GPS Compass Error whilst in flight

[BudWalker]

Hi Bud,
Thanks very much for the analysis. Some of it makes sense, however some is beyond the level of my knowledge. Lol
I just want to note, that when I got the compass error and yawned the aircraft to return, the camera gyro had some massive roll (about 20 degrees), then eventually levelled itself out after about 20 seconds. Not sure if this is anything to do with false roll and gyro values.

Do you mind if I send you my. DAT file from the last flight I did where I didn't have an issue? I'll have to send it to yo in Thursday as that's when I'll get home from work. The funny thing with this one is that I did not have any errors or indications when I flew near the power lines, however when I uploaded my file to healthy drones, it shows the track where I flew near the power tower (W-X) as red weak signal.
On the map you can visually see the power lines, however the red weak signal only came when I flew over the power tower. When I flew near the actual lines (D) it doesn't show any signal degradation, all green.
Here is the link to healthy drones:

HealthyDrones.com - Innovative flight data analysis that matters

Cheers


Sent from my D6653 using PhantomPilots mobile app

The roll and gyroX values aren't false. They reflect what actually happened. The response to that is what caused the compass error.

I'm always ready to look at another .DAT. Especially if someone else goes to the trouble of retrieving and then providing a Dropbox link. I'm not sure what you want me to look at though. DatCon isn't smart enough to know about uplink or downlink signal strength. I think HD determines signal strength from the number of packets per unit time arriving at the RC. I tried doing something like this DatCon but it never worked very well. The Litchi app provides a .csv that has both uplink and downlink signal strength. But, I haven't been able to find that in the .DAT.

From my own flying experience there is a degradation in the video downlink when the the P3 is close to the power lines near my house. I think this is probably due to the corona discharge on the power lines.

 

[flyNfrank]

The compass error in this flight is even more of a mystery than @robinb 's first flight (FLY135). I checked all of the values that DatCon knows about. The only thing I found was abrupt roll and gyroZ values at the time (85 secs) of the compass error.
View attachment 64556
I've seen this in other flights where gusty wind conditions generate sensor values the FC didn't predict. The roll and gyro values were not accompanied by stick inputs. However, in those flights the sensor deviations were much larger. Although in this flight there were other instances with large values for roll and gyroX these happened while the A/C was on the ground and not moving. BTW, I ran DatCon at 200 Hz and submitted that .csv to CsvView.

Believe it or not I had a flight very similar to this. The compass error was accompanied with an abrupt change in roll and gyroX values. And, it occurred close to some power lines. I have since flown close to those power lines many times attempting to recreate the problem. Nothing - I've had to conclude the power lines were just a coincidence. I've not had an unexplained compass error in the roughly 90 flights since that incident.

My advice? If this is an isolated incident treat it as a false positive as I described in post #26. But, if it's a common occurrence maybe it's time for a trip to DJI repair.

Maybe the difference being the one flight there was X-amount power surges traveling through the lines, and on the other times near the power lines there was no duplicate surging taking place??

 

[BudWalker]

The roll and gyroX values aren't false. They reflect what actually happened. The response to that is what caused the compass error.
.....

@SuperRC I was a bit hasty with this response. You're right. How do we know that these values are a reflection of what actually happened? Why isn't possible that the gyroX sensor had a hardware glitch? My response was based on my preconceived idea that the sensors don't give bad data. So I looked at the magnetometer data. The AC was facing -9 degrees magnetic and the geomagnetic inclination is -64 degrees. So if the AC rolls it would be expected that there would be a change in magY as well as magZ This what happened.

I'm pretty sure these values are what actually happened.

 

[SuperRC]

@SuperRC I was a bit hasty with this response. You're right. How do we know that these values are a reflection of what actually happened? Why isn't possible that the gyroX sensor had a hardware glitch? My response was based on my preconceived idea that the sensors don't give bad data. So I looked at the magnetometer data. The AC was facing -9 degrees magnetic and the geomagnetic inclination is -64 degrees. So if the AC rolls it would be expected that there would be a change in magY as well as magZ This what happened.
View attachment 64598
I'm pretty sure these values are what actually happened.

Thanks for the reply Bud.
But unfortunately I don't really know what those parameters and figures mean. Can you maybe explain it in more layman's terms?

 

[BudWalker]

Thanks for the reply Bud.
But unfortunately I don't really know what those parameters and figures mean. Can you maybe explain it in more layman's terms?

Sure. The question is - are the gyroX and roll values real or was there some hardware glitch that caused those values. I.e., did the AC really roll. The magnetometers are a different set of sensors that can be used to determine if the AC rolled. The magnetometers and gyros are independent. If the magnetometers also showed the AC rolled then it's a good bet the AC actually rolled.

The heading was -9 degrees - lets say the AC is headed 0 degrees magnetic just to make it easier. That would mean magY (the magnetometer aligned along the Y axis; i.e. aligned left to right) would be 0. This is because magY is perpendicular to the geomagnetic field. The geomagnetic inclination at the site is 64 degrees - both the X and Z axes are not perpendicular to the geomagnetic field. This means that magX and magZ will both have non zero values. Now suppose the AC is rolled 90 degrees, i.e. the Z axis and Y axis are swapped. This would cause the values of magY and magZ to be swapped. The AC rolled less than 90 degrees so magY and magZ would not totally swap values. But, they will change if the AC rolls. The exact formula for the new values of magY and magZ likely has cosines, sins, etc. The important point is that magY and magZ will change if the AC is rolled while headed close to 0 degrees. Since this happened it's a good bet that the gyroX values are correct and the AC actually rolled.

 

[SuperRC]

Sure. The question is - are the gyroX and roll values real or was there some hardware glitch that caused those values. I.e., did the AC really roll. The magnetometers are a different set of sensors that can be used to determine if the AC rolled. The magnetometers and gyros are independent. If the magnetometers also showed the AC rolled then it's a good bet the AC actually rolled.

The heading was -9 degrees - lets say the AC is headed 0 degrees magnetic just to make it easier. That would mean magY (the magnetometer aligned along the Y axis; i.e. aligned left to right) would be 0. This is because magY is perpendicular to the geomagnetic field. The geomagnetic inclination at the site is 64 degrees - both the X and Z axes are not perpendicular to the geomagnetic field. This means that magX and magZ will both have non zero values. Now suppose the AC is rolled 90 degrees, i.e. the Z axis and Y axis are swapped. This would cause the values of magY and magZ to be swapped. The AC rolled less than 90 degrees so magY and magZ would not totally swap values. But, they will change if the AC rolls. The exact formula for the new values of magY and magZ likely has cosines, sins, etc. The important point is that magY and magZ will change if the AC is rolled while headed close to 0 degrees. Since this happened it's a good bet that the gyroX values are correct and the AC actually rolled.

Thank you very much for the detailed reply. This makes more sense. This is some pretty high tech stuff lol. I'll just try to stay away from any power lines.

Sure. The question is - are the gyroX and roll values real or was there some hardware glitch that caused those values. I.e., did the AC really roll. The magnetometers are a different set of sensors that can be used to determine if the AC rolled. The magnetometers and gyros are independent. If the magnetometers also showed the AC rolled then it's a good bet the AC actually rolled.

The heading was -9 degrees - lets say the AC is headed 0 degrees magnetic just to make it easier. That would mean magY (the magnetometer aligned along the Y axis; i.e. aligned left to right) would be 0. This is because magY is perpendicular to the geomagnetic field. The geomagnetic inclination at the site is 64 degrees - both the X and Z axes are not perpendicular to the geomagnetic field. This means that magX and magZ will both have non zero values. Now suppose the AC is rolled 90 degrees, i.e. the Z axis and Y axis are swapped. This would cause the values of magY and magZ to be swapped. The AC rolled less than 90 degrees so magY and magZ would not totally swap values. But, they will change if the AC rolls. The exact formula for the new values of magY and magZ likely has cosines, sins, etc. The important point is that magY and magZ will change if the AC is rolled while headed close to 0 degrees. Since this happened it's a good bet that the gyroX values are correct and the AC actually rolled.

Sent from my D6653 using PhantomPilots mobile app