After finishing the first round of repairs and upgrades to the PCBs, I couldn't just put them back into a dirty box. Inside the head was all grungy, corroded, rusty, etc.

Now everything is cleaner than any time since it left the factory. I went a bit crazy - removed almost everything, brushed, vacuumed, wiped down with alcohol, and where applicable - sanded, magic erased, secured peeling labels, etc. All the small parts went into the tumbler, I didn't have high hopes. But after only a couple of hours those horrible rusted screws look quite good. You'll see later when I put it all back together.

This is the good stuff, available at Loblaws. Before the best I could find was only 70%. It quickly removed the permanent marker.

FLASH LIVES!!

After weeks of repairing and rebuilding practically everything between the wall outlet and the slave display, only a pinhead can appreciate the sense of relief and accomplishment felt at this sight. In the end there was just one problem - inadequate cleanup on the very first socket I desoldered, shorted a couple of pins. Everything else seems to be working perfectly on the first try.

Thanks again to Steph for sending Leon's test ROM!

Quoted from tinyrodent

:

Thanks again to Steph for sending Leon's test ROM!

Thanks for the kind words...
but really,  my contribution was rather insignificant compared to all the work you've done...

Congrats on the restoration.

Cheers,
Steph

Well I briefly had the pleasure of seeing all 5 displays working, nice to know that none need replacing. However after plugging in some driver board connectors and trying again, the MPU lost its blanking signal. Removing the driver board connectors did not restore the blanking signal. Everything else still works, and I can "cheat" the displays on again by jumpering the 1J3 display blanking connector pin to +5V.

I tested the components of the blanking circuit with a logic probe and found both sides C32 look good, but C31 input is low. I think that means either Q5 (2N4403 transistor) or IC23 (556 timing chip) is at fault. Not sure what the "normal" results with a logic probe are for these components, probably will try comparing with Firepower tomorrow.

Q5 shows a slightly low reading on one side with a DMM diode test - 648 vs. a value closer to 700 on all the other in-circuit 2N4403s I compared with. Doesn't seem conclusive...

First I realized that TTL setting should be used on the logic probe, not CMOS. This changed the results, showing Q5 and all the related caps with readings that match the (working) Firepower MPU. The inputs of IC23 look correct but the output pin 9 is low, indicating a probable internal failure.

Testing with the DMM on diode setting, red lead on ground pin 7 and black lead on output pin 9, shows a value of "1" on the faulty chip, indicating a short. On the good chip, the value is about 700. Testing for resistance again shows very low value on the faulty chip, and about 5K for the good chip. Hopefully this is the correct diagnosis, next steps will be to install a socket, then try a new chip. It will probably be mid-January before the replacement arrives...

If you're testing in circuit then you may be getting a false reading.  If C62(on a system 3 or system 6 board) is failed shorted then it'll give you the same reading.  On those boards you could remove C62 to see what happens.

According to the system 4 schematics that are in the Flash manual, it's also C62...  Bascially it's the cap between ground and the blanking line.


Mike

Thanks Mike I am testing in-circuit and the results are a bit confusing. Part of the problem is a low quality scan of the system 4 CPU assembly and logic diagrams. The Flash manual I've got only contains system 6 info which is quite different. So I've been trying to use system 3 documentation, which is similar but not identical to system 4.

If anyone can point me to a readable system 4 assembly and/or logic diagram I would be most grateful. The blurry one is here http://www.pinballfixers.com/pdf/Williams_System_4_CPU_Schematic.pdf

Mike I think you might be right, it's C62 or the rightmost green component in the photo. There is no resistance across its leads. Measuring the equivalent component on the sys 6 shows about 5K ohms (both are in-circuit measurements).

You can just remove 1 leg and give it a try.  It'll work without the cap (but you should replace it when you get a chance).

Mike

I guess it must be the timing chip after all. C62 tests ok when disconnected, but pin 9 of IC23 is still shorted to ground. The blanking circuit is pretty simple, I don't see any other explanation for the symptoms. As a sanity check I again jumpered 1J3 pin 4 to +5V and the displays work fine. R70 only measures 350 ohms in-circuit but I assume that's because of the faulty chip.

Not sure what else I can test with the board still in the machine, so it's back to the bench.