Parts of this reminds me about an issue I had around the same time, I worked on porting A/UX the Unix port for the Mac II, we reportedly got half of the early p

Parts of this reminds me about an issue I had around the same time, I worked on porting A/UX the Unix port for the Mac II, we reportedly got half of the early production run, they came without plastics in cardboard boxes, with schematics and PAL equations (PALs were relatively new then) I found and they fixed a bug in the PAL equations (didn't handle 24-bit writes some new 68020 instructions could make) so I had some cred with them when we ran into a weird bug, very occasionally you worked with the IWM chip (used floppies) the keyboard would freeze up - but only on some machines. Managed to figure out that the machines that worked had an ADB (keyboard) chip with markings on it, the ones that didn't had no markings - Apple swore they were the same, eventually told us the ones that didn't work were the ones the were doing the manufacturing run with ..... bug from hell ..... turns out there was some circuitry in there they were quite proud of, when you accessed the parallel IO (VIA) chip it tweaked the clock phase a little to give you faster access, part of working with the IWM chip involved setting up the timer in the VIA chip to time when the next sector would go by, we'd poll the timer fast enough to tweak the clock on every clock cycle resulting in the ADB chip which was connected to the VIA being clocked too fast .... we replaced that code with a simple delay loop