Problem with Sparky and Compass

[Jörg Rohde]

I have a problem with Sparky and also Quanton with the compass.
The compass in the GCS and also on my TX - its connected over HottTelemetrie - runs away. That could be 90° when I goes out. When I´m indoor the Quad is laying on the ground the compass in the GCS goes till 30-40° from the direktion.
i have made:
6 point cal., nose down cal., and set the home location manual - there is no GPS on board - it´s a small 200mm Quad.
Before I have it similar on an bigger one - running compassrose.
When I´m out I fly the quad direct in front of me, still standing in the air and I can see on my TX an misdirection that goes till 90°.
If it would be a problem with the power there has to be a misdirection but constant when I´m still staying in the air or depending on my throttle, so I know.
Could it be a tempreture problem? Between in and out here in Germany we have ~20°C.

Why I use compass without GPS? I want to test if I could navigate the Quad with compass when it´s a little far away that I could not see the direction from it to get it back.

[peabody124]

Without a log it's pretty hard for me to answer your question well. If you think it is a temperature problem we have temperature calibration: 

https://github.com/TauLabs/TauLabs/pull/333

20 degrees is a pretty good sized change

[Jörg Rohde]

Here is an log from the quanton board. It´s the same problem.
There was one rotation of the compasrose of the GCS but the quanton lays here still half an meter beside my Notebook.
The HomePos is set but no GPS connected.

[Jörg Rohde]

After making the log I let the Quanton connected with my NB. When I looked on the GCS at 15000 Flightime(seconds) the compass stands still in the right direction. Now after 54000 seconds its still pointing in the same right way.
I make an reset without disconnecting the board over the firmware tab from the GCS and it´s starts rotating again.
Am I the only person with this problem? With this phenomena?
Therefor the Spaky and the Quanton have differnt MagSensors it couldn´t be a driver problem or I´m wrong?
Can you say me where you compute the compassdirection from the sensor and the position?

[Francis Velasco]

i have the same problem on my FlyingF3, the mag sensor also drifts, but ever so slightly. ive tried recalibrating it and everything, but it still persists, but my quad has had several crashes, so i cannot tell you if its a software/driver problem, im fine with this little drift for now

[peabody124]

I'll try and look at the logs this weekend. Sorry I didn't have a chance today.

[Carsten Böhme]

Jörg, I still think you should check whether ARMING the board stops the drift. I've had several occasions where the board hat considerable yaw drift like in your case (50° per minute) while unarmed, which stopped immediately after arming.

[Carsten Böhme]

Besides, if you take a look at data objects, gyros and magnetometer, or configure scopes for magnetometer or gyros, you will notice that neither mag nor z gyro is actually drifting. The only value that is drifting is attitude actual, yaw.

[peabody124]

Looking at your log I see what you describe. There is a residual bias in the gyros (2 deg/s). I can also see the mag is having some effect since it slows down as it passes the correct heading.

As Carsten commented, I would expected after arming it would lock back onto the correct heading. Is that the case?

[Jörg Rohde]

Sorry that it last so long but I want to test it again.
Arming doesn´t change the wrong heading.
At home one Sparky stand still and the other was drifting east - the Quanton and a revo with TL drifts west.
When I was out for flying the Sparky point till 90° in the wrong direction - themperatur? Now there are 15°C different in/out.
Arming doesn´t change - throttle doesn´t change.

A other question, can you write in the app-store the version of Next/Master it´s working with when you make the next update?

[peabody124]

That's quite odd. Can you reproduce this without changing temperatures? Have you performed 6 point calibration?

Could you collect a log, keeping the board still on the table, and repeatedly arming it? Maybe arm/disarm it 5 times, let it sit for 10 minutes, then repeat the arming process.  Possibly it is related to the fact Germany has a much steeper magnetic declination, because Vinz has mentioned this before and I've never been able to reproduce it.

[Jörg Rohde]

Here is a fresh generated .log
Its a Quanton in the 3D printed case from carbo. So I could make a exact 6point cal.
The Q is placed on a wooden table, conncted with USB Labtop, 5V BEC, Graupner GR16.
You will see the compass running, before arming and after 10 min also. There was no temperature change.

[Lilvinz]

Magnetic declination at my place seems to be minimal:

http://magnetic-declination.com/Germany/Hannover

[Jörg Rohde]

You clicked here:
Latitude: 50° 18' 1.6" N
Longitude: 10° 12' 2.9" E
Magnetic declination: 2° 11' EAST
Declination is POSITIVE
Inclination: 65° 56'
Magnetic field strength: 48771.1 nT

[peabody124]

Ok that isn't a crazy declination. The problem does seem to come mostly in Germany, which is odd. Probably just sampling bias.

I'll try and analyze that soon but it might be this weekend - hectic week.

[Jörg Rohde]

Today I want to test the Bt Modul on my FF3 to make a todo/totest.
I took my quad with the FF3 on it and connect it to my notebook. Christmas or the week after I had installed the Master2312 on it with 6Point Cal and put the Gyrorate to 380Hz.
I don´t use the quad the past weeks why it´s a little big for indoor flying and when I go outdoor I use my small one´s too.
Today the direction of the FF3 was correct, the compass points in the exact direction and it was not running anymore like the other boards Quanton,Sparky, Revo.
But I killed the FF3, the CPU and now it´s only usable as a Stlink I order a new one and than I will test again.

[Jörg Rohde]

Next test: I took the new F3 out of the box put the Bootloader on it and then Master2312, make my setup and the 6 point cal.
Result, the board lays here on the desk without the copter, is the North direction differs 5° but no running of the compassrose only initialization of the sensors don´t work every boot.

The mag of Quanton and Revo are the same 5883 but the Sparky has 9150 with integrated mag. But the f3 has  the 303 mag - is the driver different from the other three?

[peabody124]

I'm really struggling to understand this from the log files.  When I look at some basic things, this is what I get:

So the bias is starting at the value of the gyros which seems like it is trying to account for things. As it is going the gyro bias is oscillating a good bit. However you don't see the same result in the gyros themselves, which is strange.  It is like your AttitudeSettings.BiasCorrectGyro is false, but when i look in your log that is not the case.

For laughs could you erase all settings and simply reboot, see what happens? It's strange because I've soldered quite a few Sparkys and don't tend to see this issue, but you and Vinz have both seen it.