Vibrations induced after autotune

Ran autotune for the first time (V3.2 stabe). Was very happy with the new PIDs at first as the responses seemed crisp and fast. Then on the next flight I noticed the GPS-mast oscillating during hover. Checked the dataflash log and it indeed shows Gyro AccX and Y to be significantly higher than before. Confirmed this by looking at earlier dataflash files as well as by changing the PIDs back to their original values. See attached pictures.

My conclusion is that autotune may be a “mixed bag”. It produces super “crisp” PIDs (too much so?), but does so at the expense of vibrations(?). As my X8 is carrying a camera, this is undesirable which is why I intent to stay with the old a values. Possibly, I may choose to “soften” the autotune values. Besides that I do not want vibrations on my video, I’m also worried about the negative influence these vibrations may have on the flight performance as well as wear and tear.

Thoughts?

I have this problems too!
After autotune the PID values are much too high.
The induced oscillations are visible in vibrations analysis.

My conclusion is autotune is not working for my Y6B.
I switched back to default values and tuned in flight with CH6 Option.
Of course PID tuning is very difficult and i don’ t manage it very good yet.

Marc

Wow… just curious, how was your original pids vs autotuned pids?

Before and after PIDs:

What’s interesting when looking at the logs is that the induced vibrations seem to be of high frequency, compared to the normal “before” vibrations.

There’s an considerable jump from the original pids to the tuned ones.

My frame is a HK x650F with 800Kv and 12.5x6.5 / 4S , but a little bit heavy ( 2.2KG ) and my autotune results were less severe that yours .

Maybe your pids are high and you could be about to experience perceptible vibrations. I would suggest you to drop then around 10% and see if it helps.

For comparison, see my (yesterday) tuned values :

It’s been my experience that Autotune pushes the PIDs to the maximum region of safe flight which often leads to very touchy and overly responsive flight inputs. For this reason I agree with Tabascoz to detune the PIDs by 10% increments or until you find the sweet spot for your airframe.

I’m not in front of my Y6B at this time, but I do recall having similar autotune PIDs around 9.5 for Stabilized itch and Roll and that was too much. I believe I found my sweet spot 6.5-7.0.

My concern is that autotune seems to push PIDs too high. I like to see it perform a calibration from default factory to maximum safe and then provide you recommendations as to where you want the PIDs to be set at.

i already reported this way ago during beta testing, but it seems that the issue could not be found or it just not had any higher priority.
My opinion is that the old Autotune 3.1.5 was working very good on most plattforms, the PID’s were tuned already pretty good.
The new Autotune 3.2. on many plattforms leads to vibrations which will “eat up” the higher PID’s value estimated by the Autotune.
As i spend a lot of time trying different way’s to get good results i can gave the only advice, as workaround, to set down the D value’s until the vibrations are gone. Usually D values higher then 0.01 are unrealistic.

On your second printscreen we see the D Value for Pitch up to 0.02, i think that this is the problem. Try to set it down to 0.01 or slightly lower until the vibrations are gone.

i’m very happy that now more people are having same issue as this could lead to higher priority for this.

Hi,
Thanks for this. I have just found this and I have known that the perfect tune would result in slightly lower PID values. Recently I have developed a better understanding of how to do this and I hope we can do better in this area.

I realise this is a long time after your initial post but I am interested if you are still in a position to test this?

Thanks

Hi Leonard,
good to here from you, i’m ready to test newer versions as soon as winter is passing and the temperatures are growing over 10° :wink: