Quadcopter - DIY

Hi,
it seems that we have a serious issue here, this glitch lead to a lost copter last week.
I thought a slight drift was no problem but when I was very far away I could not see the orientation and the drift took my copter.

@Alex we should take a look at this problem: Maybe it is a broken Sensor but it can also be the code, people are reporting it since 1.8.

I noticed an issue with acc and gyros in my old shrediquette project. We did fixed the noise of the ACC by correctiong it's values with gyro values:

I am not sure, maybe it is already done like this in the multiwii code:

The gyroscopes can only measure differences in rotational speed. Integrated over a long time (e.g. 11 minutes of flight) the angle calculated from gyro's alone loses precision. Here, the acc comes into play: It always knows the angle of the copter, but it reacts pretty slowly. And it contains quite some noise. We have to combine the fast signal of the gyroscopes and the absolute precision of the accelerometer. In the end, you get the best out of both worlds.


If someone has bluetooth it might be good to know it the angle is comouted wrong, so we know if there is a problem with the calculated angle via the acc...
I will let the copter hover in my basement for a few minutes and check the angle via bluetooth - I will report back if I find something unusual.

I think this is a priority 1 issue now!!!

Nils

Hi,
I think I now know why this problem could happen.
This is due to the difference between the configured acc1G value and the real value coming from the sensor when the multi is flat in flight.
I think the average value could be a little different and the current algorithm is very sensitive to this.
This explains also why this problem is not reported on every config.

I will post a patch tonight. It should correct every ACC drift problem.