My 1979 Williams Flash (System 4) that was working recently has started displaying some odd booting behavior and will not start up.

It boots in one of 3 states:
1. Pulsing most coils and about half of controlled lights at about 2.5 to 3 times per second. It's not a full pulse that is strong enough to lift drop targets. GI lamps are on, displays do not show anything or might flash a number or two.
2. Single pulse of coils, then just GIs on. No displays.
3. Just GIs on. No displays.

The MPU board diagnostic LEDs seem to be random. They can be both off, both on, both blinking, or top on and bottom blinking in any of these states.
I'd say #3 is about twice as common as the other two states.

It worked for a while before showing this issue, but had some intermittent periods of starting in audit mode, most likely due to a dying CMOS. This problem showed up temporarily and then resolved to booting in audit mode after reseating most of the cables. Replacing the CMOS chip fixed the audit mode issue and I was able to play several games without issue. Then the display shut off mid-game and I haven't been able to get it to boot properly again. I have tried reseating the cables multiple times and haven't been able to get it working again, so I'm not certain it's actually a cable issue.

All boards are original, and have had all Scanbe sockets replaced with SIPs, remaining sockets are tight enough that I need to use a puller to pull chips, 40 pin connectors/headers are new, all pin headers are new, new CMOS, new capacitors, new remote battery, no visible battery electrolyte damage, about 25% of cable connectors have been replaced/repinned so far.

I'm not really sure what would be causing this or where to start. I'm hoping that someone might have dealt with similar behavior or might at least be able to point me in the right direction. I'm not even sure whether I should focus on the MPU board or the driver board.

An update since I started looking at this: I went through and re-seated all the chips and reseating one of the small chips in the upper-left section of the driver board caused it to stop firing the coils and only blink the lights. I suspect this might be an additional issue I just created, since all other behavior is the same. EDIT - just a blown fuse, coils fire again.

I haven't yet, I'll give that a try to see if it shows anything useful. Maybe if I see differences in the behavior of the lights it would help to show whether it's more likely a MPU or driver board issue.

I pulled the driver board off and I still see the same diagnostic LED behavior. It varies each boot, randomly going between both on, both off, both blinking (about same 2.5-3 hz as coil/lights issue), or top on and bottom blinking.
This seems to indicate to me that I'm looking at an MPU board issue. Do others agree? Any ideas?

Edit: pressing the diagnostic button generally didn't do anything. Sometimes it would flash the flights once or turn them off, but it didn't seem to be doing anything consistent or useful.

Some additional things I have learned that don't raise any flags to me, but might be useful to someone more knowledgeable:
When I remove one or both RAM chips, it is still able to exhibit the same issues. The problem can also still occur with the 4-to-16 decoder removed.
The problem can still occur with the driver board lamp PIA removed. I'm assuming I'd see the same without the other driver board PIAs, but I don't think it's worth adding more pull/place cycles to them.
When I remove any of the ROMs, CMOS, or the driver board PIA, the diagnostic LEDs just come on solid and nothing happens (the problem cannot occur)
When I remove the 2.5A solenoid fuse to have it just fire the controlled lights, I can hear that the flipper relays are firing along with the blinking lights.
I probed the blanking pin (4th from the left) on the 40 pin connector and I see about 110mv (DC avg) on it with my DMM. I haven't dragged out my oscilloscope yet.
While probing the blanking pin, I inadvertently put enough pressure on the connector to push the spring connector away from the pin and the flipper relays stopped firing while the lights continue to blink.

Improvements since earlier posts: all connectors/cable pins on the power and MPU boards have been replaced, along with about half of the connectors/cable pins on the driver board. As I noted before, all pin headers and the 40-pin have been replaced, along with all scanbe sockets.

If I can track down an EPROM that is compatible with the programmer at work, I might be able to get access to that lab to write a Leon chip, but that is going to take some time between getting the chip, getting access to the programmer, and actually writing it.

I welcome any suggestions or ideas.

I finally got it working tonight.

Here's what I think happened:
The one remaining Scanbe socket I found on the board while debugging this was for the 6875 clock chip. I somehow had not noticed the ramped edge of that one chip when I was re-socketing the Scanbe sockets on the rest of the board. This was probably the source of the pulsing on / off and resetting issues.
After discovering that socket, I pulled the board and replaced a few components including the crystal, the clock chip socket, and repinned some more cables. When I plugged it back in, it was completely dead. I messed around with swapping out some chips and continued re-pinning cables for a few days.
Tonight I starting ohming out the chips pin-by-pin and found that the 6875's pin 13 did not have continuity to the processor or the resistor below the processor.
I pulled the board and under a magnifier I could see that the board trace had pulled away from the pin hole. I was able to solder the trace back to the pin and it suddenly started working again.

Thanks for following along as I debugged this. I'm glad I was able to get it working without having to go through the process of burning test ROMs.

Now I can get back to debugging the sound board...