At the ASVARO Electronics Flea Market I bought an HP-5216A frequency counter.
The first problem with it is the odd power connector. I went to Halted Surplus and they had the connector, so I picked one up and went around the store trying to find something that fit and found these. I drilled out the holes a little more, spread the Molex pins a bit, and soldered them to the AC line after pushing the wires through the shell. The wire near the fuse goes to the fuse so I assume that's the hot wire. Center is obviously case ground.
The second problem was my fault: I removed the two screws holding the transformer cover in place and when I pulled the cover off, two nuts went "clunk" inside. Don't do that. He's how to recover, in 15 easy steps.
The least and most significant digits don't light. The MSD was a bad nixie. The LSD may be the oddball 1820-0092 decoder/driver chip. More news as it develops. The driver chips are scarce and expensive; it might be possible to fabricate a replacement from a 74141/7441 and some open-collector driver such as an ULN2803.
After an hour of operation, it's tripping the GFCI. It worked fine until I ran my espresso maker on the same circuit, so perhaps a spike got it. On a non-GFCI circuit (not touching case!) it works fine. I removed the double disc capacitor across from the AC line inputs to ground, but that wasn't it. Now I need two 0.01uF 250V Y2 capacitors. DVM shows open from "neutral" to ground, 125 Ohms from hot to ground, which changes to 175 ohms if I switch the transformer to 220v line input. GFCI doesn't trip if the fuse is removed. Leakage current is about 12mA to ground when on, 1.2mA when off. It's beginning to look like it's the transformer.