In photo #3 it looks like you have the 5v input from the ESC (red wire) connected to the 3.3v output of the onboard voltage regulator. It works OK that way? I would think connecting the red wire to the 5v input pin would be better.
When I read this and realized what I had done, my initial panic gave way to curiosity since the board is still somehow working as expected. Yesterday I scoped the 3.3V line and the signal line. It's 5V on the 3.3V bus and the PWM signal is 5V P-P. Basically I'm backfeeding 5V into the poor voltage regulator's output. More importantly, according to Adafruit's schematic of the QT Py, that 3.3V bus directly supplies VDD to the SAMD21. How have I not blown up the board's brains all this time?
You're right, supplying ESC power onto the 5V bus is probably safer
In that case do we forgo the recommended protection diode due to the enforcement of polarity by the ESC connector itself?
I see you added the 2Mb SPI flash memory chip on the back. Are you using the Haxpress version of Circuitpython to take full advantage of the extra memory?
I intend to, but the usual CIRCUITPY drive doesn't show up after I load the Haxpress firmware. Given the 5V goof above, I now suspect the Flash chip did not fare as well as the SAMD...so back to the soldering table I go
But at least the memory chips are cheap and I bought a handful of spares.
The Flash chips I got are actually the bigger 16MB ones because I expect to store a little over 2MB per session. I figured, why not get the largest possible one? According to Adafruit >2MB chips are supported, they just decided to make the pads in the smaller footprint. It remains to be seen whether it works. The pin pitch is exactly the same, the body is just a bit wider. I carefully manipulated all eight pins so they land perfectly on the pads.
Their team is busy with daily firmware updates...when I first programmed this, it was in beta. Then I noticed a Release Candidate 0 on Friday. Looks like they're close to releasing mature firmware soon.
I have been recently testing a version of the timer that adds an analog accelerometer to boost the power in the vertical climb. Based on the initial test flights this looks like a big improvement. On the QT Py it would be super easy, no soldering, just plug in an I2C accelerometer breakout board.
I agree! I have some ideas on that front as well, just haven't had the chance to return to executing them.
Cheers,
-Andrey