Rudder at climb

[Yuri Pederi]

Hi all,

We installed engine gearbox with bigger prop to increase UAV thrust in the hot weather.

 

Today we had strange anomaly of autopilot behavior. During the climb autopilot also increased rudder and UAV started to slip and became unstable so we had to take over in manual control. We suspect that it might be driven by wrong sign in Rudder from Y acc. PID or wrong measurement of the yaw accelerometer.

 

I'm attaching the files. Please check the log file log20151019_174331 at 159-165 sec., log20151019_162350 at 61-67 sec and 89-94 sec.

 

I'm not sure what drives the rudder behavior. Compass is disabled. Could you please help me to understand the root cause and suggest the correction actions?

Regards,

Yuri

[Kevin Surminski]

Hi Yuri,

In flight, the rudder from yacc loop is active. if you plot the throttle versus y accelerometer there is some correlation between high throttle and high y accelerometer value so when the throttle goes up the vibration increases and that causes an accelerometer bias. Your next course of action is to improve the vibration damping of the autopilot. We have had good results using Enidine wire rope isolators. http://www.enidine-aviation.com/Products/Wire-Rope-Isolators/

Regards,
Kevin Surminski
MicroPilot Support

[Yuri Pederi]

Hi Kevin,

Thanks for the insight. We improved vibration isolation of the autopilot and eliminated throttle impact on Y accelerometer. Please see attached the log-file. We used the Vibration Analyzer and I wanted to check with you if we need further isolate autopilot. The magnitudes in G's taken at full throttle show the biggest peak-to-peak of 4.33 on X-Acc. Is there a benchmark we could use to work on further dumping the vibration?

Regards,

Yuri

[Kevin Surminski]

Hi Yuri,

In the vibration analyzer, you probably want the maximum values of acceleration to be much less than half of the limits on the sensors. On G2 autopilots' the accelerometers limit is +-5 G's. Therefore peak to peak should be less than 5 G's. Otherwise there is a chance they might saturate. I see a pk-pk for x accl is 4.33 which is acceptable. pk-pk=3.95 for y accl which is acceptable. And pk-pk=3.85 for z accl  which is also acceptable.

You'll also want to look at the average value to see if there is a bias shift. The closer to zero for X and Y, and -1 G for Z the better. Your averages look good at 0.02 for X, -0.01 for Y and -1 for Z (all in G's). 

I also looked at your datalog and it appears that y accl looks much better now. Throttle doesn't affect it very much anymore and the magnitude is much less. I think you have solved your vibration issue and further vibration damping is not required.

Sincerely,

Kevin Surminski

MicroPilot Support 

[Yuri Pederi]

Hi Kevin,

Thanks a lot for your help. In the Rudder from Y Accelerometer PID the values are negative. In the earlier log the Y accelerometer and Rudder work in the opposite directions. Is it correct?

Regards,

Yuri

[Kevin Surminski]

Hi Yuri,

Yes, the rudder from Y accl PID gain values are negative. This is correct. Compare with default.vrs. If you ever have any doubt about  the signs of gains, run the VRS checker. It will flag a failure if any  PID gain has the incorrect sign.

"In the earlier log the Y accelerometer and Rudder work in the opposite directions. Is it correct?"

I believe the rudder was acting in the correct direction. You can confirm this by testing it on the ground in CIC mode before takeoff. Yaw the UAV left and right and watch  the rudder. When the tail moves left (nose moving right, yawing right), the rudder trailing edge should move left. If it moves in the opposite direction, use the swap checkbox for the rudder in the servos tab of the VRS editor to reverse the rudder direction.

Sincerely,

Kevin Surminski

MicroPilot Support

[Yuri Pederi]

Thanks Kevin, the issue is fixed. Appreciate your help!

Regards,

Yuri

[Kevin Surminski]

Hi Yuri,

Glad I could help. 

Regards,

Kevin Surminski