Fluke 8840A Partial Repair

[Nogtail]

I got a broken Fluke 8840A a while back as a repair project. When powered on the display was showing random segments and keyboard was unresponsive:

I tested voltages and they seemed to be bang on and didn't have any noise etc. that I'd be concerned about.

I went through the manual's display troubleshooting section, but everything looked like it was working correctly. I moved to the CPU and couldn't see any life on the IO other than the oscillator and a 4MHz output. I tried to get the CPU to run the troubleshooting program by shorting TP205 to GND which did absolutely fuck all.

I assumed the CPU was dead, but a replacement would have to come from a parts unit as it has an internal 4k mask ROM. It looked like I might have to auction off my left kidney to get a replacement CPU, so I took inspiration from @cheapskate and used some commie CPU with an adaptor board. My adapter board uses a AT28C64B EEPROM to replace the mask ROM. The firmware is 8k in size, but only the lower half fits in the CPU mask ROM. The 8840A has an EPROM (U222) which contains the upper half of the firmware which has to match the version in the CPU. U222 in my unit has version 2.5, but I was only able to find V2.3 online for the CPU so I had to downgrade U222. I don't have a UV eraser and the weather has been pretty shit, so I wasn't able to get the sun to blast the existing EPROM clean. I decided to replace it with another AT28C64B which seemed like it should be compatible.

I threw the replacement CPU in and was greeted with what looked like a functional meter... for about 2 seconds. On power on it shows the display test pattern, then it goes all weird and thinks it's measuring voltage, current and resistance at the same time while the display tries its hardest to give you epilepsy.

I tried to start a self test but if any button is pressed it'll either go into what looks like the mV range and becomes unresponsive, or get stuck in a boot loop with a bunch of relays clicking. With the CAL switch pressed it seems to boot into what looks like a working calibration mode.

The Fluke 8842 seems to have very similar hardware, so I tried loading it's firmware to see what happened. It seemed to be working a lot better than the 8840's firmware and I was able to switch modes and run the self test (which failed pretty much everything unsurprisingly).

I threw the 8840's firmware back on and shorted TP205 to GND to get into the troubleshooting program. All pins seemed to be outputting the expected signals, although everything connected to U211 had a much lower output high voltage (3.4V instead of 4.5V). I replaced U211 along with some of the other nearby 74 series logic while I was at it. That didn't seem to have any effect.

Now I'm pretty much stuck. I haven't got a clue if it's a software problem, a hardware problem or if my weird East German CPU isn't fully compatible. I'm also not sure if there's been a hardware revision that makes V2.3 incompatible.

If anyone has any ideas what I could do next that would be great!

EDIT: Seems like the GIF of the current problem didn't work, here's a video instead: