Keithley 2000 repair help

[funkyboerg]

Hi everyone!

I just got a used Keithley 2000, datecodes indicate end of 1997. On first try it powered up and seemed to work fine. I then brought it into my "lab" and it didn't want to start anymore: When I switch it on, it beeps once, all display segments are lit and that's it. It simply stays like this forever, no reaction to any button presses either.

- The inside looks ok, no obvious damages, blown caps or similar stuff. Just one bodge-wire on the main processor.
- All the supply rails and the reference are fine (although +-15V is at the limit of the spec with +-15V83).
- The oscillator for the main processor works.
- I poked all the memory chips with a scope. Interestingly the 24LC16B IIC EEPROM (U136) is never accessed.
- There is continuous activity on the bus for the RAM and OTP EPROMs, even after the display locks up.

I have the user/repair/calibration manuals and some schematics, but they don't really help for this problem. 
What do you guys recommend to troubleshoot/repair this thing? Although I have no idea when the meter was last calibrated, I'd like to retain the calibration values if at all possible.


EDIT: Possibly good news, I can get it to fire up. Just have to put it in the freezer for 10min 
So there is some kind of temperature dependent problem. My guess would be that some timing is out of spec at higher temperatures leading to faulty data in the self check at startup. Any hints to which component could be the likely cause?

Regards,
 Michael

[acbern]

can just do guesswork here. the first divices I would look for is failed el. capacitors. second, there may also be a fine crack in a pcb trace (e.g. trace to barrel separation), had that issue temp. dependent. you can check mechanical issues by carefully hitting against the boards with e,g the handle of a screw driver. simple, but has helped me a lot.
what is less likely, i would think, is that a crystal is out of spec or so.

[funkyboerg]

The caps all look good and the supply rails are also ok. I already checked the crystal, it is at 14.71MHz (with 14.7 printed on it), so that should also be fine.

Mechanical stuff might be possible, but that will be a bit hard to locate.

To test the temperature dependency, I ripped a peltier-element out of an old portable cooling box and put it onto some of the chips in the DMM. I can now repeatably boot it by cooling down one of the two OTP EPROMs (U156 EVEN). I also touched all the solder joints of and near the EPROM socket with a soldering iron, but that didn't change anything, so my guess would be that the EPROM is somehow out of spec on it's timings at room temperature.

I was hoping it would be the RAM, because that would have been a cheap and easy fix (it's standard 256kBit SRAM). But it seems I'll have to build myself something to read/write EPROMs, so I can copy and replace that chip. Of course the other solution would be to permanently mount the peltier on the EPROM, but I'm not sure if that could be considered a reasonable repair. 

[macboy]

The caps all look good and the supply rails are also ok. I already checked the crystal, it is at 14.71MHz (with 14.7 printed on it), so that should also be fine.

Mechanical stuff might be possible, but that will be a bit hard to locate.

To test the temperature dependency, I ripped a peltier-element out of an old portable cooling box and put it onto some of the chips in the DMM. I can now repeatably boot it by cooling down one of the two OTP EPROMs (U156 EVEN). I also touched all the solder joints of and near the EPROM socket with a soldering iron, but that didn't change anything, so my guess would be that the EPROM is somehow out of spec on it's timings at room temperature.

I was hoping it would be the RAM, because that would have been a cheap and easy fix (it's standard 256kBit SRAM). But it seems I'll have to build myself something to read/write EPROMs, so I can copy and replace that chip. Of course the other solution would be to permanently mount the peltier on the EPROM, but I'm not sure if that could be considered a reasonable repair. 

Another useful option is to use "freeze spray". It is used for exactly this purpose: Cool down specific components one-by-one to locate temperature dependent faults.

An inexpensive universal reader/programmer like the TL866CS can be had at all the usual places (eBay, amazon, dx.com, etc.) for $50 and up, depending on accessories, mainly sockets for surface mount ICs. I don't have one, but I want one. Dave checked it out in #411.

Even if you can read the EPROM while cold, you may want to find a known-good copy of the firmware to burn into the replacement chips. Unless you have a way of validating what you have read (like a checksum) then how do you know that every bit has read OK?

[funkyboerg]

Another useful option is to use "freeze spray". It is used for exactly this purpose: Cool down specific components one-by-one to locate temperature dependent faults.

I know, but unfortunately I don't have this stuff here. That's why I came up with the peltier element.

Even if you can read the EPROM while cold, you may want to find a known-good copy of the firmware to burn into the replacement chips. Unless you have a way of validating what you have read (like a checksum) then how do you know that every bit has read OK?

Well, the DMM does some kind of check at startup, which fails right now when warm. So I'd try to read the EPROM at different speeds and temperatures to see if I can recreate the error. I should be able to get a "good" read from a cooled chip. This data goes into a new EPROM and if the DMM boots, all is good. If not I'll have to repeat the process, which will get a bit expensive in the long run, as the EPROMs cost ~7€/pc.

There is a version of the firmware on the forums here, but apparently I would loose the calibration if I use that. and although the calibration is apparently from 2002, it's probably better than nothing.

[David Hess]

A flaky EPROM is possible.  Usually before completely failing, they can be read by lowering or raising the supply voltage slightly and some advanced programmers may even support this.

My suggestion before you found a problem with the EPROM was going to be the CPU oscillator or clock source and the power on reset circuit.