I just posted a video of a very odd problem with a System 11 (from a Banzai Run machine, but currently running F14 ROMs) on YouTube, thought people here might be interested. Maybe someone even has some ideas on what could be causing it!

For those who prefer text or generally don't like YouTube, here's the description:

Here's a Williams system 11B (running F14 ROMs) doing weird stuff. It does this every time it's started from complete cold, reliably. It runs for a while (about 20 seconds), then locks up. After around a minute, it comes back, and resumes where it left off. It does the same thing in the machine too. If the player starts a game before it locks up, it will resume the game fine when it starts running again, it just seems to pause. the scope and counter are both connected to the main clock (pin 39 of the CPU). That seems absolutely stable.

As the problem begins, the debug LED gradually starts flashing faster. I think the attract pattern also speeds up. Eventually the LED is flashing so fast it just seems to be flickering. At about 30 seconds in the video, I cut out 50 seconds of it doing nothing. When it comes back, the LED gradually slows down, and once it reaches a reasonable speed, the attract pattern resumes. This particular run is a bit unusual - it restarts, then locks up again for a couple of seconds, then restarts again. If I power it down and then up again, the problem doesn't occur, and it seems to run perfectly reliably. A day later, it'll show the problem again.

Things I've tried: Checking over the reset circuit, in case it's browning out and doing stupid things. Replaced the CPU socket, and a few other things that looked questionable. Scoping out various other lines, which look basically OK.

This so annoying to troubleshoot. I get about one fault per day, with the outage lasting about a minute. Occasionally it recovers unexpectedly, and at times it doesn't fault at all. Last thing to check - E seems stable, but I haven't managed to get a look at it while the fault is occurring.

Ah, good idea. I'll check that. It would explain a lot - for instance, why it's able to resume absolutely fine when the IRQ comes back.

You would have thought so, right? Neither seemed to make any noticeable difference. In particular, I thought that cooling the board to make the fault easier to reproduce would have worked. No luck. But now that I can concentrate on the IRQ support ICs, maybe cooling will prove more successful.

Yep, CR1 seems fine - both frequency and waveform are solid. Re the processor, yes - it's actually running the sound section CPU as the main CPU at the moment.

Thanks for the follow-up - I've been seriously busy over the past few weeks, so progress has been slow. There are three ICs directly below the batteries that have been affected by corrosion; the legs are discoloured. Two of those (or was it all three?) are part of the IRQ logic, so I'm guessing that's the cause. I'll cut them off and replace them, hopefully that'll sort it out.