Autotune:0...Ground:0...Winner? It's tuning!

First - apologies for the poor video. I'm running a $30 gimbal on a standard Martinez without tuning, and a TBS with a mysterious vibe problem. I've left prop sound in, so you can hear the motors misbehaving.

Ok, my previous forays into autotuning have not had great results; today was better.

I ran autotune on my testbed, and it was done in about 3 minutes - super quick. Looked pretty tight in the air...until I tried the settings for myself. It was way too floppy, like the APM was loose or there was a drunk driving it (not me!). It was downright dangerous. The PIDs were way soft, so I tweaked them. Was much better after that. A few tiny tweaks to go, then I can start thinking about optimising the gimbal.

The Disco took slightly longer, but was strange - too aggressive on pitch, too soft on roll. Mad oscillations when in fast forward. Pitch P was tuned down, Roll P was tuned up, and D was tuned up on both. Much better. Now I just have to find that damned vibration...it's driving me mad, and the video is unuseable! Using a TBS Loveseat, but might try switching to moongel for a while.

I think I'll continue to use autotune, even after the crashes...it's not perfect (for me! Your mileage may vary), but it's definitely a quick way to get to 80% right, then it's a quick tweak or two to get to optimal. It's also a high risk way of highlighting any deficiencies in your airframe...

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • This gimbal was hard mounted and has low stack height so pendulum effects should be minimised. It found its settings in no time at all - faster than the non-gimbal tbs - so some gimbals obviously affect it more than others.
  • I thought the auto tune was supposed to be down with out the gimbals in order to have good results, has this changed now? can we tune with the gimbal on?

  • Excellent news Jason - this will be especially useful in video shots I suspect - I can't seem to get the balance between locked in Yaw with no bounce, and a smooth, slow panning (though, there's the probability I'm just a crap pilot!) - I can use expo admittedly, but this sounds like it will allow the same slow panning. I assume I can attach this setting to a dial, so I can fly fast, then slow it all down for a shoot? Will this apply to all three axis, or can each axis be controlled individually?

  • Developer

    We added a new param called RC_Feel to the latest code. The idea is to get the craft highly tuned and adjust this parameter to get a smoother or softer feel. That way you don't detune the craft to change the feel.

    Jason

  • I suspect they'll be better quality than the RC timers! Do as some have suggested - do it in stabilise, not alt hold, and it should mean you retain full control at all times. 

  • MR60

    I'm now afraid to try autotune on Mr Red (surname of my X8) because he's with SimonK ESCs (blueseries, may be better than the RC timers)...

  • lol Hughes.

    Cause 1 (test bed): still unknown. A battery failsafe engaged (outside of the failsafe range I might add), and the land event did nothing of the sort

    Cause 2: The RCTimer ESC's freaked out, possibly due to either a simonk issue, or just generally being rubbish.

  • MR60

    So to summarize your experiments for saving our multicopters (thank you), what were your identified root causes for the failed auto tune ?

  • Apart from the PID's it chose in the end. But one set of tweaks, and it was flying great. So I still consider autotune to be useful - I'm just not expecting it to deliver the "on rails" sensation.

  • Yeah, it crashed the TBS. Or rather...the cheap nasty RCtimer ESC's couldn't handle the autotune program. I replaced them with multistars, and it went flawlessly.

This reply was deleted.