Here 3 - z offsets

Hi everyone.

I’m using a HERE3 + BlackCube (with mini carrier board) + 4.1.0 and i’m always getting this:

  • ‘‘Test: Compass = FAIL - WARN: Large compass offset params (X:-95.43, Y:-147.41, Z:406.45)
    FAIL: Large compass offset in MAG data (X:-251.00, Y:-471.00, Z:97.00)
    mag_field interference within limits (20.46%)’’

Already did calibrations (multiple ones outsideand the result is ALWAYS the same in the Z).
In flight i must say too that even with the ‘high offsets’ the quad flies very, very good (althold, loiter, auto).


Any idea what could be?

Thanks as always.

edit. The GPS is way far from ESC and power boards:

1 Like

orientation of GPS and cube is different.
make sure that the arrows on them have the same direction.

Hi Mustafe,

They are in the same direction (used the Board Orientation (AHRS_ORIENTATION)).


Can you update the here 3 GPS and even if problem persists try calibrating Relaxed


The firmware is already in the latest version and sure i could try tomorrow to calibrate the compass in relaxed mode.

Would it change anything regarding the offset? I thought that ‘default’ and ‘strict’ would be better.

Yes definitely but if the system is generating warning regarding compass magnetic interference,try using relaxed see what comes next.

The thing is that i dont have any interference from surroundings.

Already did a compass mot and the result was good.

Use UAVCAN/External Compass as first priority compass.
Then reboot your vehicle, re-Calibrate compass in default mode.
I hope it will work for you.

Its in the first priority since the initial setup:

thanks for the tip anyway!

Could this be a faulty GPS?

Like i said before the quad flies really good ( anyone can see the log), thats why this is very wierd.

1 Like

Ignore that. Auto Analysis is long out of date. Some of the parameters it error’s on haven’t existed for years.

Thats what i thought that could be a false positive.

Is it possible to any off you guys download the ‘log’ and if there is anything wrong?

Anyway thanks to all for the quick insight and reply.

The only reason you would want to relax fitness is if the Calibration fails. There is no issue here, a compass Z offset of 400 is not a problem.

Nothing much wrong but you have Rate Roll FLTT at 20Hz with FLTD at 10Hz. I suspect you forgot to update that parameter as the Pitch filters are at 10Hz. It could explain the difference in the Rate Pitch/Roll P&I’s. Unless there is some mechanical reason why pitch and roll would be different these are usually close to the same. But it could be because you lowered Auto Tune aggression from default. You typically want to use default (.1) on this size craft. If you are going to try another Auto Tune it wouldn’t hurt to configure the Dynamic Notch Filter 1st.


I used autotune aggr 0,075, since i’m using 16’’ props and always used this aggr for this frame (tarot 650mm).

Regarding the Rate Roll FLTT at 20Hz, can i change it now to 10Hz without doing autotune again?


But even the largest craft you want to use the default aggression. If you want to read if from the Master he’s talking here about a large craft. Page down to Auto Tune.

1 Like

The parameter is ATC_RAT_RLL_FLTT ? Correct?

Regarding the autotune, i changed for 0.075 based on the wiki ardu setup (i think). Gonna redo the autotune with default settings .1 as soon as i can.

Well thank you DKE for the tips and i’m gonna read that post from the Master Leonard! thx for sharing it.

Yes. It would typically be the same as FLTD. But honestly I don’t know what the effect would be with it set like that.

1 Like

This filter is there to reduce noise coming through the EKF attitude estimate. There is the noise you see in your rate loops and the bias updates. As the EKF does not use the gyro filter it is effectivly unfiltered other than the nessisary intergrator to go from rate to angle. Generally this noise is smaller than peoples rate noise but on very nicely setup aircraft this becomes signiricant. So generally the impact of the FLTT filter is not significant.

So yes, this is a safe thing to do.

Definatly do this. You have a flying aircraft. Just do a 2 minute hover in loiter with pretty much no input in anything other than throttle with batch logging and you will get a good data set for your Harmonic Notch filter settings.

Yes, 10% is the right choice for pretty much all aircraft. An aircraft needs to be very low noise for anything below 5% to even be succsessful. The 10% is the overshoot pluss noise. And you want between 5% and 10% overshoot for the best disturbance rejection. So there is rarly any reason to change the aggressivness.

Well what can i say for the tips? Thank you both!

Very helpful.

Dont want to create a new thread so gonna write in here.

So did some real flights, auto, and i’m getting EKF_YAW_RESET and EKF_PRIMARY - 1 / 0 during the flight.

The flight was ok, altitude was unstable the rest looks good.

IVN,IVE,IVD values:

IPD values:


Both XKF3.IPD (Barometer) and XKF3.IVD (Altitude) are too high, right?

Any idea what could be?