Did you evaluate the AT clock on the SiS board while running DOOM and the same hardware?

Did you replace all the crystals which were near the battery leakage? My partially corroded cyrstals needed to be replaced.

It is also possible that the AT connector on the motherboard isn't making all that firm of contact with the PSU's connector. I had an issue like that once and had to replace the AT connector.

1: Pretty much, yes. I used the VESA card I was experiencing almost immediate lock-ups with on the AOpen along with the VLB I/O card that I had been using on the AOpen. Waveform for bus clock on the SiS was absolutely rock solid on the scope. The AOpen one is quite noisy. Probe sat in there while Doom ran the usual demo3 without issues.

Also note that I have noise in some of the surrounding pins as well.

2: I replaced the RTC crystal at the edge of the board along with most of the other components in that area. Corrosion didn't reach the 14MHz crystal - and it is functioning properly (as shown in previous post).

3: I walked several of the ISA slots near the bottom of the board and saw the same measurements. To change things up, I relocated the I/O and VGA cards to the bottom ISA slots and probed the top ones. Same story.

I'll run the bench/test power supply that I've been using for the AOpen testing on the Soyo system and make sure the readings don't mimic the AOpen. If they are solid, then the PSU isn't at fault.

My money (possibly literally) is on the trace paths to the terminating resistor banks and/or the filter caps on the motherboard. It is also worth noting that there are several other places for capacitors that weren't populated on this board (various power rails for example) - It is possible there were more caps on the original version or only in design and they were omitted for this version's release as a cost-cutting measure. As a matter of fact, they are present in the front cover photo of the manual.

I too am more inclined to think the issue resides with the motherboard and not the PSU, but I'm wrong 50% of the time. Did you check the PSU for bulging caps? I had to recap my original 386 PSU about 5 years ago.

I did. Caps look good. PSU probably should still have a recap, but it is a Dell HP-P2507F3P & is built well. I'll get it plugged into the Soyo tomorrow.

Was the pinout of the old Dell AT PSU the same as standard PSUs? I know Dell swapped the pins around on the ATX PSUs, which angers everyone.

Yep. It is a standard one. Picked it up in at a garage sale along with a bunch of other stuff about a year ago. Guy did PC repair work for years.

With the same test PSU hooked up to the Soyo 486, the signals were quite stable. I am ruling out the power supply.

Noteworthy measurements to ground for each pin on the AOpen - B20, B28, and B30 each hit 27Mohm (geez) with everything still in the case (VGA + I/O. PSU not hooked up). Removal of the mobo from the case and the expansion cards gives 1.8Mohm for B20 (and the rest, most likely). The values sail upward when measured.

I have the clock signal traced from the UMC 491 IC to a 33ohm resistor to the 8 pin XT connector and beyond. I have no idea where one of those B20 pins might be terminated as I cannot tell from the exposed traces on the top and bottom of the board.

Running the probe around the board for a continuity check didn't yield much - and there are so many points to hit. I suppose I could check each resistor bank individually (there are like 25 resistor networks...) - OR - I could just toss an 11K resistor between B20 and GND to see what happens to the CLK signal.

It might be prudent to confirm that none of the onboard capacitors are shorted, including tantalum, SMD, and electrolytic. I've had this happen sporatically with a tantalum cap. There's a thread on it somewhere.

it might provide more insight if you have a, say, 20 K trimmer, and add it between your pins and GND. Then watch the scope as you adjust the trimmer.

Are you able to test the crystal oscillator out of the motherboard by setting up an oscillator circuit on the breadboard?