Everything went horribly wrong. I added the decoupling 680uF electrolytic capacitor as per the Garmin spec, and the flights consistently triggered a hard system blackout shortly after each ‘lab’ flight started. Only a full power-reboot fixed it each time. Disabling the camera and normality resumed, including the I²C errors. These only exposed themselves again during a outdoor flight, causing her to to suddenly leap 4m into the air when she was supported to be descending. I killed the flight, luckily crashing into some bushes. But that then exposed some design flaws in the frame where the CF arms attached to the folding aluminium shoulders. Net of the last few days: 2 broken props, one motor with wires cut, and an arms and shoulders damaged. That’s a very expensive few test flights.
Once the replacement frame parts arrive from Poland, I’ll fix their design flaw based upon knowledge from a previous revision of me about how to build carbon fibre archery arrows. More on that when the parts arrive.
In a couple of ‘lab’ runs, everything starting behaving perfectly again when powered directly from a 5V 2.4A power supply. That suggests Hermione’s A+ + sensors + camera are overloading the 5V 1.5A switching regulator from the LiPo causing a hard-power-off, FIFO overflows and I²C errors. I’ll try Zoe and her Zero when her PCBs arrive later this week. As well as reducing power requirements considerably, using the Zero conveniently moves the camera cables away from the I²C PCB tracks, and hugely shortens the I²C wires to the Garmin so this may well solve everything.
On the plus side Hermione was flying in X8 format very stably prior to the crash!