Glad to see I'm not alone - I have what appear to be the exact same problems here, with a BPPC. Rarely MOS works for upto an hour, and more often than not it just locks up after a few minutes. Somewhat frustratingly, now MOS won't even boot here - it gets as far as the CGX boot butterfly screen and that's it. It was booting ok on Wednesday, but for some reason it stopped after that.

I have to agree that I doubt this issue is heat related - OS4 is working perfectly on the same hardware still, as is OS3.9, obviously. Smells of interrupt handling problems to me. Maybe even the 603e specific cache TLB miss exception handlers which are software based, unlike the 604e and newer PPCs. I remember when I was fixing 603e support in the OS4 kernel, it was the TLB miss handlers that were the main problem (especially given the Motorola example code for them contains bugs!), which tended to cause lockups on OS4.