Good ideas!

I think the pump board USB connectors have been updated after your past orders. Labmaker found a different variant which seems to hold on stronger.

The temperature sensor idea is good, though it means we would have one per reactor rather than one per control computer. Originally I put the sensor on the end of a long sticking-out part of PCB in an attempt to thermally insulate it, but as you said it doesn’t work that well. Often it is as much as ~7-10C higher than it should be…

The feedback on stirring/heating could potentially work. I think it would be feasible to tell if the motor is rotating or not, but also figuring out if there is a stir bar there would be tough… Are you hoping that the motor will draw a significantly different current if it is/isn’t interacting with a stir bar? I suppose there will be SOME change in current draw, but it might be of a small magnitude and hence difficult to distinguish between inter-device variability in that component (fan current draw). If we want to implement this could add a low-value resistor downstream of each of them and then measure voltage across it (and hence current) with an ADC (which would need to be added, there are cheap/simple I2C ones that would fit).

For the water sensor change is the primarily purpose you imagine to ensure that we can always diagnose when this has caused the crash? (versus some other possible hardware issue)

Yeah, i agree cabling isa pain and something I have thought about changing. At present there isn’t too much to be done since (even if data was wireless) we still need power transmitted between them.
One option I thought of was to have a “base” to which reactors/pumps attached. Thus, all power/communication would be routed through some table top-size structure and you just plug in the pumps/reactors to the top of this. But, would take a lot of design/manufacturing and potentially could create new failure modes (e.g. liquid spills could become more damaging…)