YAW drift in hexacopter

I rebuild my quadro to hexa and use the same Pixhawk. Quadro didn’t have any problems and works very well on 3.1.5. I also upgraded to 3.2 and now 3.2.1. My hexa flights also very well except YAW drift in CW drirection.

First thought: problems with mag, so i recalibrate mags and withdraw bat away from Pixhawk (second mag is with gps on pole far away from any magfield). Yaw drift persist, so i recalibrate CompassMot with props upsidedown without any effect.

Second thought: vibrations, so I added antyvibration plate to Pixhawk mount. Yaw drift persist.

I have no more ideas to remove this ugly CW drift, that start afeter some time in air so i decide to search for help here. The day the logs was recorded was very windy. Second log is also battery performance test, please ignore FS_BAT, becouse attopilot isn’t calibrated properly and battery capacity also. Please help if You can.

I can’t use your log files, I removed the last bit so it was just .bin but it wouldn’t open. Log files are normally .log or .bin without the extra bit.

But I suggest you look at RCout and see if clock wise motors outputs are different to anticlockwise by about the same amount (each clock wise about the same, anti clock wise about the same as each other)

If there is a difference then it is probably a twisted arm that is creating a yaw torque that exceeds what the brain can cope with.

Bin logs are compressed by 7zip because of size limit of attachments added to post. You need first uncompress them.

I don’t think that yaw torque is created by arm twist, there is steel plenty of scope in all six PWM rcout to operate. Furthermore I can fix yaw drift by rcin without any problem. I don’t think that you can fix yaw drift by rc if there is yaw torque created by arm twist. I read topic about yaw torque on this forum before I post my problem.

I also convert bin to log, compress by zip my 129.log and attached it to this post again.

OK that worked. Never seen a compressed file like that…one for the memory banks.

I have seen arm twist effects many many times on mine and other peoples logs, and it IS apparent in your log although it looks like cg is slightly off centre too (left/right), but the trend is quite obvious. probably a 1/3 to 1/2 of my posts answering seem to relate to this problem :slight_smile:

Plot RCOUT…Motors 1,3 and 6 (CW) have a higher demand than 2,4 and 5 (CCW).

3 is worryingly high! So you don’t have a lot of margin for safe flight in this condition.

Have a proper look, the best way I have round is to look down the arm with the propeller on that arm across ways,eyeball the tips against the adjacent arms using a common point on each side. spin the prop 180 to confirm (rules out a bad tracking prop, if the high side is the same amount either way then there is a twist).

It doesn’t take much twist at all. Rule it out, its easy to rule out.

These symptoms could be due to an odd esc calibration where coincidentally the cw are different to ccw. I have seen far more mechanical setup issues than any other that shows this effect with these type rcout plots.

I got the same that can’t be it response just recently! and it was :slight_smile:


Edit, additioanlly if you plot desyaw and yaw, later in your log you can see desyaw chasing yaw confirming its something applying torque turning the copter. (the software is designed so that desyaw only try so hard to avoid the copter springing back/getting confused). The reason it happens at the end of the log is there is less power to fight it as your battery voltage drops.

I confirm that this was arm twist. This problem is solved. Thank You very much for solution :smiley:.

no worries, glad you have it sorted