Keithley 2400 randomly turning-off/on

[asilva10]

I have a Sourcemeter Keithley 2400 that presents this problem if if nobody is using the instrument: it turns-off and -on again.

There are days that it works without a single issue. Besides the mentioned problem, everything else is okay with the instrument.

I measured the power rails and they don't have issues during the erratic events.

Any hints?

[Jwalling]

Must be a common problem with these:
https://www.ebay.com/itm/163269198642?

[TheSteve]

How clean is the output? Replace the electrolytic caps

[asilva10]

@Jwalling: yes, it seems to be a "common" issue..

@TheSteve: So far, I hadn't tried to replace power-supply caps because issues with the power rails were not identified. I mean, I put a DMM (Keysight U1272A) at each rail and I didn't observe any low/high fluctuation via the MaxMin/Peak feature of my DMM. Unless, it is something very quick that the DMM cannot capture.

[Brad O]

I think this is a known issue with the 2400s, so I can shed some light on this:

The original design of the 2400 used tin only contacts as part of the power supply board.  It turned out that these contacts (on one connector in particular!) were susceptible to fretting, a combination of wear and oxidation in this case. If it got bad enough, the fretting would cause an intermittent connection to the rest of the instrument, which results in the instrument randomly turning off and shortly coming back on.  There was a revision to the 2400s to change that contact to gold which prevents this problem, so you must have one of the earlier 2400s.  You can send it in to be repaired, or, if you want to try fixing it yourself, you can follow these unofficial, undocumented, warranty-voiding directions:

Lightly cleaning the problem connector is likely all that's needed to make a good connection again.  I wouldn't recommend sand paper as that may destroy the tin layer, but a soft cloth or a paper towel and a little bit of force will probably be enough to remove any built up oxide.  Getting to the connector is a little tricky though: it's at the bottom of the instrument so you will probably need to remove the analog board that's immediately visible when you open the box. 

To identify the problem child connector follow the power from the line voltage input connector, where you plug into the box.  There's a wire pair with brown and blue wires that carry the line voltage and take it immediately into a board, next to where the power switch actually controls the instrument's power.  Right next to that, the same wires come out of that board and carry the line voltage over to the internal power supply that creates the SMU's rail voltages.  On the far end of that board (directly behind the SMU's front panel) you'll see two connectors with black, red, and orange wires. One set goes to the fan (which is fine) but the other set goes to a different board on the instrument (see the photo I added), that's the one with the problem.  You would need to carefully unseat that connector (being careful not to pull the wires from the connector, they're just held in with crushed metal) clean the tin contacts on both the male and female side (maybe a very small wire cleaner would work for the female connector?) and then put in the instrument back together. 

I'm not going to say I fully endorse this method (I don't think I can since I told you to open the box), but that should keep you from seeing the intermittent power supply problem. 

Edit: I found an open 2400 so I fixed some descriptions and added a picture with an arrow at the bad connector, the one I found is a newer revision with gold contacts.  Also I'll stress that removing the top analog board is not exactly trivial, so if you're not confidant in your teardown skills, better to send it in for repair. 

[asilva10]

@ Brad O:

thank you very much for your attention and time!

Yes, the connector you had mentioned is the same one I was thinking about when following your instructions.

This kind of connector is really problematic - it is possible that I will replace it. During this week, I will work on this fix and see if the problem can be solved.

Again, thanks!!

[Macbeth]

Brad O - Holy fuck! There are still some humans working at Keithley Tektronix! 

[Brad O]

Ah good, you already had all the hard part done!  Glad to help!

@Macbeth: Don't you worry, Keithley's not going anywhere, we just added "a Tektronix company" to the name.

[asilva10]

@ Brad O:

I just concluded the maintenance tasks and so far, so good. I have a Python script running collecting data for many hours via serial interface - if the instruments reboots, I will know for sure.

Since I didn't have a new connector/pins with gold contacts, I have used an old technique from my days of "mainframe" computer maintenance. First, I removed all 4 pins from the female connector - what I would do next would damage that plastic case.

Next, I tried to put solder at the specific contact face that touches the male pin at the board. And you were right: there was an oxidation process at that tin layer, although not clearly visible. However, when I tried to apply solder at that part, the solder didn't spread as usual but a kind of a write layer came off with the heat. Next, I manually removed that layer and the ugly solder mixed with it. I repeated this process 3 or 4 times until a shiny and very thin solder layer was nicely covering that contact face.

I also carefully applied solder at the crimp part. As typical, when oxidation shows up at the contact pin, it also shows up at those kind of crimped connections. Again, it was difficult to apply solder but eventually a small amount of solder was applied successfully.  I had to repeat the overall process for all 4 pins - it took me around 20min to conclude the maintenance. I usually take a series of photos of the disassembling process and they quickly guide me when assembling back.

It is important to observe that it was a little bit hard to put the connector back due to the additional solder at those pin contacts, but it was finally attached with very very good electrical contact. If something happens again at the future, I would replace that connector with the 4-wire pigtail - I only forgot to measure in order to find the part from Digikey/Mouser. However, the lead at the solder is robust against oxidation - however, it easily wears out if we have to plug/unplug this connector multiple times, which is not the case.

If something bad happens during this week, I will report back. However, my intuition is that the problem was indeed solved.

Thanks!!

[Jwalling]

Must be a common problem with these:
https://www.ebay.com/itm/163269198642?

Those all seem to have disappeared quickly.