7 replies to this topic

[Corvus Corax]

Hi

I did some Gyroscope temperature measurements. I got very weird results. Is this the right place to post it?



As you can see this is from a time compressed replay, the realtime replay would stretch over about half an hour ish or longer.

The jumps in accel raw come from shifting the frame to a new position. I had disconnected telemetry when moving the frame to the new spot to prevent peaks in the ahrs readings that would ruin the plots.

I get 50 degree with my frame's lid closed. Both GPS and OP get quite hot. I believe the AHRS has a lower temperature since it sits apart from the other components (especially the ESC/BEC which I believe to be the main heat source)

[Corvus Corax]

Here's 2 more plots while I let the temperature rise again.



I also did a vibration test during this one - visible in the accelerometer window. I expected the ESC and the engine to cause even more heat, but instead the airstream around the fuselage cooled the system down noticable

There is a third plot of the same process up to above 40 degree. There was a glitch in the accelerometer, messing up the diagram though. I have those from time to time in both the gyro's and the accels. (I blame the broken AHRS)

[Corvus Corax]

The raw logs can both be found in http://cybertrench.com/vixen, too.

[peabody124]

Sambas did similar tests and found that the gyro drift compensation covered it when he threw it in the snow in Sweeden (after changing the gyro drift rate constant). So a better thing to look at would be the gyro filtered.

[Corvus Corax]

peabody124, on 03 January 2011 - 01:02 AM, said:

Sambas did similar tests and found that the gyro drift compensation covered it when he threw it in the snow in Sweeden (after changing the gyro drift rate constant). So a better thing to look at would be the gyro filtered.

Unfortunately my AHRS doesn't have a working magnetic sensor, so it can't measure the drift around all axis. I am thinking about using the gyro_temp values for an additional linear gyro temp compensation put into the filtering code. Maybe that gets me more stable heading.

[Corvus Corax]

As I found out my AHRS temperature will be flight speed dependent. If the frame sits indoors still, it gets as hot as 50 degrees.

But if I only turn the engine on it already drops - yet the lid is closed, protecting the electronics from most of the airflow, so it will not get to outside temperature either I think. I guess auto drift compensation is fine, but I'll likely need something on top of that.

[Corvus Corax]

Here is the same data replay with gyros_filtered displayed:

rising temperature:


falling temperature:


the drift compensation does "something" - but its not capable of fixing the temp offset - not without help from magnetometers (and I don't have those)

[Corvus Corax]

This is the gyros_filtered drift without temperature compensation:



This is the same temperature transition WITH temperature compensation (note to self, coefficient for Gyro Y is still off, X and Z are spot on)



the second plot is with BiasCorrectedRaw = FALSE

This is a bit tricky to calibrate.

the formula calculating gyros_filtered was

filtered = raw * scale + bias

now it is

filtered = ( raw + gyro_temperature * gyro_tempcompfactor ) * scale + bias

it can be calibrated by taking the average raw values for gyro and gyrotemp (XYZ) at two distinct temperatures

(the further apart the better. I would suggest +40 degrees and -15)

then just calculate the gradient.

 gyro_tempcompfactor == -(raw_gyro1 - raw_gyro2)/(gyro_temp1 - gyro_temp2)

Default in the UAVObject is zero, which means no temperature compensation takes place.