Keithley 2602 dual SMU repair

[TheSteve]

Another purchase from everyone's favorite money pit, erm auction site.

This was purchased with a broken power switch and "input and output voltages are incorrect".

When I received it I saw that power push button itself was missing and output A did not work correctly - it always displayed "overflow" and had -2 volts on the output. The good news was that output B was just fine.
Fixing the power switch turned out to be pretty easy. The front of the switch was broken off and the push button cover was missing. Unlike many items like this that use a long plastic rod to a rear mounted power switch this unit has a small PCB with a connector and power switch bolted to the chassis near the front. I was able to find a compatible power switch from Digikey and also a new plastic push button cap for it. The cap I ordered was grey which matched Keithley plastic perfectly.

The internal construction of this unit is very nice. It has two SMU modules that are the same. Each SMU has two input connectors - a power connector and a control connector. It only takes 6 screws to cleanly lift out both SMU units. I swapped them side to side to verify the problem was with the SMU module and not the input power supply.
The SMU units are also easy to open - just a bunch of philips screws to remove a double sided PCB with shielding on both sides.
I compared various voltages and resistances between the good and bad units without it really pointing to a specific area/fault. After a bunch of probing around I decided to have look at the various "ECO's" these units received at the factory. There are a pair of TO92 cased transistors installed in place of SOT23 cased SMT parts. Well it was my lucky day. One of the transistors moved very easily. One of the legs was soldered to a PCB trace that had been ripped off the PCB. I ran a small jumper to the proper location on the board and the faulty SMU unit worked just fine again. I believe the 2602 received some sort of shock on the front face which broke the power switch, that same shock must have been enough to cause the PCB damage as well. I checked the second working SMU unit and the same trace was loose on the PCB but had not yet been ripped off the board.

So why does my 2602 have TO92 transistors in place of SOT23 devices? I thought maybe there was an error in the PCB layout but the SOT23 parts are correctly laid out. So if the pinout is correct what is it - power handling, or perhaps a matched pair of transistors? The transistors are a 2N3904 and a 2N3906 - would anyone actually match such generic transistors? I suppose it is possible both were repaired at some point with the through-hole transistors but that seems odd to me as it is identical in both SMU units.
I'll probably never know but as it is working again I can live with it.
Here are some pictures:

As received:


The power switch PCB:


Original and replacement power switches:


Suspect transistor pair:

[TheSteve]

More pics:

2602 with the cover removed - you can see the center mounted fan which blows on the heatsinks of both SMU units:


With both SMU modules removed you can see the main power supply and the processor/controller unit. You can also see the power switch PCB in the lower right.


Some other ECO rework - a jumper wire, an additional resistor and a pair of sot23 devices:

[TheSteve]

More pics:

It is interesting to see there is a location on the PCB for an additional IC and a network jack. The 2602A version has this network port installed(wish mine did too!). I only have GPIB and serial.

SMU unit with the top PCB shield removed, there is a second similar shield PCB on the bottom side. it also has a small fan which blows on the PCB.


The network port I wish I had:


The unit cleaned up and working great again:

[TiN]

 

Thread begs for high-resolution photos . Pity it does not have triax ports installed.

Perhaps worthy to find out what LAN PHY/MAC chip goes in there and perform an DIY "upgrade"? How hard could it be to solder little TQFP chipsy.
I'd expect firmware still have all the guts for network interface, if it allow you to select LAN settings in the menu?

What is the type VPG shunt resistor? VCS300 or 330?

[TheSteve]

The SMU unit boards would accept triax connectors, so someone really hardcore could swap out the phoenix connectors for them and then cut the back panel.
I will check on the shunt resistor next time I have it opened up.
Higher resolution pictures would be great - probably pointless unless I get something better then a camera phone for the pictures though.

The back panel:

[VintageNut]

Nice job bringing this beauty back to working condition.

There is a Phoenix connector to triax adapter. The part number is 2600-TRIAX.

The lowest measure range is 100nA. With 6 1/2 digits of resolution, you can resolve a solitary pA. I have seen up-close a user measuring 20pA of device leakage with a 2602B.

If you press the DISPLAY button, you can see all of the digits for one channel. The dual channel display truncates the measurements due to lack of space for all the digits.

Enjoy.

[TheSteve]

The 2600-Triax adapters cost a pretty penny, and you'd need two of them.

You can see the dual layout PCB here:

[Zucca]

Beauty, 40V 3A can cover a lot of stuff. TheSteve you seem to be a good EE sniper!

[technogeeky]

Another purchase from everyone's favorite money pit, erm auction site.

This was purchased with a broken power switch and "input and output voltages are incorrect".

When I received it I saw that power push button itself was missing and output A did not work correctly - it always displayed "overflow" and had -2 volts on the output. The good news was that output B was just fine.
Fixing the power switch turned out to be pretty easy. The front of the switch was broken off and the push button cover was missing. Unlike many items like this that use a long plastic rod to a rear mounted power switch this unit has a small PCB with a connector and power switch bolted to the chassis near the front. I was able to find a compatible power switch from Digikey and also a new plastic push button cap for it. The cap I ordered was grey which matched Keithley plastic perfectly.

The internal construction of this unit is very nice. It has two SMU modules that are the same. Each SMU has two input connectors - a power connector and a control connector. It only takes 6 screws to cleanly lift out both SMU units. I swapped them side to side to verify the problem was with the SMU module and not the input power supply.
The SMU units are also easy to open - just a bunch of philips screws to remove a double sided PCB with shielding on both sides.
I compared various voltages and resistances between the good and bad units without it really pointing to a specific area/fault. After a bunch of probing around I decided to have look at the various "ECO's" these units received at the factory. There are a pair of TO92 cased transistors installed in place of SOT23 cased SMT parts. Well it was my lucky day. One of the transistors moved very easily. One of the legs was soldered to a PCB trace that had been ripped off the PCB. I ran a small jumper to the proper location on the board and the faulty SMU unit worked just fine again. I believe the 2602 received some sort of shock on the front face which broke the power switch, that same shock must have been enough to cause the PCB damage as well. I checked the second working SMU unit and the same trace was loose on the PCB but had not yet been ripped off the board.

So why does my 2602 have TO92 transistors in place of SOT23 devices? I thought maybe there was an error in the PCB layout but the SOT23 parts are correctly laid out. So if the pinout is correct what is it - power handling, or perhaps a matched pair of transistors? The transistors are a 2N3904 and a 2N3906 - would anyone actually match such generic transistors? I suppose it is possible both were repaired at some point with the through-hole transistors but that seems odd to me as it is identical in both SMU units.
I'll probably never know but as it is working again I can live with it.
Here are some pictures:

As received:


The power switch PCB:


Original and replacement power switches:


Suspect transistor pair:

I wanted to point out that even the modern brand new devices like DMM7510 have TO-92 2N3904 and 2N3906 devices in them which Dave questioned in a teardown.

Any idea what they are used for in this setup? Even in very old keithley gear, like my 197, they are in matched pairs and are used for input protection and to connect the driven guard traces to ground.

[VintageNut]

The 2600-Triax adapters cost a pretty penny, and you'd need two of them.

You can see the dual layout PCB here:

Ok. Got it. I am thinking that the PCB you show may be dual use and the same for the 2636 of the same era. If so, you may also want to also find the pcb that has the LO triax pads.

You may spend as much for the 8 triax pcb connectors as is the cost of two of the 2600-TRIAX adapters.

The phoenix connectors are nice if you wan to fan out to some breadboard and you do not care about measuring pA.

All in all, a very sweet score. I absolutely love the KE 2600 series.

[HighVoltage]

Great repair thread of your 2602
You have a good hand in finding the good repairable instruments!

[TheSteve]

Great repair thread of your 2602
You have a good hand in finding the good repairable instruments!

In this case I just got lucky I guess. It was a best offer auction and I threw in a bid, they accepted it a day later. I thought for sure someone would have used buy it now.

[TheSteve]

One other tidbit of info I just remembered that might be useful to post. If you perform a firmware update on the 2600 series via the serial port expect it to take 1.5 hours. It transfers the flash image via 9600 bps serial which is a very slow process. The programming of the flash takes all of 30 seconds once done. So ensure your computer won't power down/fall asleep during the data transfer.

[SeanB]

Going to guess those transistors were selected out of the batch for lowest leakage current, and then used there. Easier to get low leakage on the bigger package, with the larger mass of epoxy reducing photoelectric effects to the die in the package.

[TheSteve]

SMU's often run quite warm, which can make diagnosing based on component temp a little tough.
Here are some thermal shots of my unit, taken 5 minutes after power on with the output turned off. Normally there would be a small fan blowing air over the components.

For TiN - the current sense resistor is a VCS301T 0R1000 1%

[rx8pilot]

Nice one!

Short and misplld from my mobile......