Failed autotune, almost crashed

Hi everyone,

I had an issue yesterday whilst running autotune on roll and pitch. The copter completed the roll axis but failed the pitch axis almost immediately, ending up in major oscialltions after the 2nd twitch and descending rapidly. Luckily the only thing that broke was a carbon fibre tube on the leg.

Could anyone have a look at the logs and let me know if there’s anything obviously wrong? I have noticed an imbalance as left side motors (2 & 5, 3 & 8) are not working as hard as the right side motors. I’m not sure whether that is within allowable limits.

Any other suggestions would be greatly appreciated.

I am running a running a 1400mm X8 frame with a pixhawk cube black. Motors are T-Motor u8 lite 100KV, props are T-Motor 29.2 x 9.5. ESCs are Flame 60A HV, running at 12S.

Log:
https://drive.google.com/file/d/1LEYuXWMlcU-70s0q1fXu-3nN5jAGrR_h/view?usp=sharing

Thank you

Did you follow the required tuning instructions BEFORE doing the autotune?

That is mentioned in the warning at the very top of the autotune documentation page.

It was a bad idea to fly that craft on default parameters let alone run Autotune. And that flight log is of little value.

Don’t understand much from the log. I guess it is not the one from the autotune.

Really sorry, uploaded the wrong file. Correct log is here:
https://drive.google.com/file/d/1LEYuXWMlcU-70s0q1fXu-3nN5jAGrR_h/view?usp=sharing

I have followed the tuning guide, parameters were not default, apologies for any confusion. The copter is flyable in all modes, I’ve done a yaw autotune before with success and good results. The pitch and roll are twitchy, hence the recent run of autotune. I have played around with ATC_INPUT_TC to smooth rc feel, which does make some difference but I think it could be better.

Ah well. The 1st thing I do is extract the parameter file from the log hence the comments about default parameters that don’t apply. I don’t see anything obvious other than a twitchy tune. Perhaps lower the Autotune aggression further and try again.

Yes, sorry about that! Okay, I will give it another try with a lower Autotune aggression. Just wanted to check if there was anything else obviously wrong that could have caused it. Thank you for having a looking through the log.

On a related note, do you have much experience with manual tuning? If so, was it just a case of fly in stabilise and do some manual twitches and land and check logs for des roll/pitch vs gryo roll/pitch?
Thanks!

Yes. If the craft is basically stable I start with Rate Roll/Pitch kP on a Live tuning pot centered on where it’s at now with maybe a 50% increase/decrease range. Then from a rough Hover give it some aggressive pitch/roll in Stabilze at the mid setting. Switch to Loiter for ~10s or so, adjust the pot to the low setting switch back to Stabilize and repeat. Set the pot to the high setting and repeat. Then in the log you can see the Stabilize sections identified with Loiter between them. Look at Des/actual roll/pitch, change the P value to the best and make I equal and do it again. This is time consuming.

After that move to Stabilize Roll/Pitch kP. Then maybe back to Rate Roll/Pitch kP.

2 Likes

Thank you for that, I may give it a try if I find autotune at 0.05 isn’t providing acceptable results. Am I correct in assuming once you have Rate Roll/Pitch kP where you want it, you then adjust Rate Roll/Pitch kD?

That’s an interesting question. Some start with D. On small quads that on the 1st flight are an oscillating mess you have no choice but on larger craft I have done D last.

That makes sense. Thank you for all your help, much appreciated :smile: