Leaky chopper - Keithley 177 repair

[Swainster]

Hi,
I've decided to seek advice from the community to help reduce my backlog of stalled repair projects, starting with most recent first. As you may have guessed from the title, this concerns a Keithley 177 DMM.

First, a quick introduction to the patient. The K177 "microvolt" DMM is based on the 4.5 digit version of the popular and venerable intersil voltmeter ADC chipset, however Keithley added "microvolt" capabilities by designing in a preamplifier before the intersil ADC. This needs to be extremely stable as it precedes the ADCs built in offset compensation (if that's the correct term). Hence Keithley implemented a chopping scheme based on 2 op amps, a discrete fet chopping switch and a jfet demodulation switch.

So my problem is a very unstable reading in the 20V and above range. It actually reads spot on in the 2V and below ranges, and ok-ish in resistance (may be some issue here but not as obvious as the higher voltage ranges). I didn't note down the current readings but off hand I seem to think that they were OK, but take that with a pinch of salt.

Initial observations: The voltage error is always greater in magnitude than the expected voltage i.e. positive voltages get more positive and negative voltages get more negative. The error is constantly drifting in a random direction. The voltage at the rear analog output terminals always matches the displayed voltage (taking into account the range setting). I.e the problem is before the ADC chip. The incorrect voltages can also be measured at the output of the input divider I.e. before the chopper, however there is no significant leakage within the divider/range switches (confirmed by wiring in an external divider which showed the same problems).

At this stage, the symptoms are pointing to excessive input bias current in the chopper amp, so I pulled the chopper fet and did some impromptu measurements, namely Vgsth and leakage current. As best as I can make out in my home lab (a tiny desk in the spare bedroom), the only parameter out of spec was the gate leakage on the input side, which was in the 10s of pA, rather than sub 1pA as expected. Everything else seemed spot on (1pA is about the limit of the sensitivity of the tools I had to hand).

However, this discrepancy seems too small to explain the observed voltage offset. Nevertheless, assuming that the feedback side of the chopper is less sensitive to leakage than the input side (due to the input divider resistance) I swapped the input and feedback chopper fets - they are in a single can which I flipped upside down.

This did indeed reduce the problem, however it didn't eliminate it completely, and pretty much brings us up to date. So my question to the floor is how shall I proceed from here:

1) Order a NOS chopper fet, which will cost a significant fraction of what I paid for this meter in the first place, and may not be guaranteed to fix this issue as I don't know enough about chopper topology to say that all the excess leakage is coming from this fet. It seems to me that fet gate leakage must only be a small contribution to the total chopper amp leakage.

2) Dump the original chopper amp and replace it a modern Zero drift/autozero type op-amp such as ADA4522, which I already have in stock. The possible issue with this being input noise current putting me back to square one. I haven't done the maths yet, but it's only a 4.5 digit meter, so may work out to be ok, plus i have other op amps in stock so there may be something better in my shelves.

3) Any other suggestions?

Anyway, sorry for the long post - hope someone finds it interesting

PS I checked the input protection for leakage and it was fine
PPS I forgot to take any pics, except for above mentioned  input protection, which is just 2 diodes so I won't bother posting it.

[Dr. Frank]

Hello,
I find your post interesting because:
1. I used this DMM in the beginning of my physics diploma, I think as a pre-amplifier with analogue output and maybe with a GPIB interface (not sure), anyhow, 34 yrs, ago
2. I have many very nice and capable engineering colleagues in Singapore, in our company
3. Have been to SGP and really like the people and this town/state.

The problem most probably is located in the switch, see cutout,  as the lower ranges work properly.
Try cleaning that part.
And leave that chopper and its FETs alone, please!!

Check the other modes and ranges, esp. Ohms, if you find something peculiar, to narrow down the root cause.

Frank

[Swainster]

Hi Dr Frank,
many thanks for your kind words! I'd like to make a disclaimer that I'm not originally from Singapore but i am married to a Singaporean and I have lived here for most of the last 22 years and fully agree with your sentiments.

I was hoping to avoid messing with the switch which is why I tried to eliminate it by substituting in an external divider for some testing. Having said that, cleaning this part is certainly a job worth doing as if it isn't the current culprit, it's a prime candidate for future shenanigans.

It looks like the only way to clean it properly is to remove the whole thing (it appears to be a single assembly) and dunk it in IPA. I'm not keen on putting the whole PCBA into IPA as older PCBs seem a lot more fragile than modern ones. I seem to recall that one of the other K177 threads mention a switch cleaning procedure or best practice, so will have another look through them. This might be a good excuse to get an automatic desoldering gun...

[Swainster]

Just to follow up on this, I never got as far as cleaning the gang switches. I thought that as I was going to be be doing some cleaning then I'd socket the chopper pre-amp ICs, then clean the whole PCB. However once I had fitted the sockets and put the old ICs back in then the drifting readings settled down and everything was working fine. I'm now wondering if the problem was actually caused by some aged solder joint. Either way, I can now complete the alignment procedure and electrically everything is looking good.

So next job is to address the cracked front panel. I was originally going for a laser cut piece of acrylic to replace the whole front panel, but once I had checked my 3D model for fit using the 3D printer, I decided to stick with 3D printing, and just fit a small acrylic lens over the display cut-out. This should eventually look more original than a plain flat piece of acrylic, plus I can cut the simple lens by hand instead of sending it off to the laser cutters. I'm also taking this opportunity to fit safety banana sockets, so I can directly use modern test leads. At least, that's the plan - the only tricky part should be transferring the old vinyl/polycarbonate label... worst case I guess I could get a new one printed.

[Dr. Frank]

Oh, very nice, that you also own a FLUKE 544X.

In such old devices, sockets pose a big problem, as well as whiskers.
Therefore, unplugging and re-seating all ICs is a good idea.
If  the metallic surfaces of both the IC or the socket look very matte, instead of shiny, then a  chemical cleaning or mechanical polishing might be helpful.

Bad solder joints can easily be visually inspected.

Frank

[Swainster]

Just to tie up a loose end, I've completed the front panel restoration and put the K177 back into "service" as one of my functional multimeters. The final mechanical parts of the restoration included:

• Swapping out the captive US power cord for a captive UK power cord - I didn't bother with fitting an IEC socket as I like the way the cable wraps around the feet and is gripped by a slot built into the rear of the casing
• Fitting blanking plugs for the missing handle
• Replacing the cracked front panel with a substitute 3D printed in ASA filament

The original front panel 'vinyl' label was transfered to the 3D printed substitute, and I used a printed 'lens' as a template to score some 2mm red acrylic with a craft knife. The resulting acrylic lens was a tight enough fit that I didn't bother with any other fixing method. The result is not as dark as the original so perhaps 3mm thick acrylic would have been a better choice, however this would more likely need some double sided sticky tape or other mechanical means to secure it in place. To the eye, the display is perfectly readable so I will leave it as it is.

If anyone else wants to have a go with the 3D printed parts then I'm attaching the models here. The mesh files can go straight into your favorite slicing software, and I've also attached the step files for people who want to do some modification.