For the end of October, the weather is more like full summer here in the UK – ambient temperature is more than 20°C in the shade, and in the sunshine, probably nearer 30. Phoebe with her perspex dome, and black heating element (resistor) took just a few seconds to warm up (it normally takes a good 30s), and the flight met full expectations*.
Then I ran a second flight – she leapt into the air almost immediately as her temperature now was only just under her 30°C calibration temperature, and because her temperature continued to climb, so did the errors in the calibrated sensor readings and she drifted.
I’ll take a triple pronged attack strategy: first, I’ve colo(u)red in the heating resistor with a silver pen I have after checking it’s not electrically conductive (although this had little effect); next, I’ll recalibrate her at a higher temperature – perhaps 40°C; finally, I’ll add some code to kill a flight if the temperature exceeds some arbitrary bounds around the calibration temperature.
* expectation still means drift currently; my next step on that now has to be to deploy Chlöe. Her sensors are soldered to a PCB which is bolted to the RaspberryPi which is bolted to the B+ PiBow case. I’m hoping that this rock solid build will help getting rock solid calibration at fixed temperatures.