Plane 4.0 release

urlu75 · June 22, 2020, 9:30am

Hello

Give a try to 4.0.6 beta 1.
Seems ok, auto mission for takeoff ok, tried for a small amount of time autotune and then as I have had a video issue I switched to cruise mode. I was not looking to fix my issue and the plane crashed in a tree in a small hill. Terrain follow was enable and map was download with the terrain.py before flight. I am not very familiar with log analysis and was not able to check if something went wrong and if terrain follow correctly worked.

Log file here

Thanks for any help

tridge · June 22, 2020, 9:35pm

thanks very much for reporting this. The terrain system was trying to download terrain data from your GCS but it failed to do so. It didn’t have any terrain data, so didn’t follow terrain.
The real bug was that there was no arming check failure for the lack of terrain data when you had TERRAIN_FOLLOW=1. I have fixed that ready for the next beta. My apologies for that bug. You should have had an arming failure.
I hope the plane survived!
Cheers, Tridge

tridge · June 22, 2020, 9:52pm

I’ve just released 4.0.6beta2. This is a minor update over beta1.

prevent jump to circle mode in TAKEOFF if flying for less than 10s
fixed handling of out of range on LightwareI2C Lidar
fixed init of HoTT telemetry
added support for mRo Pixracer Pro, Holybro Pix32v5
added arming check for terrain data healthy if needed
fixed compass ordering bug
increased monitor thread size to 768

Many thanks to those who have tested this release!

urlu75 · June 22, 2020, 10:11pm

thanks @tridge for your analysis.
That’s weird because I put all the .DAT file on the SDCARD manually and I’m sure that the area was covered. Maybe some files are corrupted. I will try with the beta2 to see if arming check reports a warning with the same setting.
Also what I was wondering, could it be an issue if terrain spacing is set to 50 on the FC and 100 in the .DAT file ?

Thanks and no worries for the place, only foam

Naterater · June 23, 2020, 3:31am

Just wanted to bring your attention on something I found when testing 4.0.6 beta: https://github.com/ArduPilot/ardupilot/issues/14639

tridge · June 23, 2020, 5:58am

I don’t see an issue report from you? I only see a “works for me” posting.
Can you post again?

urlu75 · June 23, 2020, 8:54am

1/ Upgrade from beta 1 to 4.0.06b2.
2/ Arming check warning about terrain waiting for data.
3/ Re download the same area with the last python script and copy the fresh data to SD.
4/ Arming check all good.
Thanks for adding this to the arming check list !

tridge · June 23, 2020, 9:43am

thanks for testing this, much appreciated!

tridge · June 23, 2020, 9:44am

yes, that would do it

Naterater · June 23, 2020, 10:41am

Sorry; https://github.com/ArduPilot/ardupilot/issues/14639

tridge · June 24, 2020, 3:00am

thanks Nate! I’ve reproduced in SITL. It only happens when MIS_RESTART=1 is set, which is not the default. I’ll work on a fix.

pompecukor · June 24, 2020, 6:39pm

Hi Andrew,

Is it possible that the behaviors for LGR_OPTIONS has been changed yet again? Maybe a behind the scenes rebase from copter? (like from here: https://github.com/ArduPilot/ardupilot/pull/13638)
Remember you fixed it a few months ago in PR, so we started using option 2. Now it is only deploying, not retracting.

Thanks
Alex

hwurzburg · July 5, 2020, 5:41pm

@tridge TX Tuning issue in master. went to TX tune roll on my C1 TVBS this am…here is log: https://www.dropbox.com/s/zr2wnzirms6krf4/2020-07-05%2008-26-50_badtxtuning.bin?dl=0

although TUNE_PARAM set to Roll_PI…it appears to be changing Pit_PI and reporting as such in the message log…there is even a param save, but nothing actually changed from my param before the flight when I compared back on the bench…twilight zone time…

Alexey_Degterev · July 6, 2020, 8:41am

Hello. Can vibrations at high throttle cause this crash?

here log of flight
https://1drv.ms/u/s!AuFPVI_eDaZbhMhjfNGQeLrVaB9-lQ?e=VgaUCc
100% crash happens if i launch by holding the wing. If i launch holding the fuselage it succesfully 100% even if there are the same vibrations
In this log first succesfully by holding the fuselage and two fails holding the wing
https://1drv.ms/u/s!AuFPVI_eDaZbhMhkWP2OP9S7mGKejg?e=A2VKIo

mboland · July 6, 2020, 6:13pm

You can even hear the vibrations in the video.

If your IMU’s are clipping it means they are over their maximum G rating and will return corrupted data = crash.

You just need to fix your vibrations.

tridge · July 6, 2020, 10:56pm

the compass is really way off. That needs fixing, or disabling. I suspect that is your main issue.

Alexey_Degterev · July 7, 2020, 1:28am

thanks for reply! Could you explain how you understood this? with this graph?

also bad compass can cause crash in fbwa mode? i think fbwa not using compass, only gyro,accelerometer and may be baro

Don_Gilles · July 8, 2020, 1:07pm

Didn’t get an answer in the ardupilot on planes forum so I’ll ask here:
What is the recommended setting is for compass_learn on a new matek F765 build running arduplane 4.06 beta2? I just installed a new BN880 (replaced my BN220 just because I had one and I like to tinker) in my mini-talon and was going through the settings and searching on this parameter I didn’t find any definitive answer.

0 Disabled
1 Internal-Learning
2 EKF-Learning
3 InFlight-Learning

Thanks,

lazydude · July 9, 2020, 5:04pm

Answered by Atx_Heli in the arduplane forum:
the default, which is none…only inflight can actually be used, the others are experimental…and only inflight if you cant possibly pick up the vehicle to do the compass calibration rotation “dance”, and even then the Large Vehicle Mag Cal section of the wiki is the preferred method…

Advanced User marked parameters should only be changed from default if you are one, and therefore know what they really do

Naterater · July 9, 2020, 6:11pm

Pretty interesting roll demand oscillation happened in 4.0.5 stable. Definitely not a good thing: https://github.com/ArduPilot/ardupilot/issues/14789