Here, I share a repair report, hoping the it might help in case of similar issures.
During my initial research, I found several reports of Fluke 165x having unreliable RCD measurement, this might be the same issue.
I recently bought a defective Fluke 1653 for a good price, mainly to check my own house installation after renovation.
The description mentioned "All tests working but calibration not possible". I was also provided an email from Fluke, stating that RCD measurement needed repair in order to proceed with the calibration.
When I got the instrument, I tested it for a while, trying out all settings and ranges. Indeed everything was working fine, except for the RCD tests.
I realized, that using the low current range (e.g. 30mA), I was not able to trip my 30mA RCD, but with higher current ranges (e.g. 300mA) the RCD tripped. However the measured tripping current was way off (90mA)
Next step was to measure the AC current while RCD tripping using the different current ranges, to find out what really happened.
Catching the right moment with my HP34401 turned out not be be easy, as the device sets a fast current ramp (especially if your multimeter is powered by the RCD you are tripping...
)
Still, I was able to observe some kind of linearly increasing AC current ... but serveral orders of magniture to low.
At least, this explains, why the RCD was not tripping.
For the higher ranges, the AC currents started very low as well, then jumped suddenly. At this point the RCD tripped.
Smells like range switching, with one range broken ....
Serching for "fluke 165x service information", you will find an extensive service manual, which helped me a lot, as it provided schematics and explanations.
As I don't think, it is legal, I don't provide a link here.
This instrument provides two differnt RCD measurements: it can measure the trip current and the trip time.
For that purpose, it has an active load consisting of 4 MOSFET, operated in linear region.
The load is controlled by a µC and the current is measured by a shunt.
Indeed, there are two measurement ranges using a 1\$\Omega\$ shunt (R36) and a 10\$\Omega\$ resistor (R62).
The 1 Ohm shunt can be switched off by MOSFET Q19 for low current ranges.
The rest was easy: I took the meter appart and measured R62 - and yes, it was serveral kOhm.
As I did neither observe cracks in the resistor nor burn marks, I can't say why it failed.
Could have been overload or a small fracture due to thermomechanical stress or shock.
Replaced, reassambled and everything worked perfectly.
One additional hint: the ribbon cable connectors are rather fragile. Care has to be taken to connect them correctly.
I had the issue, the after reassembly, the meter showed Err 1 without leads. The L connector was dead for current measurements.
It turned out, the the ribbon was a little bend inside the connector resulting in the L-Line relay not to switch.
BTW: I would be very much interested in the detailed description of the communication protocol, including the calibration commands. In the service information it is referred to as "165X Serial Interface Specification document"