AutoTune Issues

Log Analysis -- deprecated Copter Logs -- deprecated
1

I have a new quadcopter which I first flew on Sunday with 3.0.something, and upgraded last night to 3.1 so I could try autotune. The quad is not super well tuned, just enough that it wont shudder about when in stabilise mode, however when accelerating or decelerating rapidly it would wobble/shake quite a bit for a second or two.

There was no wind last night, so I rushed over to an oval before it got dark to try autotune. I had only ever flown the quad in stabilise mode, so I had no idea how well alt hold would work. It turns out, not so well… it was very shaky you could hear the motors constantly changing speed. I switched the quad into auto tune however it would fail after a few rolls to either side.

For those who read this, I found that AutoTune Failed would display when my tuning was just too terrible to hold the quad still. I dropped the alt hold PID from 1.0 to about 0.85 and it made it better, but still shaking about, i also dropped all my other P values for rate and stabilise pitch/roll. AutoTune now ran.

AutoTune however could not figure out the pitch values. It tried and tried and tried. My voltage started rapidly dropping, going from 13.8v to 10.6 in a matter of seconds - one of the cells in my Zippy Compact 5800mah 4S had died… completely died. Luckily I had two batteries in parallel, I was able to land it however I never had any tuning results as AutoTune failed. Given that I put 5560mAh into the surviving battery I don’t think I could have continued for much longer. The quad had spent around 20 mins tuning, most of which was spent on the pitch.

Is there anything I can do to reduce tuning times? I have a new 5800mah battery on the way which should hopefully arrive before christmas.

I know that I need to reduce the alt hold P further so it doesn’t work so hard, but beyond that I’m stumped as to why the pitch couldnt figure itself out.

Is it possible to get the values which it calculated for roll, or where it got to on pitch? It was starting to look quite good I thought and if possible I’d like to start autotune next time from where I left off as I figure this will reduce the tuning time considerably?

I’m flying a Hobbyking H4 with D3536/8 1000kv motors, 2x 4S 5800mah packs, Afro 30A escs, APC 10x4 props, genuine APM 2.5 (onboard mag) powered by the current sensor/power module thingy. The batteries are both running down the roll axis (so from front to back) which would make the pitch much slower to respond than roll due to the extra mass/inertia.

My log file for the flight is available at: rris.com.au/logs.zip

2

good that you included logs, thanks!

Normally altHold not working is due to vibration although it doesn’t appear too bad at the beginning of your log.

What I do see is copter is quite overpowered because it seems to be hovering at about 25% of throttle. If that’s correct then you can certainly put a bigger battery on it, it may even fly better that way.

With a very overpowered copter you’ll generally need to reduce the gains. So I suspect your Throttle Accel P could be reduced to 0.5 (from 0.75) and I reduced to 1.0 (from 1.5). That might help a bit.

The copter’s Rate Roll/Pitch P values are far too low. They’re down at 0.0385…I’d bump them up to 0.08. You can probably make Rate Roll/Pitch I the same.

3

Thanks for your response rmackay!

I tried again, before you posted. Logs at:
rris.com.au/logs2.zip

This attempt would not do more then 2-3 iterations before giving in. I had a light wind tonight, perhaps 1.5-2.5m/s but not constant. I was running autotune when there was a gap in the wind mostly.

Yep, very overpowered… my eCalc calculations were way out! I’m thinking of going to 10x3.8 props as I have some of these already.

I currently have two 5800mah 4S batteries on it, I don’t actually have any more and I’m not sure I could fit them if I did - I have 3 new batteries on the way (to replace the one which had the cell failure yesterday and give me 2 flights worth of batteries)

I tried throttle P at a range from 0.5 to 1.0 without any audible difference to the pulsing. I dont know if this is in the logs, I tried to add “PID” to the log through the console however I dont know if this actually did anything you can see. I did not try changing the I term.

I tried changing the Alt hold P term from 0.5 to 1.0 but it didnt do anything except make it lose altitude quite a bit hehe.

When I was initially testing inside, with Stab values all at 4.5, I couldnt go any higher than 0.0385 on the P term of the rate pitch/roll before it would oscillate wildly and I was unable to takeoff.

I’m a complete newb when it comes to Quads, I’ve flown my TRex 600 for years, and many many fixed wing planes for a decade or more… but PID tuning is all new to me so I’m very grateful for any guidance you have.