Altitude sensor, drift due to P4 heating up?

[herkam]

Hi!
I did some hovering today to try to figure out what the pressure sensor (altitude sensor) is doing. My conclusion is that it drifts due to the heat buildup in the P4.
From a cold P4 the heat build up will lead to an altitude change of around 8-9 meters, it always changes to greater altitude. If I just replace the battery and start again the altitude drift stays at a moderate 2.5 meters. It stabilizes after 4 minutes from a "warm" start, and maybe 6-7 minutes from "cold" start.

I had the vision sensor turned off in litchi and use the log feature (really great by the way) to plot how things where changing.
Temperature was around 15°C.
If I didn't touch the sticks the P4 would slowly descend so I adjusted the height with the stick to keep it constant relative ground.
The P4 was hovering and the wind wasn't to bad so I rule out any dynamic effects.
The altitude was reported in feet so the changes in the plot is in steps of 0.3 m, there is a another altitude called "raw" in the log file that is reported in dm but excel plot was made using feet. I checked the two values and they are identical except that the "raw" value has a bit higher resolution (dm instead of feet).

Y-axis is meters and X-axis is seconds.

Cheers
Hakan

 

[GMack]

Interesting data.

I often wondered about the "Calibrate when cold" idea since so much other electronics stuff is calibrated at working temperatures, or at least with some warm-up time. I would theorize that calibrating at operating flight temps would be the best method, but to do that one would have to know how much cooling is happening inside the drone during flight.

 

[herkam]

It would be interesting to do a "cold IMU" calibration and see if they create a compensation map for the altitude sensor during the heat up of the P4. But that requires that they have a temp sensor that measures a temp quite close to the pressure sensor, would be interesting to know!
The IMU seems to be heated to 60°C so they don't have to use a map with compensation values at different temperatures, they just need compensation values for 60°C, nice solution!

Cheers
Hakan