Kit's Machine

[Doddy]

5V for the BoB and 12V for fans is provided by the original PC power supply.

Kit, you're one person on here that I really wouldn't want to teach how to suck an egg. But that picture of your BoB - firstly addresses on solution that crossed my mind - you're trying to smooth the supply to the BoB with that cap hanging off the PSU (I would suggest a low esr cap in parallel with the floating electrolytic - but there'll be enough of that on the board anyway)... so that's all good. But, the one thing that does worry me is I can't see any isolation on the BoB (in the form of opto-isolators). It could be a different package that I'm used to - but the geographic positioning of devices on the board doesn't appear to support that. You say you're using relays on the inputs - I can't see the detail of this in the figure, but assume you're using a NC or NO contact?, hopefully you have it that in the non-triggered state the input is hard-tied to ground? (i.e. not floating?). Have you the connectivity to use a C/O contact between 0V and 5V, so the inputs are hard-tied in both states? Also, I've experienced so much crap with proxy diagnosis of noise on BoB inputs that I wouldn't now entertain any without opto-isolated inputs. Not for the galvanic isolation, but simply it's good to have a device that needs proper current flowing to switch, rather than the usual HCT high-impendance inputs. From the UK I'd be throwing 6 quid at China for a cheap replacement BoB to trial.

Again - I'm confident you know what you're doing, and you sound like you're addressing this top-down with removing noise from source, just worth conisidering going bottom-up and looking at the noise susceptibility on the signalling to the PC.

UCCNC, btw, is the way forward :)