Board fabrication order is in progress so I am less stressed and have a little time to look at other things. That DE CPU board will take quite a few days of continuous work to get to a stage where the schematic is transcribed.
So ... a little progress tonight. I'm not going to declare the findings as definitive because I only have a sampling size of one.
On a whim (or a hunch):
- Swapped the 74LS244 in U1 and U2 on my CPU board (good thing I build these with everything socketed) with a 74HCT244.
- Logic levels should be compatible and everything appears to work on the display but the sounds went absolutely crazy with just this change.
- Changed the Fujitsu 74LS138 with a Texas Instruments 74LS138 in U26 on the sound board. This reduced the level of craziness but there were still errant problems.
- Tried the 74HCT244 with an OEM board. Works perfectly.
- Put back the reproduction board. Reduced level of sound craziness.
- Swapped back the to the Fujitsu and high level of sound craziness.
- Swapped back to the 74LS244 on the CPU board. Everything back to normal.
Conclusion ... IC incompatibility / variance due to manufacturer and silicon type. HCT is logic level compatible but may not be fully timing compatible. I know I read somewhere about issues on Rottendog CPU boards that use HCT ICs and replacing the HCT family with the LS family fixes the problem. Rottendog uses HCT presumably because they are cheaper and more readily available.
My first suggestion to anyone encountering this problem is to make sure you have an OEM (original Williams CPU) board.
@ Crash - Did the machine(s) that exhibited failure have OEM CPU boards?
@ Coyote - I know the TZ is not OEM (green LEDs as ChrisH has noticed). I hope the results of any testing at other machines provide evidence to back up the conclusion I made above (on a sample size of one).