I picked up another 34401A from ebay. This unit was rejected by the cal lab with 'failed 1000V adjustment’ sticker on it. It passed the self-test and all modes and ranges worked fine.
So I hooked it up to the calibrator and fed 500V. The overcurrent protection tripped immediately in my calibrator and multimeter went out completely.
I cycled the power and now it would not turn on at all. After digging quite a bit I found 5V VCC rail almost shorted to ground. Ok FLIR time and sure enough the U150 was about 170°C.
There was no more obviously bad or fried components or anything like that so I replaced the U150. The DMM worked again. I hooked it up to the calibrator once again and started
increasing the voltage bit by bit. At around 450V it happened again with exactly the same result - the U150 was dead. Then I started a complete and thorough check of the frontend
and I found the culprit. This was the first time in my experience.
The K101 had a very small burnt spot on the case where it touched the E100 surge arrestor (with one side of it connected to the HI terminal).
I checked the K101 relay and metal shield / housing had resistance to the coil about 14Ω. So it was a double breakdown between the E100 and relay housing and also between relay housing and coil!
Perhaps the prime cause was too little clearance between the E100 and the K101 and some sort of vibration rubbed the paint of the relay housing and initial high voltage breakdown occurred.
Now I will check the clearance in each 34401A and encourage you to do so as well!