Georacer
(George Zogopoulos)
May 7, 2025, 8:48am
1
Attendees (unique) : 12
master
← HishamGhosheh:jhemcuf405pro
opened 11:22PM - 04 Apr 25 UTC
Built/Tested the bidirectional dshot version based off `Copter 4.6.0-beta5` / b1… 36aa2b6daab5384c35325deb5adff8ca37b90f

This version of the board has I2C pads exposed.
Tested on my quad. Flies well
**Tested working:**
1. Bidirectional dshot with multi source notch filter
2. Compass / Baro / ELRS (500hz) / GPS (~16 sats / hdop 0.67)
3. POSHOLD tested in windy conditions (40 km/h wind) / RTL
4. Blackbox (can be choppy, possible due to high rate / sharing DMA with I2C for the bdshot version)
5. RGB led
6. Buzzer
7. Analog VTX (IRC Tramp) + OSD
8. Current sensing
9. ESC passthrough
10. Turtle mode
**Test quad**
- 3.5" 4S
- JHEMCU GHF405AIO-HD 40A F405 Baro OSD Dual BEC Flight
- LANNRC 1404 3800KV 3~6S
- HappyModel Crux35 3.5inch
- HGLRC Zeus nano VTX 350mW (IRC Tramp)
- HappyModel EP1 ELRS receiver
- CADDX ANT Nano 1200TVL
- Walksnail WS-M181 GPS M10 GNSS BUILT-IN QMC5883 Compass

**Test binaries**
Build off `Copter 4.6.0-beta5` / b136aa2b6daab5384c35325deb5adff8ca37b90f with cherry-picked commits
[jhemcuf405pro-df41f4f.zip](https://github.com/user-attachments/files/19623945/jhemcuf405pro-df41f4f.zip)
**Additional performance info**
Full parameters list (change extension to param)
[full parameters list.txt](https://github.com/user-attachments/files/19610938/full.parameters.list.txt)
Sample flight BB (very windy day so vibrations are expected)
[jhemcuf405pro_bb.zip](https://github.com/user-attachments/files/19610958/jhemcuf405pro_bb.zip)
Filters hover section

Filters gyro with multi source notch filtering

IMU Spectrogram pre filter

IMU Spectrogram post filter

Filters response

BB Baro

BB ESC RPM

Log GPS HDop

LOG GPS NSats

Log IMU Gyr

Log MAG (sharing DMA with SPI3/Flash doesn't seem to affect it)

PIDS Pitch

PIDS Roll

PIDS Yaw

UTC0707
Andy : Apparently NODMA is now correctly marked.
And only one IMU is mentioned in the README.
So probably all comments are fixed.
Merged!
master
← peterbarker:pr/remove-bad-baro-stuff
opened 12:46PM - 06 May 25 UTC
- the HAL_BARO_DEFAULT for the LPSx25 is unused and ancient (crazyflie specifies… this driver in its hwdef, so it can be done else-wise)
- HAL_BARO_* defines are unused, so comment them out
- HAL_BARO_DEFAULT defaults to HAL_BARO_NONE in AP_HAL_Boards.h, no need for other definitions to this value
- remove unused defines of various baro names in hwdef.h files
- remove defines of HAL_BARO_DEFAULT to values that don't exist / were never checked against / were killed before esp32 stuff was merged into master (eg. icarous)
- remove defines of baro variables that are the default anyway
-
UTC0710
Approved
master
← andyp1per:pr-blheli-passthru-fixes2
opened 03:52PM - 22 Mar 25 UTC
soft serial input was both racy and prone to resetting the line mode incorrectly… .
soft serial input was decoding input with trailing 1's incorrectly
soft serial input could miss data when timing out and restarting
DMA output could be left in a bad state
UTC0713
Andrew : I couldn’t get master to connect to BLHeli, but with this PR I managed to once .
Betaflight manages every time.
Andy : Are these AM32 ESCs?
A : Yes, they are. You can still flash non-CAN firmware on them.
Andy : AM32 in general doesn’t work with any of our releases. BLHeli generally does. BlueJay somewhere in the middle.
A : I noticed the last (stop) bit of an 8N1 packet from AM32 was half-width. I fixed it but it broke compatibility with many config tools. The issue seems to be that CPUs bit-bang the serial but they don’t have time to do the next thing, if they linger the full duration on that last bit.
It would be nice if the flight firmwares weren’t so pressing on the ESC to send/receive the next byte.
Do we have enough threading priority?
Andy : We probably do.
A : What if we shut down most other threads during passthrough?
Andy : Perhaps it will help with timings.
master
← peterbarker:pr/param-write-mavftp-reuse
opened 07:07AM - 06 May 25 UTC
Closes https://github.com/ardupilot/ardupilot/issues/29985
This work sponsore… d by Freespace Solutions
UTC0729
< Discussion on param ftp internals, to address an occasional issue on param ftp load >
master
← andyp1per:pr-fstrate-motortest
opened 09:46AM - 20 Jan 25 UTC
Motortest currently doesn't work with the fast rate thread and I suspect that it… was somewhat accidental that it did before. The issue is synchronization between the fast rate thread and the main thread when dealing with the motor test. This fixes the synchronization issues and allows motor test to work normally with the fast rate thread.
Reported here: https://discuss.ardupilot.org/t/unlocking-the-potential-of-faster-attitude-rates-in-copter-control/120743/54
UTC0737
Andy : I’m still getting an instant watchdog.
It’s most likely a race condition.
A : We need Randy for this, unfortunately.
master
← peterbarker:pr/roll-from-ahrs-not-cosphi
opened 10:28AM - 30 Apr 25 UTC
Issue https://github.com/ArduPilot/ardupilot/issues/7817 pointed out that the or… der in which we are retrieving the euler angles from the transformation matrix is incorrect.
Testing shows that there is a discrepancy between these values and the values retrieved directly from the AHRS.
Since we are simply attempting to retrieve the roll angle, extract it directly from the AHRS.
Closes https://github.com/ArduPilot/ardupilot/issues/7817
Ping @luweiagi
UTC0738
Merged!
master
← peterbarker:pr/clang-warnings
opened 05:03AM - 01 May 25 UTC
btiwise-op stuff; we intentionally use & to avoid short-circuiting.
Also fixe… s for `-Wunknown-warning-option`
UTC0743
We still can’t use clang in our everyday builds.
master
← peterbarker:pr/param-set-filter
opened 08:37AM - 06 May 25 UTC
Add `MAV_OPTION` bit to ignore setting of parameters via MAVLink.
Add a LUA s… cript which will honours `PARAM_SET` - based on a a whitelist of parameter names.
```
STABILIZE> Lua: State memory usage: 4824 + 10598
AP: Scripting: restarted
AP: param-set-filter script loaded
STABILIZE>
STABILIZE> param set DISARM_DELAY 5
STABILIZE> AP: param-set-filter: param set denied (DISARM_DELAY)
Flight battery 100 percent
AP: param-set-filter: param set denied (DISARM_DELAY)
Failed to set DISARM_DELAY to 5 (invalid returned value 3.0)
STABILIZE>
```
This work sponsored by Freespace Solutions
< Discussion on ways to lock down parameters against accidental changes. >