Keithley 776/R repair

[zortness]

I have had this Keithley 776/R for a few years, but it stopped working on me the other day, after re-arranging my workspace. This has the TCXO and 2.4GHz options.

I'm not an EE by trade or education, so it's possible I'm missing some easy diagnosis steps here. Any suggestions are appreciated, and I'm here to learn!

Unfortunately any copies of the manual I've found online have essentially unreadable schematics. I'm throwing a request out there in case someone has a printed manual to scan or a higher resolution digital copy so I'm not as blind here.

As far as the problem itself:
* no display, beep, or any action when turning the power on from the front panel switch
* I noticed the +/-5V regulators were very hot
* I've verified there is power running into and out of the mechanical front panel switch when it is enabled
* I've de-soldered the +12V, -12V, +5V, -5V jumpers on the bottom of the main board, and validated that the two stages of the power supply are working in isolation: first the +/-12V, then the +/-5V, measuring within spec on output side of jumpers. No excessive heat without the 5V jumpers closed.
* While poking around, before re-soldering the +/-5V jumpers, I noticed that the +5 far side of the jumper has direct continuity to ground.
All of the others have a few kOhms at least. My first thought is a tantalum cap went bad, or maybe a reed relay is stuck?

I've had to replace a couple of the Tantalums that had randomly popped in the past, but none of the remaining are showing any outward signs of failure.

Anyway, it's an inexpensive attempt so I went through and replaced all of the tantalums on the main board (not on the tcxo yet, but I was testing with it unplugged). None of the removed caps showed any signs of failure on the LCR meter and the +5V rail is still being pulled to almost zero (after reconnecting the solder jumpers), so the problem is elsewhere.

Onto removing and testing the reed relays?

[zortness]

Ok, I think I got it figured out. Unit is functioning again.

My unit came with all of the inputs in the rear, which means there's two long runs of coax from the front of the main board to the back panel. I have added my own front panel breakouts, but that wasn't the issue. One of the original runs of coax, from the Channel A input, had its shield pierced by an IC pin on the 5V rail trace that goes along the very front of the main board, very close to where the coax solders onto the board.

I was hunting for the issues based on the very low resistance between the 5V power trace and chasis ground (0.45 to 0.15 Ohms). I was following the 5V trace trying to methodically measure the resistance at each point. The resistance didn't actually change, but as I was holding the board up to the light to follow the trace, I moved the coax cable out of the way and the resistance suddenly jumped up into the kOhms range.

The coax in question is the white shielded in the photo. The black coax is what I had added previously for the front panel. (Don't worry, the 2.4GHz R module is a single run to the option unit, I just migrated that one from back to front).

Other things I had tried:
* I actually purchased replacement reed switches because they seem rare and I wanted to snag some, but while looking at the datasheets I was able to intuit which pins should be in which state. None of the coil pins were shorted to ground or each other, and all of the latches seemed to be in the right spot while unpowered. I guess now I have spares in case one does eventually die.
* I pulled out the power FETs on the 5V rail and tested them outside the board
* I pulled out the 5V protection diode and tested it outside the board
* I actually bought a cheap thermal camera to see if anything other than the 5V supply FETs were getting hot, which didn't lead to anything

[zortness]

Just to update this epic journey. The unit stopped working again shortly after my last post. It would read mostly random frequencies at the input, whether an input was being provided or not.

Backing up, shortly after I got the unit, a tantalum capacitor on the 12V rail died and shorted to ground. That was easy to identify and fix. Temporarily replacing it with an electrolytic cap to verify it worked, then I ordered a bunch of replacement tantalums and replaced every single one on the main board.

Now after this short on the 5V rail, the +/- 5V supply area was getting really warm and the unit had started reporting random frequencies. At first I suspected the TXCO board, as I couldn't see a signal on the output. After mucking about for a while, I dug back through the manual and realized it outputs 500 MHz, so of course my scope can't see it. Using the spectrum analyzer, I eventually validated the output of the TXCO board.

I have been frustrated that the scans of the original manual have essentially unreadable schematics. However, I eventually just slowly read through it and found some of the triage checklists. Going through that, many of the voltages were wrong, specifically tied back to the +5V line. This prompted me to once again de-solder the main components on there and re-check them. This time, the PNP transistor on the +5V side had two pins internally shorted. This seemed to let the +~7V out of the rectifier straight into the +5V rail, which probably sent the feedback circuit into fits, and made the analog reference voltages jump all over.

Replacing the transistors on the +/- 5V power section with similar stuff I had in the parts bin has brought the unit back into working order.

Fingers crossed it'll last a bit longer this time.

[garrettm]

I have the exact same model and option. IIRC, Keithley didn't actually make this specific frequency counter, its a re-badge of a Tabor Electronics unit.

https://www.eevblog.com/forum/testgear/kontrontabor-6010-universal-counter-still-worth-getting/

Anyways, I have a Fluke 81 pulse/function generator (which, ironically was also made by Tabor, but whose original design was done by Datron--who then got absorbed by Wavetek who was then assimilated by Fluke...) that had a tantalum fail shortly after buying it second-hand. I replaced it with a higher voltage part, but decided not to go wholesale slaughter on the remaining tants and its been fine ever since. I don't think one tantalum failing is reason to replace all of them. Maybe if two or three were to fail I'd do so.

[garrettm]

I forgot to mention, my 776/R had a bad electrolytic cap on the TXCO board when I got it. Unfortunately, I didn't take pictures of the repair (the cap had bulged and leaked out), but I'd suggest checking your TXCO board for bad caps too.

[zortness]

Thanks for the heads up. I also didn't know the units were made/shared with Tabor, so that gives me some additional resources. I did a cursory check of the TXCO board, but I'll give it a closer inspection.
Luckily no electrolytic leaks at this point.

[zortness]

Oh wow, thanks so much for pointing out the Tabor 6010. The manual from Tabor, while still not a spectacular scan, is infinitely more legible than the Keithley scans.

[garrettm]

I think the correct Tabor model is 6020. Yeah, thankfully, the Tabor 60x0 manuals have better quality scans of the circuit diagrams compared to the Keithley manual. Unfortunately, it looks like Tabor reused the exact same scans of the circuit diagrams for both the 6010 and 6020 manuals (I didn't check the 6030)... So no cherry picking the more legible diagram between them.