So this is a mobula 7 running 3s, so it rips. Sometimes on turns with this particular quad, the quad picks up thrust as I am lowering the throttle. I thought this was due to the frame flexing and the FC trying to adjust due to that, but I strapped a carbon fiber bottom plate to the quad and it still seemed to give me throttle occasionally as I for instance would bank left. I can hear the motors spinning up as I drop the throttle. It's not all turns and I can't quite pin it down.
I am certain it is not loose parts or cracks in the frame.
Wondering if this is an airmode thing because I'm turning? Should I adjust something there?
Wondering if lowering my base throttle speed might be wise since I'm running the quad on a disproportionately large battery?
Thanks for your thoughts!

Airmode will allow you to maneuver the quad at low throttle, so that is normal. But just guessing, maybe your P terms are too high, you could try reducing that on roll. 3S on a 75mm is lots of power so I could see it overshooting on a fast roll. Too much feed forward might also do the same. Or if you have set too much ant-gravity or too high dynamic idle maybe that would also contribute. If you are applying sharp bust of yaw at the same time, that might also inadvertently over throttle the motors. If you have a blackbox log and dvr that may or may not shed some light.

give it a try on increased gyro filtering, move the slider a bit to the right. it might need some more filtering, some hardware related noise can come from mamy things - maybe filtering does solve it easy.

Thanks. I'll start with the P value and try lowering that. Unfortunately the symptom isn't always there and takes some hard flying to notice, so sussing it out may be tough. No blackbox for this drone, so i'll just be feeling it out. Dynamic idle is set to 0, so I'm guessing that's not it? Antigravity sounds like something that could cause an issue like this, but I don't know enough about it. Will look into that and see if adjusting that has an effect. Thanks for the pile of ideas to try!

sounds like i term windup, not pterm related

if it flew well befor with that tune, start by the filters and check the hardware if not successfull would be my way

I lowered the P value, but that didn't seem to effect the symptom. Then I adjusted I term and I do think that it dramatically reduced the uncommanded thrust inputs. The only time I encounter the problem now is when I am nose down diving, then the quad tries to throttle itself up and sort of floats down instead of falling. What to adjust now?
Thanks!

disable anti gravity feature and increase iterm again to get "stable dive"
whoops does fly nice without that feature - thats a possible way, but it seems like more of a plaster than a true solution.

im not sure if you run into a hardware related issue or just a tuning issue for using it with 3s.
did it fly nicely once with 3s and that tune which gave the "flyaways"?

why did you set dynamic idle to 0?
sure that doesnt match, dynamic idle should be adjusted by spinup the motors without rotors and do the math for the different load with props for the best results.

i would suggest to go int bfc, run command tasks in the cli tab and show us the result. maybe you just stress the cpu to much by features and processing speed?