Repair help for a not so old Keithley 2000

[Hole]

Hi!

I got an Keithley 2000 from ebay. Date codes look like somewhere around second half 2004, PCB has 2015 on it, front 2010.

Everything is *very* clean, nothing looks used, no dust, no spiders, it still has the protection film on the display. On the outside, ok, someone really didn't like it. Perhaps it was rack mounted and removed by a physically challenged or more experienced in agriculture.

Problem: it only displays "No Comm Link", no beep, no smoke, nothing. As far as I see this message is displayed by the display itself, so the base unit seems to be dead. Wiggling the connector did not help.

I poked around in power supply section, not sure where to look for.

CR 102 gives 49 V, CR 103 10.2 V, U144 5 V, U125 15 V.

U119 gives -9 V and is hot (ok, warm. I touch it and feel the temperature.). As far as I see in repair manual it shall give -15 V.

Before starting to poke more, anyone has a suggestion where to go for?

As far as I see, the caps look great. No swell, no fluids, nothing. Clean as can be.

Thanks!

[Macbeth]

Just a quickie - I remember getting this when messing around changing firmwares. The PROMs are PLCC socketed, labelled ODD and EVEN. Perhaps they have come loose from a bit of bashing in the post? Try reseating them.

[Hole]

I didn't remove them, only gave them some thumb presses... Still "No Comm Link".

[Smith]

NO COM LINK means the display unit is working, but it doesn't get any data from the CPU. You may try flashing the EPROM's. I know TiN had a unit with the same problem (another unit, but basicly it had the same internal type 2000 setup), and reflashing the EPROM's made the unit come back to life.

You might also check the caps, especially the ones near the voltage regulators. they tend to leak, and eat away the PCB. It's always a good idea to replace them anyway.

[TiN]

No comm means display board ain't getting love from main processor. Check the firmware (remove chips, read with programmer such as TL866) and verify its ok.

Your -15v LDO giving -9v is a bigger problem, so try to find who is sucking all the power, seem you have bad opamp or zener somewhere. I'd replace electrole caps anyway, 12 years is not that fresh.

[Hole]

Ok, caps replacement will be the first thing.

The firmware path is difficult, no programmer here...  Finding the bad opamp looks like I will first dig into the schematic to understand whats going on.

For me it sounds like the not-start-problem has nothing to do with the failing -15 volts.

So thanks so much, I will update this thread if anything moves.

[Hole]

Ok, removed PCB from case. It has a stamp that confuses me. 1435 could be read like 2014 week 35...

But anyway. The PCB is super clean, nothing to see, no holes, burns, patches, hand soldered, absolutely super.

[picburner]

As rightly Smith and TiN said, NO COM LINK means that the cpu fails to boot properly and doesn't communicate with the display circuit.
The cpu initially runs a diagnostic testing all the hardware, eprom, ram etc.
If a diagnostic fails it enters a loop and locked in this state.
If the eprom is bad, the cpu isn't able to run a program successfully, then intervenes the watchdog which resets the cpu constantly.
You probably have two distinct faults one in the digital section and one in the analog section.
The analog section could never function properly with -15V voltage lowered to -9V then you look for the reason.

[TiN]

It's easy to check date of manufacturing by date codes on ICs, such as DG411 switches etc.

[Hole]

So I desoldered all Elkos and gone throu them with my trusted Chinese $15 component tester:

C280 1     46 µF  ESR 0,64 ohm  0,8% Vloss
C279 2     45 µF  ESR 0,66 ohm  0,8% Vloss
C104 3     99 µF  ESR 0,24 ohm  0,4% Vloss
C131 4   2138 µF  ESR 0,13 ohm  0,9% Vloss
C148 5   2131 µF  ESR 0,15 ohm  0,9% Vloss
C146 6   2138 µF  ESR 0,22 ohm  1,9% Vloss
C156 7   6435 µF  ESR 0,10 ohm  1,6% Vloss

Doesn't look bad, or?

Board is as clean as it can be around the caps.

Resolder them or replace? I tend to replace them, as they are out already...

[Smith]

Although they still measure OK, they can still be bad or leak. Just replace them, next time you could be restoring the PCB because of leaking caps.

I really had caps before who measured quite good, but failed immediately at normal working voltage, which of course is much much higher than normal ESR meters work at. I even had disgusting leaking caps once that measured ok in ESR and capacity.

[Hole]

So. Some times later I found some time to poke around inside my K2000.

Caps are changed. And while changing caps I learned that the K2000 has two grounds, AGND and DGND....

So I rechecked the power rails, all within specs I would say: -15.95V, 15.95 V and 5V to AGND. 36,3V and 5V to DGND looks fine to me.

The analog voltage regulators become pretty hot but not that hot that they are untouchable. Seems ok, too. No other parts are hot.

I removed the 27C020 PLCCs, cleaned contacts, inserted back, no change.


My current verdict, following your advices: might be the 27C020 PLCCs are dead.

Any chance that someone can burn me those? Buying a EEPROM burner for that single case use is a little bit too much money.

Oh, and I have a "2000-803A19 ODD und 2000804A19 EVEN" firmware. Can I use the A20?

[Macbeth]

So. Some times later I found some time to poke around inside my K2000.

Caps are changed. And while changing caps I learned that the K2000 has two grounds, AGND and DGND....

So I rechecked the power rails, all within specs I would say: -15.95V, 15.95 V and 5V to AGND. 36,3V and 5V to DGND looks fine to me.

The analog voltage regulators become pretty hot but not that hot that they are untouchable. Seems ok, too. No other parts are hot.

I removed the 27C020 PLCCs, cleaned contacts, inserted back, no change.


My current verdict, following your advices: might be the 27C020 PLCCs are dead.

Any chance that someone can burn me those? Buying a EEPROM burner for that single case use is a little bit too much money.

Oh, and I have a "2000-803A19 ODD und 2000804A19 EVEN" firmware. Can I use the A20?

You really are going about this the wrong way - a TL866 is peanuts (like $30?) and believe me is a tool you will really end up using everywhere! That little 8 pin SOIC that holds the calibration data? Yeah, you can take a backup using your TL866 and reprogram it when you inevitably fuck it up (like changing from A19 to A20 - a big no-no unless you can afford a full Tek/Keithley recal!)

Pissing about buying OTP's and sending them back and forth at EU postal rates instantly is bad economy, but then how do you trust the guy who programs the device (I do not mean trust as in steal your money, I mean as in he burned them right). You could also end up blaming someone for sending you duff shit when he actually sent you something working perfectly. (Oh lord have I been a victim of "No good deed goes unpunished!")

I myself have cocked up some burns. Thankfully I chose the cheaper PLCC FLASH rather than ancient OTP and so could just overwrite them.

Also, just reading your original EPROMS using a TL866 and comparing them with a binary A19 will reveal if they are duds or not which I am highly suspect of.

Go on, spend a few Euros on a TL866 you tightwad!  Also don't be so tight as to get the barebones without a decent pack of adapters including your PLCC!

[Hole]

TL866 is about 50 € in EU.

[Macbeth]

TL866 is about 50 € in EU.

Is that all? Worth every penny.

[TurboTom]

...

Any chance that someone can burn me those? Buying a EEPROM burner for that single case use is a little bit too much money.

Oh, and I have a "2000-803A19 ODD und 2000804A19 EVEN" firmware. Can I use the A20?

If you like, I can flash the EPROMs for you (provided you've got a functional code file). Please PM me for details.

Cheers,
Thomas

[Hole]

So.

After replacing the EPROMs with two SST39SF020 I get the same error result like before.

I did some probing. Reset/Halt looks fine, goes up to 5V. Clk looks fine, 14,7 whatever MHz, as far my DSO can tell. There is no action on any (ok, I did not check every) line of the U151/U152-Ram lines beside one short spike at startup.

Any advices what to do next?