Keithley 2000 display repair

[aix]

The display on my Keithley 2000 has stopped working (stays completely blank). I would appreciate some advice on how best to troubleshoot this.

Here is what I have learned so far:

• Except for the display, the meter is functioning just fine: I can take readings over GPIB.
• The keyboard (which is part of the same assembly and controlled by the same MC68HC705C8A) also seems to work: the unit reacts to key presses as it would normally: relays click etc. I haven't tested the keyboard beyond pressing a couple of buttons and listening to how the unit reacts.
• The +37V rail that powers the VFD reads close to zero volts. (Pins 2/3/6/7 of U101: LM317M.)
• When I disconnect the display assembly from the main PCB, the output of U101 goes to +36.5V and can supply at least 200mA (as tested with a DC load).
• Supplying +37V directly to the display assembly from a bench power supply puts the power supply into current-limiting mode (at 200mA).

This leads me to the high-voltage part of the display assembly. It consists of a pair of HV5812 drivers and a VFD. There's also a pair of diodes on the back of the board (CR401 and CR402) whose purpose I don't know but that test short (in-circuit, so not sure what to make of this).

If you were in my shoes, what would be your next troubleshooting step?

Thanks for any suggestions and happy holidays!

[coromonadalix]

Check your tantalum cap (s)  if there is more

I would have desoldered the hv5812 one after the other by hot air to see if the shorts goes away with one or the two removed ??

I dont think the 68hc705 is short since its 5 vdc powered ??

Here in canada they dont cost too much (hv5812),  under 5$ cad

[aix]

I too think that the 68HC705 is fine since it can read the keyboard and talk to the main board. I'm pretty sure it doesn't use the +37V rail, although weirdly the +37V supply is routed to that part of the board (for reasons I haven't quite figured out).

My inclination was to remove/replace the HV5812s (I already have spares on hand). I am just a bit unsure whether my hot air rework skills are up for the job, given how close the glass of the VFD is to the two chips. Need to figure out a heat barrier of some sort?

Checking the tantalum cap is definitely a good idea, thanks. Will do that next and report back.

[aix]

The tantalum (C413) is fine (measures 21μF, 0.3Ω ESR).

So I'm guessing the next step is to remove the HV5812. Any practical advice on how to protect the screen and the ribbon cable from heat? (Or any suggestions for other things to try first?)

Thanks again.

[coromonadalix]

Some aluminum foild can deflect the heat ??  2-3 sheet thick would do fine, or Kapton tape

But the best would be to unsloder the vfd and remove it completly  .... my 2 cents   just to be sure you wont affect it, scratch it, and have plenty of room to work

If youre not to experienced, maybe someone here near your location with the right equipment could do it ??

I would have done for free to help you, but where worlds appart  loll

[martinr33]

You could remove the VFD first - easier than pulling the chip. It would be good to eliminate some kind of in-glass short.

Once it is off, if the 37V is still bad, you can pull the high side driver chip.

If the 37V goes good, time to suspect the glass.

[aix]

Anyone got any tips for how to remove the VFD without destroying it?

As far as I can tell, it's sitting on double-sided tape. Use dental floss to cut through that and then wick off the solder from the leads?

[Rasz]

You could remove the VFD first - easier than pulling the chip.

are you serious?

hotair: desolder ribbon cable, put a barrier (foil, knife blade, piece of metal sheet etc) between chips and vfd and take them off

[martinr33]

Use a solder sucker on the pins. Make sure that evey pin is loose in the hole when you are done by making it move with a small screwdriver. If the pin is loose, it will spring back easily. Sometimes, you have to break them free of vestigial solder with a push. Or you may have to refill the hole with solder and try again.

Once that is done, you can carefully remove the glass. I usually lever them off slowly.

[Inverted18650]

when I pulled the VFD on a couple, I put a dab of fresh solder on each pin first then used a solder sucker as described above.

[aix]

Thanks for your suggestions everyone.

I ended up removing the VFD and both driver ICs, only to ultimately track the problem down to a shorted ceramic cap. 

The display is now working again, but I can't help thinking that there must have been an easier way to track down the root cause. Thermal imaging maybe? Any other ideas?

[coromonadalix]

Glad it work again, 

For trouble shooting  when its possible, track down any schematics if available,  do power supply lines checks, checks for shorts in any signals / data lines ...

Get datashets for any ic's / transistors / mosfets  etc ... you see on the pcb,  may partially serve as service manual ...

And sometimes an 0.05$  part can give you lots of troubles  loll

When i troubleshoot,  i check all capacitors types on the pcb, if i get some low or stranges values, i remove them and recheck, and good lcr meter with serial / parallel mode test help a lot.

And sometime you can have luck  loll

[Rasz]

Thanks for your suggestions everyone.

I ended up removing the VFD and both driver ICs, only to ultimately track the problem down to a shorted ceramic cap. 

The display is now working again, but I can't help thinking that there must have been an easier way to track down the root cause. Thermal imaging maybe? Any other ideas?

cold spray, IPA, lips

[aix]

When i troubleshoot,  i check all capacitors types on the pcb, if i get some low or stranges values, i remove them and recheck, and good lcr meter with serial / parallel mode test help a lot.

Would you mind elaborating on how to use LCR parallel/series mode for this sort of troubleshooting?

[coromonadalix]

I dont want to go off topic  since your k2000 seems repaired

a few links
http://www.techni-tool.com/site/ARTICLE_LIBRARY/BK%20Precision%20-%20How%20to%20Use%20an%20LCR%20Meter.pdf
https://community.keysight.com/thread/25877
http://conradhoffman.com/hoffman_effect.htm
http://conradhoffman.com/capchecktut.htm
http://conradhoffman.com/cap_measurements_100606.html

[martinr33]

It is quite difficult to hunt down a short like that, especially if it only shows up at higher voltages.

Usually, there are too many suspect parts to make it easy. What you can do is measure the voltage across the rails at various points on the board. Where it is lowest, that is where the short is. You can also trace the voltage drop along the circuit - the voltage drop will follow the current path to the fault.

The voltage will be in the millivolt range. You can also use an external supply. The local supply may have enough ripple that you can use a current tracer like the HP 547A, which is a tricky useful piece of gear if you can find one.
https://www.eevblog.com/forum/reviews/hp-547a-current-tracer/


There are also instruments for tracking shorts - usually an audio osillator that responds to the resistance between points. As the frequency increases, you are approaching the short.