Micoair H743 V2 harmonic Notch filter advise

ArduCopter Copter 4.6
,
1

I was unable to suppress the noise from the drone using the bdshot 4.6.3 firmware.
Does anyone know where I could have set it up incorrectly? I know a few axes where the vibration is above the recommended range. I will try to identify and work on that..

FCU = MicoAir743v2
[Edit]ESC = BLUEJAY 0.21.0, DShot 600
Motor = Skystars 1404 3800KV, Poles: 9N12P

INS_HNTCH_ATT,40
INS_HNTCH_BW,20
INS_HNTCH_ENABLE,1
INS_HNTCH_FM_RAT,1
INS_HNTCH_FREQ,77
INS_HNTCH_HMNCS,1
INS_HNTCH_MODE,3
INS_HNTCH_OPTS,2
INS_HNTCH_REF,1
INS_LOG_BAT_MASK,3
INS_LOG_BAT_OPT,4
MOT_PWM_TYPE,5
SERVO_BLH_BDMASK,15
SERVO_BLH_POLES,12
SERVO_DSHOT_ESC,1
1 Like
2

What frame did You use? BTW - vibrations are more related to frame than FC itself (except how it is mounted).

3

What frame do you use? Try something like that and lower your PIDs twice, but still it looks bad, you need to find the source of vibrations. Switching to RAW IMU logging can help to see overall picture of noise.

image
image1115Ă—1199 126 KB

INS_GYRO_FILTER,115
INS_HNTCH_FREQ,150
INS_HNTCH_BW,30
INS_HNTCH_ATT,30
INS_HNTCH_REF,1
INS_HNTCH_FM_RAT,1
INS_HNTCH_HMNCS,1
INS_HNTCH_OPTS,2
INS_HNTC2_FREQ,75
INS_HNTC2_BW,30
INS_HNTC2_ATT,40
INS_HNTC2_REF,0
INS_HNTC2_FM_RAT,1
INS_HNTC2_HMNCS,7
INS_HNTC2_OPTS,0
INS_HNTCH_ENABLE,1
INS_HNTCH_MODE,3
INS_HNTC2_ENABLE,1
INS_HNTC2_MODE,0

1 Like
4

Thank you, it has some improvements. At 23Hz, there is a spike.

INS_HNTC2_OPTS,2  # I have tried 2, multiple sources, result almost the same
5

image
image1293Ă—686 159 KB

There is so much motor noise that it’s probably creating too much for the filters. Drop your PID values (especially for roll) and see if that helps.

Try:

ATC_RAT_RLL_P, 0.101
ATC_RAT_RLL_I, 0.101
ATC_RAT_RLL_D, 0.003

Might be worth doing the same for pitch, but it’s not as bad.

1 Like
6

Don’t need second notch, the 23 peak you see probably is oscilation, drop your PIDs twice as Allister suggested. Enable fast instead of medium attitude logging in LOG_BITMASK.

7

The 2nd example today of the futility of configuring a notch filter when there is massive output oscillation present.

2 Likes
8

@Jai.GAY Users who bypass both the ArduPilot wiki and AMC get into this kind of trouble.

Pick one of the two and use it.

9

Apologies, earlier I have trouble confirming the ESC firmware type, I based on here say from a staff it was AM32, after much effort, it is confirm Bluejay and I help him upgraded to 0.21.0. I thought their drone could be overpowered, We did a test by putting dummy weight the motor RCOU “oscillation” become flatter and shift up to higher percent. I will give the PID values you suggest a try tomorrow, and very appreciate for looking into my flight log. I also feedback that their frame rigidity is one source of concern.

10

I will half the PID and disable second notch a try tomorrow as well.

Noted.

11

@Allister , @mike_tk , and those spending valuable time looking into my flight logs. Thank you so much.
The following values make the difference: the drone is now ready for Quiktune and Autotune. Vibrations are now limit to 15 m/s/s max. Before that, they will have to improve the frame.

ATC_RAT_RLL_P, 0.101
ATC_RAT_RLL_I, 0.101
ATC_RAT_RLL_D, 0.003
ATC_RAT_RLL_SMAX,25
ATC_RAT_PIT_P, 0.101
ATC_RAT_PIT_I, 0.101
ATC_RAT_PIT_D, 0.003
ATC_RAT_PIT_SMAX,25
ATC_RAT_YAW_SMAX,25
INS_HNTC2_FREQ,84
LOG_BITMASK,180221
PSC_ACCZ_SMAX,25

image
image1279Ă—868 18.4 KB

image
image515Ă—377 4.87 KB

image
image515Ă—377 4.86 KB

image
image515Ă—377 4.48 KB

image
image515Ă—377 4.19 KB

12

image
image1236Ă—667 24.7 KB

image
image1280Ă—447 32.8 KB

Use the web tool for the filters. You’ll be able to get better results.

1 Like
13

The fixed notch filter does not need multi-source.

2 Likes