HP 3457A Ohm measurement leaking repair

[xi]

Today I decided to try to find out why my HP 3457A was displaying ohms values after some time with nothing connected to the input terminals!

I am posting here in case someone gets the same problem

Symptoms:
- after a few minutes in Ohms mode, with nothing connected to input terminals, the DMM displays a decreasing ohm value (starting at >100M Ohms and ending near 90k Ohms).
- 2 wires and 4 wires modes do the same
- there are no rear terminals on my DMM
- switching ranges manually do the same, and displays about 90k in 300k range / 0.267M in upper range / 00.278M in upper range / ...
- I placed a capacitor across the input terminals in order to periodically and quickly measure the voltage on inputs with another multimeter without influencing the HP 3457A: the voltage was slowly decreasing to reach a final voltage depending on the selected range (eg 0.9V for 300k range / 0.3V for 3M range)

All this let me think that the current generator was more or less working, but something was slowly shorting the input...

Repair:
The schematic below shows the input terminals (HI / LO), the input path and the Ohms current source.


I started by checking the ohms current source with the help of the service manual: everything seemed to be OK, except that the output of the current generator (pin 15 of U101 "Input Hybrid") started at about 8V each time I selected an ohm range, then slowly decreased, typical sign of something that is consuming current on the output of U101 (trace of the oscilloscope is connected on pin 15 of U101). Notice the value on the DMM that starts at "OVLD" then continuously decrease to reach about 90k Ohms:


Then I checked all the relays that could provoke a current leak by doing a wrong connection: everything was ok.
Then I suspected C101 to be "resistive", or worse: the input 13 of U101 or Q111 to be damaged: various tests didn't show anything wrong here.
Last component suspected was Q212 (Q212 is connected to ground and could create a current leak through Q215 base): and guess what: Q212 was not soldered with the right orientation! (notice the tab position compared to the drawing):


Q212 is a JFET transistor (2N4117A) used as a high performance diode (reverse current leakage is really really low compared to a diode), but soldered the wrong way, it's Gate was unconnected, and the Source and Drain were connected to base of Q215 and to ground respectively, thus creating a real current leak of Iout to ground

I checked the solders and this transistor have never been unsoldered, meaning that this issue is here since the manufacturing of this HP 3457A ... were the components inserted by hand on this DMM?

Anyway, everything is working fine now that I have resoldered Q212 with the right orientation I don't expect someone else to get the same problem, but who knows?

I have also reassembled all the schematics of the service manual of the HP 3457A, it's much easier to read. Here is the link of this document: multimetre_HP_3457A_service_manual_reassembled.pdf

[coromonadalix]

some thru holes components where indeed put by hand,  human errors .... but your meter wasnt fully tested, or tested long enough for this trouble to show   lucky you, it was an simple repair once the trail was followed.