Need help! Severe shaking on roll axis


First of all - i did not make a build in 3 years, so a lot of practical knowledge (of what little i have had) evaporated… But, i got to say, i never had such a problem before.

It is a small build, 250x300mm X - Armattan chameleon LR 7 inch frame. It uses 30 amps Armattan BLheli_S ESCs with Velvet 2300kv motors.

FC - pixracer, with some additional stuff on board - gps, lidar, runcam, etc.

i get and cannot get rid of the severest case of roll axle shaking - here are link to logs before, i do not want to pollute this request with my assumptions, i tried to recover pids to what they supposed to be originally as well, i cannot figure out what is the issue.

Log has alt hold and stabilize modes samples. alt hold with lidar was sitting very well at same altitude, 0 issues holding alt from what i can see. motors were oscillating, and after this loiter motors were crazy hot.

it was done with 6 inch 3 blade props, as i suspected 7 inch props were too overpowered - with 7 inch it hovered with 3A current, with those 3 blades it hovered at 6A current.

Could somebody smart please look at this and indicate what can be done here, if anything? it does not seem it is even able to get to do autotune because of this shaking.

Responses to sticks were very good and smooth. Holding direction was good. I just cannot figure out what is going on with this shaking, all my attempts to play with PIDs produced 0 result so far, i am puzzled as i do not know what i am doing. :slight_smile:

autoanalyzer output is below:
Log File C:\Users\paul\AppData\Local\Temp\tmp952C.tmp.log
Size (kb) 7072.115234375
No of lines 92459
Duration 0:04:17
Vehicletype ArduCopter
Firmware Version V3.5.5
Firmware Hash 27229c83
Hardware Type
Free Mem 0
Skipped Lines 0
Test: Autotune = UNKNOWN - [?] Autotune 43006-71292

Test: Brownout = GOOD -
Test: Compass = FAIL - FAIL: Large compass offset params (X:-492.23, Y:-501.53, Z:918.50)
FAIL: Large compass offset in MAG data (X:-492.00, Y:-501.00, Z:918.00)
Large change in mag_field (40.91%)
Max mag field length (643.36) > recommended (550.00)

Test: Dupe Log Data = GOOD -
Test: Empty = GOOD -
Test: Event/Failsafe = GOOD -
Test: GPS = FAIL - Min satellites: 0, Max HDop: 99.99
Test: IMU Mismatch = GOOD - (Mismatch: 0.14, WARN: 0.75, FAIL: 1.50)
Test: Motor Balance = GOOD - Motor channel averages = [1293, 1288, 1291, 1299]
Average motor output = 1292
Difference between min and max motor averages = 11
Test: NaNs = GOOD -
Test: OpticalFlow = FAIL - FAIL: no optical flow data

Test: Parameters = FAIL - ‘THR_MIN’ not found
Test: PM = GOOD -
Test: Pitch/Roll = GOOD -
Test: Thrust = GOOD -
Test: VCC = WARN - VCC min/max diff 0.441504v, should be <0.3v

For comparison - i just placed 7" props on and did alt hold flight. it had problems keeping stable altitude as i think it hovers at the min motor speed. shaking is there but frequency is way lower than with 6" props, motors also very warm but not burning hot like with 6" props. Obviously, problem is the same…

Anybody? Pls?
Any ideas about possible origin of the issue?
I found this post, OP says he replaced pixhawk and it went away. Looks like same oscillations i see in my RCOU channels, 2 to 3 hz nonsense that heats up motors.

I just undid this frame and placed into there a usual betaflight fc - soldered it to same escs, motors, same power source used for pixhawk - it lifted up and flew like a dream no issues. Placed back pixracer - same oscillations. It got to be something software related. Pixracer now sits on hard plastic mounts with silicon thin inserts in mount holes on 2.5mm steel bolts, it is literally almost part of a carbon frame now with minimal protection. Before it was on semisolid rubber mounts - so i removed them thinking it may have been in resonance.
What is this thing? Vibrations seen in log are almost none. Nor sure what to do next here.

Your motor output is weird ,all have a limited travel, i.e. RCOU.C4 goes from 1170 to 1500

Looks to me that the parameter MOT_PWM_TYPE is wrong , it is set to 6 try to set to 0, normal.

It could be also an issue related to Esc calibration.
I would go for software Esc calibration with Blheli suite.

It was left by mistake as i tried to set 3.6rc2 on dshot - it was same on it, so esc protocol seems not to be a factor. I then altered it to 0, recalibrated escs and got (with 7 inch props) a carbon copy of log i had with 6 inch props - same 3hz or so jigsaw sent to motors. Only improvement i got after that - it now keeps altitude fine on 7" props. Oscillation still there but less severe than on 6".

Again, i just typed that - i just flew this model on betaflight fc. It is not escs or motors, it worked fine on betaflight.

Btw here is a link to a similar, may be, issue. Not sure. Sounds like my build - an overpowered light racer.

Did you check with Blheli if the esc has the exact calibration range ?

Not yet bur it is not escs, i am pretty sure. Will try some time next week.
In esc calibration and motor compass test rotation of motors is perfect, response to rc throttle is smooth and is visually same as on any other model i made. I had one model before with esc issues and it was visually apparent in any scenario.

Odd stuff. It is a pity it gives me such a hard time, it really worked out nice with this 7" chameleon frame - 2 full 35mm stacks and transmitters in the tail, 2 gps, lidar and all works fine except it refuses to fly :slight_smile:
Such a minor bummer. I can of course alter it into iNav build, but it is not nearly same as arducopter.

You do not get the point .
If the Esc have not the same calibration, lets say one has 1000 - 2000 range while another has 1050-1950 especially with overpowered copter you will have a severe rapid oscillation even if taking the motors one by one they run just perfect.

So, issue it resolved, flight is smooth as silk now on same hardware.
The thread below was the key, it was all about params.

The key was lowering P gain, But and a big but, the mission planer extended tuning does not allow it to be lowered any lower than 0.080, and it needs to be around 0.050.

Go in to full parameter list and tune these: