" (was called Turtle Mode)

TO DO - Describe what this does, how to setup and which ESC firmware supports this.

NOTE: Ensure the ESC Firmware supports DSHOT commands to Reverse the motors. NOT all latest ESC firmware has this support. Check with the ESC Firmware developers/Docs for the ESCs you are using.

Description by brycej:
In its current form (as of last week, 8aug17, or so). You need to flick the mode switch and then arm. Throttle and yaw sticks don't do anything. Pitch and roll will control which props are spinning based on stick deflection. With the pitch and roll centered no props are moving.
Take a good guess on the side with free props and try and flip it over. If it isn't flipping, try a different pitch or roll. Don't just jam the stick over and leave it if it is stuck. I haven't burned up any escs in quite a bit of testing, but I'm just someone will manage it with some effort....
Disarm after flipped. Undo flipaftercrash mode and rearm to fly away.
Other tips: Has worked well to get me out of trees.

Try it with props off on the bench first. 16.63 from official blheli hasn't some issues reversing all the motors because it is a little more stringent then 16.67 from the unofficial blheli fork. Newer betaflight versions have some fixes in to help that.
Also, try it in short grass LOS so you can see how it works...
Can't fix broken props or save you or if long grass, so doesn't always work.

RCs 4 or below inverted the mixer and had all 4 props reversed and spinning. That works too but it is a little more messy on the ground and requires some finesse.

Yaw Jump tuning

Post by ctzsnooze:
The jump up when abruptly stopping / reversing a hard yaw spin is because two motors must go full on to generate maximum yaw torque. That's like applying 50% throttle.
acc_limit_yaw delays the speed at which yaw can accumulate, but not the maximum amount, so as you say it kind of delays the onset of the jump, but if a big correction is needed it will still jump at the same speed. As you say, slowing the rate of onset of yaw moves using a lower acc_limit_yaw will reduce yaw responsiveness to stick inputs.
pid_sum_yaw indirectly limits the maximum speed the motors will be requested to attain during a yaw move. That will limit the maximum possible yaw rate, and reduce the climb rate to the same extent. Basically jump and yaw are inextricably linked. But this should retain crisp yaw responsiveness for inputs that don't hit the limit.
RC5 has a number of measures to reduce yaw spin to the moon problems. I haven't had a bad one since going to RC5.

Reversing the prop rotation also seems to me to reduce the amount of spin acquired when clipping a gate with the outside of a front prop. This reduces the chance of getting a big spin and a big climb after hitting a gate.
Keep in mind that if you clip a gate and spin at high rotational rates, and if you run a lot of yaw P, two of your motors will go full on to oppose that spin, and may need to stay on for some time before regaining control.
I configure my sticks so that equal yaw and roll stick deflections result in a properly banked turn with the quad at 45 degrees. Hence I set the same rates for yaw as for roll. However because yaw responsiveness is inherently weaker than roll, I do need to run quite a lot of yaw P, which does cause fast yaw climbs after clipping gates.

FPV Camera control via OSD

Consult wiki and RCG thread to see if your camera requires special handling. See: https://github.com/betaflight/betaflight/wiki/FPV-Camera-Control-(Joystick-Emulation)

Experimental Slew filter- Removed in RC5

https://github.com/betaflight/betaflight/pull/3983
More info here:
https://github.com/betaflight/betaflight/pull/3909
https://github.com/betaflight/betaflight/issues/3959#issuecomment-326430286 https://github.com/betaflight/betaflight/pull/3950
https://github.com/betaflight/betaflight/issues/3893

Arming Flags

Are now in the CLI STATUS. Example: Arming disable flags: RX LOSS CLI
On F4/F7 STM32's the Arming status will print out names whereas on F1/F3 STM32's this is a simple Bit Field.
See Arming Sequence and Safety for details.

Rate Profile and Profiles

From Woody_99's experience:
I will share what I have learned, and also that 3.2 is different in a couple of ways from previous versions.

  • rate profiles are definitely separated from regular profiles in 3.2
  • in 3.2, if I enable rate profile select on a 3 way switch, BF will use all three rate profiles no matter where the slider is positioned in the adjustments screen. Previously, if I just wanted to have two rate profiles tied to a single profile, the slider would control whether the rate profile was selected by switch position. Not a bad thing, just a change to be aware of.
  • Need to use a different slot now as well. Previously I could use the same slot for various different adjustments using different aux channels, but now different adjustments need a different slot as well.
    Post with a Adjustment Tab screen shot

CLI command changes in 3.2

Removed CLI Commands

feature -VBAT
feature -FAILSAFE
feature -CURRENT_METER
feature -BLACKBOX
feature -SDCARD
feature -VTX

set align_mag
set bat_detect_thresh
set blackbox_rate_num
set fixedwing_althold_dir
set frsky_vfas_cell_voltage
set gyro_use_32khz
set mwii_ibat_output
set servo_lowpass

New CLI commands

TO DO - list all new CLI commands with a description of what they do and how to use them (links are good) and what the Default and optional values are.

feature -DYNAMIC_FILTER

beeper BLACKBOX_ERASE

GLOBAL SET

set beeper_frequency = 0

[0..16000]

set blackbox_record_acc = ON

[OFF..ON]

set camera_control_key_delay = 150

[100..500]

set camera_control_mode = HARDWARE_PWM

[HARDWARE_PWM, SOFTWARE_PWM, DAC]

set camera_control_ref_voltage = 330

[100..400]

set dashboard_i2c_addr = 60

[8..119]

set dashboard_i2c_bus = 1

[0..2]

set esc_sensor_halfduplex = OFF

[OFF, ON]

set ibatv_offset = 0

[-16000..16000]

set ibatv_scale = 0

[-16000..16000]

set led_inversion = 0

[0..7]

set motor_pwm_inversion = OFF

[OFF, ON]

set report_cell_voltage = OFF

[OFF, ON]

set vbat_detect_cell_voltage = 30

[10..50]

set motor_pwm_protocol = ONESHOT125

[OFF, ONESHOT125, ONESHOT42, MULTISHOT, BRUSHED, DSHOT150, DSHOT300, DSHOT600, DSHOT1200, PROSHOT1000]

Per PROFILE:

set crash_delay = 0

[0..500]

set crash_dthreshold = 50 :degrees/second/second

[0..2000] dterm crash value, zero (off) by default, set to at least 1 to enable crash detection

set crash_gthreshold = 400 :degrees/second

[0..2000] gyro crash value

set crash_recovery = OFF

[OFF, ON, BEEP]

set crash_recovery_angle = 10 :degrees

[0..30]

set crash_recovery_rate = 100 :degrees/second

[0..255]

set crash_setpoint_threshold = 350 :degrees/second

[0..2000]

set crash_time = 500 :ms

[0..500]

set horizon_tilt_effect = 75

[0..250]

set horizon_tilt_expert_mode = OFF

[OFF, ON]

Name changes

yaw_motor_direction [1, -1] ==> yaw_motors_reversed [OFF, ON]

sport_halfduplex ==> set tlm_halfduplex

tlm_inversion' ==> tlm_inverted

spektrum_sat_bind_autorst ==> spektrum_sat_bind_autoreset

dfu ==> bl

current_meter_type = ADC ==> current_meter = VIRTUAL

d_lowpass ==> dterm_lowpass

d_lowpass_type ==> dterm_lowpass_type

d_notch_cut ==> dterm_notch_cutoff

d_notch_hz ==> dterm_notch_hz

digital_idle_percent [7.00] ==> dshot_idle_value [450]

yaw_accel_limit ==> acc_limit_yaw

yaw_control_direction [1] ==> yaw_control_reversed [OFF]

anti_gravity_thresh ==> anti_gravity_threshold

battery_meter_type ==> battery_meter

blackbox_rate_denom [4] ==> blackbox_p_denom [32]

ibat_offset [0] ==> ibata_offset [0]

ibat_scale [400] ==> ibata_scale [400]

Changes in the New Configurator GUI (V3.2.2)

Receiver Tab

Added "Stick Min", "Stick Center" & "Stick Max" ---these are simply MIN_CHECK, MID_RC, MAX_CHECK.

Home

Getting Started

Release Notes

Tuning Tips

Setup and More

Troubleshooting

Development

Boards

Page Templates

Clone this wiki locally
You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.