Got me a Micro Apex frame, you can check out all the build details at https://rotorbuilds.com/build/27939

Building it was quite frustrating at times, but tuning it was a breeze because the Apex frame really has very little resonance. Here's the unfiltered roll gyro graph:

   

You can see there's a nice "valley of silence" from 50 to 150 Hz, which is great because trying to filter anything there would kill the propwash information and/or add big delays.

There's a single peak of noise at 270 Hz and even though the standard filters would crush it perfectly, I still added a manual notch filter (center 270 Hz, cutoff 250 Hz) which doesn't add much delay at all. This way the dynamic notch filter is free to kill any other (temporary) peaks.

So with the gyro lowpass filters disabled, the manual 270 Hz notch filter, standard RPM filters and D-term filter slider at 1.5 I'm getting a really clean response. For the PID tune I just put the Master slider at 1.5 and FeedForward at 1.2, nothing more. This results in great stick tracking:

   

And this is what it looked like while tuning:




Hope to get some more flight time soon, I need to get used to my higher Sbang Sbang rates!

May I ask what version of betaflight are you using?

If I try anything close to your pid and filter values the kwad just flies to the moon. Betaflight 4.2.9 iflight twin g fc, tekko 32bit esc, BetaFPV 2004 3000kv 4s

Need to figure out what I did wrong

Hi ojwright845

I'm also simply on BF 4.2.9. There could be lots of things that increase the noise and make the PIDs go crazy;

• Did you use damping grease like I did?
  
• Is the RPM filter configured correctly? Did you set the correct number of magnets on the bell?
  
• Did you make sure there are no wires running over the gyro(s) or near them?
  
• Is the FC not mounted too tightly or loosely? Nothing else that can bounce around or vibrate too much?
  
• Perhaps the motors are less smooth, a little unbalanced or the wide stator has more inertia?
  

Also, if you're using BLHeli32.8 you should know that there's a bug where it doesn't report the RPM as precisely as it should. So you may want to go back to 32.7 to be safe or try the 32.8.2 test version which has just been released. On my Roma F5 I haven't noticed much of a difference, but I guess it may affect a tune that's right on the edge.

Could you post your blackbox logs with gyro_scaled info? Is your noise graph as clean as my first picture?

If your gyro data does look clean, it may simply be that the your motor and prop combination need lower PIDs to get the correct tune.

Another thing I just thought of: did you disable D-min?

Because I didn't clarify in my first post that I did disable it. If you leave it enabled, the D-max is higher than the fixed D value when you disable D-min. So when you have a bit of noise the "smart" D-min code could think you're in a propwash situation, pump the active D value up to D-max, which causes even more noise in a feedback loop and there you have your rocket to the moon...

(01-Sep-2021, 12:04 AM)L0stB1t Wrote: Another thing I just thought of: did you disable D-min?

Because I didn't clarify in my first post that I did disable it. If you leave it enabled, the D-max is higher than the fixed D value when you disable D-min. So when you have a bit of noise the "smart" D-min code could think you're in a propwash situation, pump the active D value up to D-max, which causes even more noise in a feedback loop and there you have your rocket to the moon...

Totally missed your reply....but thank you so much! Actually you saved my quad. I flashed BLHeli 32.82, disabled Dmin, and all the noise is gone. No more flying to the moon.

Hey I just built one of these pretty similar to yours it has the mamba mk3 f722 basic mini with f40 esc ot has fpvcycle 21mm 2300kv motors dji and crsf I can't seem to get it to fly right it's a clean build but it sure doesn't act like it if I cruse it seems ok but I can see jello and on hard throttle pumps I get a twitch or jerk I'm running all 4.2 default pads and filters I enabled bidirectional dshot rpm filtering but haven't adjusted anything bbl shows alot of vibe on the pitch axes but the others are fine could maybe u share your bf and blheli setup with me or offer a suggestion it's my first 4" build but I have built and tuned several other 5" quads thanks man

Hi joey

On default pids and filters it should always fly pretty smoothly, so I'm guessing something is wrong mechanically or electrical. Check your 4 motors individually in the configurator - do they spin at a similar RPM at certain throttle positions? Check without props or with reversed props (so they push the kwad down) but be very careful.

Also check the gyro in the configurator sensors tab. I've had FC's before where one axis didn't react quite right...

Well I've checked and rechecked this is like my last option I've repositioned wires checked and rechecked there are still spikes in the pitch gyro traces everything is new one of the motors does sound a little less smooth than the others but not putting out a lot of vibration not enough to really effect it I wouldn't think

The vibrations are only on the pitch axes so there has got to be a reason why what would be your best guess

I wish the fc fit onto the screw a little looser than it does think that could be it maybe if fits too tight but you would think that would effect all axes sorry for rambling btw

Really weird that only the pitch axis is problematic. Bad gyro is suspect #1 to me. Maybe you could try and rotate the FC 90 degrees, change the gyro orientation in Betaflight and do a test flight like that. If the problem is then on the roll axis you can be 99% sure it's a bad gyro and not a motor, esc or mechanical problem.

That's true the thing is I had to mount it backwards so 180 already be cause that's the only way it would fit it is a mamba stack I just got it bad gyro was my first thought I already contacted raceday I'm just trying what I can sorta wishful thinking even tho we both know I've tried everything to little or no effect hopefully they will make it right I've never had an issue with them before I ordered a tiny hawk Fram from them one time and I guess it got smashed in route they just sent me another one even tho I'm sure it wasn't they're fault any way thanks for giving me your opinion yeah the spector graph in Blackbox explore looks really good on all the axes except the pitch I'll try to include some picks

You can see which is pitch I'm sure

Ok I guess I can't do it from my cellphone for some reason