Autotune quits 15 seconds into the second (pitch) axis

I’ve got a mini APM 2.6 board on a 250 sized copter and have been having a hard time getting it tuned in. Running autotune, it gets through the roll axis first which takes maybe 3 minutes, but then stops twitching only 15 seconds into the tuning of the second pitch axis. Additionally, the settings are not saved after disarming.

This behavior happens on ArduCopter V3.2-rc2, and now having the same problem with rc4.

[attachment=0]2014-08-07 19-17-32 24.bin[/attachment]

Also, the frame is extremely stiff, with 6mm thick carbon fiber arms and running HQ carbon reinforced props so there should be virtually no flex in this system.

See here what will solve problems on a Mini:

Thanks sevet. I had been reading about the changes Leonard was working on to get mini copters to tune better and was able to test it out tonight.

I ran Autotune on the new 3.2-rc5 build about half a dozen times and it completed without issue every time. I tried with and without changing INS_MPU6K_FILTER to 42Hz and it seems to record higher PID values when set to 42Hz, however the data is somewhat inconclusive with only a handful of tests.

Overall I feel the autotune results are very tame compared to what I get on my larger 400-size quad with 8" props. I increased ATC_ACCEL_RP_MAX to about 220K, and ATC_RATE_RP_MAX to 48K. The responsiveness to stick commands feels very crisp and responsive, although the stabilization feels a bit sluggish. It’s especially noticeable when dropping altitude fast. I will follow up with some manual tuning but it seems to be a good start.

Attached is latest autotune log, with increased ATC values.

Only thing I can suggest is trying manual tuning. Try increasing D until you get rapid oscillations, then back it off. Then increase P until you get oscillations, then back it off, then set I equal to P. Do this over soft grass (lawn).

See if you can get a better tune. I don’t have a 250 myself, so I can’t give much advice here.