Level drift - alarming angles etc - observation

[shikra]

I guess this is a good time to post...

I have a couple of boards that suffer from drift which I stopped flying until had time to investigate plus have been helping a couple of others diagnose their drift issues too.

In all the cases so far, the ones with bad level drift effect is caused by coming from the gyro....

By bad drift mean putting into level mode it tips at an angle and shoots off to the side. Sometimes the angle is alarming... Thing is - there guys are not noticing any significant issues flying in acro mode. Replacing the acc sensor made no difference, but replacing gyo has cured.


To test....
Calibrate ACC and gyro. Check good on GUI - gyros reading zero
Fly for 5 minutes
Connect up to the GUI - check the gyro numbers. If they are not zero when still then time to replace the gyro.

You might also notice the issue by just leaving for 10 minutes connected to the GUI on the bench, but flying seems to bring it out more - maybe its vibrations or something.

Another observation - WMP seems to be worst..Now I think its better to replace with complete IMU board. 6 out of 7 have been this now. Waiting to hear about the 7th...

Cheers,
G

[Noctaro]

Hey,
i posted some times ago in this threat ->
viewtopic.php?f=8&t=671&p=11157&hilit=accumulative+drift#p11157

I came to the same conclusion about drift coming from gyro. But i never was able to see this drift in gui.
Constant 0 when not moved. No drift after flight, and even not if connected around 20 minutes.
Until now, i do not exactly know where the drift came from. Replacing my WMP+ with ITG3200 did the trick.

greetz Noc

[copterrichie]

Being we are on the bashing of the WMP topic, I also have had a number of drifting issues and even inflight failures using the WMP. Never really fully understood what was happening, I figured it was because I was using 5 volts to power it.

[shikra]

Yes NOC I recall now. Re-reading the thread it was same issue.

Another thing I noticed - on 2 boards if its drifting bad in flight, some rapid left/right shaking of sticks seemed to "reset" the gyro centering for a while. Tried some moving average gyro center compensation code. It improved it a bit, but never perfected in flight as moved on to better boards until I received a couple of boards from a guy who was just about to give up! Couldn't let that happen!

[shikra]

So have another 2 controllers recovered from the bin with a new lease of life! I think that means another copter needs to be born!

[Sebbi]

Bias drift of gyros is a pretty common thing. It happens over longer periods of time which is why MultiWii is calibrating the gyro bias ( null level when not rotating) on power up. Drift can be random, but is most of the time a function of temperature. Sensors like the MPU6050 are mostly factory calibrated and compensate for different temperatures, etc automatically.

If we want MultiWii to compensate for older gyros (like the Wii ones), we would need a complicated calibration method (compare to http://autoquad.org/).

[copterrichie]

and the heat makes total and perfect sense!