I recently was digging through some old things and came across my old GUS (revision 3.74) that I've had since the 90s and had a wave of overwhelming nostalgia that caused me to jump into sourcing all the bits for a great mid-90s retro box to put it in. I finally put everything together, and then was greeted by hard freezes when running GUS setup and running ultrinit.exe. Replacing the card was out of the question due to the high cost of GUS cards these days. I found a few leads for people with the same problem: the Gravis Ultrasound Classic problems thread here and the Gravis UltraSound card, can't get it working!! thread on VCF.

Here's what I went through to bring this card back from the dead, in order of cost and effort involved:
- Didn't help: Cleaning all contacts: ISA connector and all socketed chips.
- Didn't help: Replacing one suspicious cap that was dented, but still tested OK.
- Didn't help: Replacing the 19.7568MHz crystal from a place in the US with a $10 minimum order quantity, so I had to order 22 of them. This thread showed me that it was possible for these crystals to go bad, and now I see from Need help repairing a Gravis Ultrasound Classic 2.4! the poster there ran into the same minimum order quantity issue. Whoops.
- Didn't help: Replacing the Lattice PLD chip with one I reprogrammed from a dump I made the original chip as well as the dump in the Gravis Ultrasound Class Rev 3.74 Missing IC thread. I had a suspicion about this one because the PLD's datasheet mentions a 20 year retention for the programmable cells, and we're well past 20 years since these were made.
- Worked!: Purchasing surplus replacement GF1 chips from China. I actually found 3 different sources for these, two of whom said they could source pretty large quantities. I was extremely skeptical, but one was able to source and ship quickly via DHL, and I had GF1s in hand from them in less than a week. Quite impressive. I'm still waiting for supposed GF1s from the other two sources to arrive, as they shipped via much slower shipping methods and now we're in the middle of many companies taking off for the Chinese New Year holiday.

The one on the right is a close match for the one on my GUS v3.74. The one on the left looks quite different, and at first I thought it was a fake or was sanded/rescreened but from its date code (late 1994), it's actually a close match for the later GF1 chips on cards like the GUS Extreme and Primax Soundstorm Wave. I tried the chip on the right first and the GUS worked right away.

So, now, I have one extra GF1 chip and potentially 3 more coming my way, 2 PLD chips, and 21 19.7568MHz crystals. Now to figure out what to do with them. I'm tempted to start an ARGUS-like project with the orginal GF1, but the companion GF1D1 chip is very hard to source.