We don't have facilities to cool chips, etc (not even any freezer spray!) but as none of the components run hot, that isn't an issue.

As I keep saying, I believe it's the oscillator on the button PIC and the issue is simply down to comms timing.

You cannot verify it using any of the front panel controls, as all of these will interfere with correct reception of dimmer commands by the front board. Yes, this is a design flaw, but the dimmer was added very late in the design cycle using spare pins on the PIC, hence the nasty interactions between non-IR input and front board commands.

Hugo