Second tethered test flight of the Raspberry Pi controlled quadcopter - yet another terminal crash.
On the plus side, this time, the blades didn't slice the LiPo battery cable in two, so no risk of explosions, and I have some diagnostics to ponder. Having the video gives good hint: I think the PIDs for pitch, roll and yaw have gains set too high (in this case the roll) nleading to a chaotic, and increasing left / right oscillation. Since my code won't cope at all well with 90 degree swings in any direction, this is my primary suspicion of the root cause of the crash. Sadly, I can't test the theory until I have yet another set of blades!