I need help troubleshooting a quad. I'm getting severe shaking a mid-throttle range. It's essentially unflyable. Here is a video:



Here is a link to the log file:

https://www.dropbox.com/s/33sa8y35beqy62...4.BFL?dl=0

An good example of the issue is prominently visible at around 0:57



You can see big oscillations in the Yaw axis. My yaw authority is actually slipping at this point since motor 4 seems to be bottoming out. The shaking comes in suddenly and quite predictably around a throttle stick value of 1300 - 1400.

I think this might be a hardware issue. I'm thinking one of the ESCs or a motor. But I'd love to be able to narrow it down. This is my first time actually looking at a flight log. Is there a way to figure out what component is causing the issue? Or is there maybe even something else happening here?

Build

• Frame: Armattan Chameleon Ti
  
• FC: Holybro Kakutue F7 AIO
  
• ESCs: Holybro Tekko32 35A (4x)
  
• Motors: Xhover XH2207-2500KV CINEMATIC
  
• Props: Azure Power Johnny Freestyle
  
• VTX: Holybro Atlatl HV V2
  
• Reciever: Turnigy X6B
  

Context

As you can see these are supposedly quality components. This is my 3rd quad and it was supposed to be "The Good Build". To my ongoing frustration it never flew right. ESCs had problems from the beginning. I had intermittent issues flashing on one of them. The quad had motor de-syncs and dropped out of the sky. I kept working on it. Replaced one faulty ESCs. It got better. Replaced another ESC that was also suspicious and it maybe got better? But never good. Then I took a longer break. I got back into the hobby recently. And it seems it's a lot worse than I remembered. Wanted to finally get down to the bottom of this and make this one fly nice.

Help!

First good step - you didnt ignore it.

I havent looked at log, but huge vibration in video. Wow.

Im assuming you
1. checked each motor without prop for vibration on BF motors tab
2. made sure frame is tight
3. used fresh props, and props snug

(13-Mar-2021, 10:01 AM)romangpro Wrote: First good step - you didnt ignore it.

I havent looked at log, but huge vibration in video. Wow.

Im assuming you
1. checked each motor without prop for vibration on BF motors tab
2. made sure frame is tight
3. used fresh props, and props snug

1. Yes I did. Motors sound smooth on the bench.
2. It's solid. My other beater quad is a tangle of wires and hot glue in comparison.
3. It's using Popo props without nuts. When I do another test flight I can try adding nuts. But I have some experience with what happens if a Popo prop doesn't properly lock and that's not it.

If you haven't already solder in the biggest capacitor you can find room for. 1000uf if you can fit one.

Also, do you have rpm filtering functioning?

(14-Mar-2021, 05:33 AM)2000drz400e Wrote: If you haven't already solder in the biggest capacitor you can find room for. 1000uf if you can fit one.

Also, do you have rpm filtering functioning?

There are four capacitors on this thing - soldered directly to each ESC separately.

The quad is currently on BF 4.0. RPM Filtering was officially supported with 4.1 or so. I will flash it. But look at the oscillations. I'd be shocked if that's something software can smooth out.

New flight. Updated to newest Betaflight (4.2). Flashed ESCs. Turned on Bi-Directional Dshot and RPM Filtering. Changed to my Rates. But kept everything else at defaults. Running both loops at 8kHz. Also added Prop Nuts in addition to POPO as suggested.





Logs:
https://www.dropbox.com/s/05lylgn54o124l...6.BFL?dl=0
https://www.dropbox.com/s/nwdf75gyw3nf4w...7.BFL?dl=0

Takeaway: To my surprise the extreme Yaw Shakes did disappear. Makes me think that maybe they were due to some Dshot1200 issues? I had to step down to Dshot600 because that's how BF 4.2 now works.

However, the Quad still has significant vibrations I cannot account for:



Now across all axes and also appearing and disappearing suddenly. Not as bad as to lose control. Looks kinda like propwash just not in situation where you'd expect propwash.

I also decided to log out DSHOT RPM Errors to see if maybe there is some problem with an ESC. They are being tracked at the bottom in the debug graph. I am getting up to 10. Which as far as I understand is 0.1%. The Errors seem to be related to the Noise but usually appear after a delay. Making me think that maybe they are actually caused by the Noise? Not sure how to interpret this. Would love some feedback.

I did also notice that ESC 3 is running at a non-zero Error rate for longer. The other ESCs drop to 0 when there flying is smooth. ESC 3 has some more persistent, background Error rate. But then again, it's like 1 or 2 so 0.01%. Is that bad? Is this expected? Not sure what is happening there. 

Also, something weird happens in the second video. I still have some vibrations in the beginning. But then eventually the Quad stars flying more smoothly. And the RPM Errors also disappear for the rest of the flight?

This is all very confusing and frustrating.

What do would you suggest to try next to pin down the issue?

Ok nevermind. Flew the Quad again today. It's back to doing this again. I have changed nothing:

check your cpu usage in the cly with "tasks"
you might get benefits in performance by decrease to 4k/4k and dshot300.

tpa feature can help for midthrottle oscillations.

a capacitator id you didnt have allready.

(23-Mar-2021, 01:26 PM)hugnosed_bat Wrote: check your cpu usage in the cly with "tasks"
you might get benefits in performance by decrease to 4k/4k and dshot300.

tpa feature can help for midthrottle oscillations.

a capacitator id you didnt have allready.

Hmmm. This is an F7 FC. I checked "tasks" and and the total maxload was at 119%. Weird. I set it to 4hz PID loop and Dshot 300. Can't change the Gyro. This is what it looks like now.



Why is total maxload still so high? Is this normal?

As mentioned above, there are 4 caps on this - soldered individually to each ESC.

its normal, gyro looptime is fixed to 8k seems bf4.2

im not sure about, seems a huge gap between average load an max load. i will need to connect a build to bfc to do a comparison.
anyway 119% will decrease the perfomrnce for sure. the developers suggest to stay below 60% avg load (if i remember well). i would give it a go on 4k/dshot300

Ok so new flight. The changes I did:

• DSHOT300 at 4kHz PID
  
• Added extra vibration dampening to the stack
  
• Re-routed some cables just in case they interfere with the gyro or something
  
• Now logging at 2kHz and "Gyro Unscaled" in Debug
  

Results:



BFL: https://www.dropbox.com/s/av8g33ocdyp648...6.BFL?dl=0

The quad flies again and I don't see the twitching anymore. My running theory is that it was the controller lagging out due to too high cycle settings. But who knows, maybe it will start twitching next flight. Fingers crossed this was it.

Now that the big twitching is gone it still doesn't fly very nice. There are more issues:

• I see constant noise in gyro unscaled on roll. Gets worse with throttle. Pitch and Yaw are looking better. Seems like the filters are handling it. But where could this be coming from? Is this abnormal?
  
• I feel the Quad "slipping" on punchouts. If feels like it yaws or pitches.
  
• Generally despite using my rates the flight behavior feels off. I find myself mis-judging yaw behavior. I struggle finding the sweet spot when mixing the inputs.
  

Things are exacerbated because I also upgraded my radio. But I didn't experience the same flight feel issues in the sim so this is probably a tune problem. Any suggestions?

Unsolder all 4 motors.

Connect battery. ARM.
Hold in hand and make whooshie jet noises.

Does BB still show pitch noise?
You could have 1 warped unbalanced motor..