Keithley 2002 repair (Help please)

[SelectLOL]

I think you probably jumped the train soldering new parts. From your previous photo it looks like you are in full treat with leaked electrolyte. So ALL components from area around 5cm in each direction from where cap was must be desoldered, cleaned with IPA, all PCB via's soldered thru, and cleaned again with Acetone, then distilled water, then IPA again, dried, cleaned again. Then put new parts back in. This procedure is mandatory for any old (15+ year old) Keithley 2001/2002, unless you want to continue "repairing" it till cows come home 

I have old video on my kei2002_u2 unit from 2016. This should show you the process. I still have this very meter, and it's happily calibrated and used as my daily bench meter.

Hello TiN thank you for joining in. I have seen most of your videos and whats on your website, amazing work, thank you alot for that I am guessing that the caps have already been replaced once, besides one that somebody deemed to not be worthy to replace. When my new parts arrive (MOSFET and Resistor) I will do what you said to the parts close the Capacitor that leaked. And after that I can continue with my measurements to find more defective stuff. 

[SelectLOL]

Excellent Job SelectLOL.

Thank you, thank you but the job is still to be done  Luckliy I have so many awesome contributors here, I am quite confident 

Please check this

https://www.eevblog.com/forum/repair/this-keithley-2001-need-to-be-saved!/msg2883864/#msg2883864

before ordering a new R100 which is difficult to find anyway... such a unicorn...

I do not like thermalfuse, and I would avoid them....

Amazing, I have read the whole thread, it was as funny as it gets  What is the current status of your device, did you manage to clear every error? Is the resistor you installed holding up fine?

And to the Resistor, I have seen that stupid thermalfuse on marco reps's video: https://youtu.be/HVDQeJTEHFI?t=344. And luckily I have ordered the exactly same resistor as you did, I say luckly because when it worked for you, it hopefully will for me too 

Marco reps didnt go into detail which resistor he has used for his repair in the end. But at https://youtu.be/HVDQeJTEHFI?t=432 one can clearly see that he had the same idea 

The digikey link for the resistor R101 Ke2002:  https://www.digikey.ch/de/products/detail/ohmite/HS10-470R-J/5307525

I had to put a little more thought to replace the MOSFET Q101 Ke2002. The obvious parameters like cont. Current, max. Wattage etc. were easy to sort and chosse from for the right part. But I was afraid of the driving range of the feedback circuit that drives the FET. I chose 2 FETs with similar input voltage sensitivitys to test. My thinking might have gone over the board, but Id rather be caucious and select the right part (or buy two/many) to not delay the repair 

So the links to the ones I bought: https://www.digikey.ch/de/products/detail/taiwan-semiconductor-corporation/TSM13ND50CI/9602845 and https://www.digikey.ch/de/products/detail/alpha-omega-semiconductor-inc/AOTF12N60L/6946477

I will test the first one first and if it seems to be operating fine, I wont bother to change it for the other one.

The parts will probably arrive Monday and the journey will continue then I will keep you all updated, thank you all very much 

[Zucca]

What is the current status of your device, did you manage to clear every error? Is the resistor you installed holding up fine?

It was fixed before I moved to USA. Once I will setup my lab here in USA properly, I will let you know if it is still okay.
I need to do a calibration anyway....

Yes the power resistor without thermalfuse I got is simply perfect, even the size is perfect!

[Per Hansson]

I had to put a little more thought to replace the MOSFET Q101 Ke2002. The obvious parameters like cont. Current, max. Wattage etc. were easy to sort and chosse from for the right part. But I was afraid of the driving range of the feedback circuit that drives the FET. I chose 2 FETs with similar input voltage sensitivitys to test. My thinking might have gone over the board, but Id rather be caucious and select the right part (or buy two/many) to not delay the repair 

So the links to the ones I bought: https://www.digikey.ch/de/products/detail/taiwan-semiconductor-corporation/TSM13ND50CI/9602845 and https://www.digikey.ch/de/products/detail/alpha-omega-semiconductor-inc/AOTF12N60L/6946477

I will test the first one first and if it seems to be operating fine, I wont bother to change it for the other one.

Just a note on the MOSFET, the most important specification in this implementation is not RDS(on) or the max Drain Current (Id) you can go with much better values than the original and it will not cause problems.
What is important is that the Gate Charge in Columbs (nC) is similar, because that is how much drive is required for the device to switch, which is of course matched to the U100/U101 photovotaic opto-coupler's drive strength.
And following that the important parameters are the actual switching voltage (Vgs th).
Just doing a quick search on Mouser I found these ones in stock: Vishay SIHF30N60E-GE3 and Infineon IPA65R045C7XKSA1 where the latter seems like the best match.
Also worth keeping in mind these are physically smaller than the original package, but maybe that is not a problem, the proper package would be TO-247 but I think the availability of an isolated MOSFET in that package is much less...

[SelectLOL]

What is the current status of your device, did you manage to clear every error? Is the resistor you installed holding up fine?

It was fixed before I moved to USA. Once I will setup my lab here in USA properly, I will let you know if it is still okay.
I need to do a calibration anyway....

Yes the power resistor without thermalfuse I got is simply perfect, even the size is perfect!

Ah amazing, I wish you all the best setting up the lab there 

Mine came today it is really a good fit, smaller than I expected. I will be posting some pictures later on. 

[SelectLOL]

Just a note on the MOSFET, the most important specification in this implementation is not RDS(on) or the max Drain Current (Id) you can go with much better values than the original and it will not cause problems.
What is important is that the Gate Charge in Columbs (nC) is similar, because that is how much drive is required for the device to switch, which is of course matched to the U100/U101 photovotaic opto-coupler's drive strength.
And following that the important parameters are the actual switching voltage (Vgs th).
Just doing a quick search on Mouser I found these ones in stock: Vishay SIHF30N60E-GE3 and Infineon IPA65R045C7XKSA1 where the latter seems like the best match.
Also worth keeping in mind these are physically smaller than the original package, but maybe that is not a problem, the proper package would be TO-247 but I think the availability of an isolated MOSFET in that package is much less...

Thank you very much, thats exactly what my seniors told me too. The Gate Charge and Vgs are very important. I have had a look at your MOSFETS, the latter one to be specific is the one, which I will do a little overview with:

Value		Vishay SIHF30N60E-GE3		TSMC TSM13ND50CI		Original 2SK1451
Vgs th.		2.8V		3.8V		2V-3V
Gate Charge		85nC @??V		39nC @10V		???nC @?V

The table above shows a little comparison between the one you recommended, I bought and the best I could find for the original FET from left to right (pdfs attached).
The Vgs voltage of the one you recommended seems to be good, would you say mine wouldnt work as well/is out of range? When buying I compared with the output of the opto-coupler, which seemed to be able to drive even much larger FETs perfectly fine. Should I stick with my FET or change it?

The package for my current one fits pretty fine, it smaller then original but its shape is identical, so it presses nicely against the wall of the Keithley.

I will make a post to what happenend when I started up later on. Thank you very much again 

[SelectLOL]

Hello everyone, here is the latest update to the thread

My parts have arrived and I already soldered them in. Pictures in the attachments.

The Ke2002 booted just fine as expected. Without the weird bootlooping, that I had before.

Currently I dont have the Flir on my hands, I will have to lend it out tomorrow again. I measured a little with a temperature probe and the FET and Resistor seem to be at ambiente. The Regulators are a little hot, about 65°C, but I couldnt measure it accuratly.

I have measured the +-15V rails and they seem to be bang on. About +-150mV difference and totally in spec.

I dared to do a self test for funsys, the errors are as following:

401.1 Regulator
405.2 Abs ckt. *10
405.4 Abs ckt. *10
405.6 Abs ckt. *10
406.6 Sample/hold
409.6 /500 freq cmp
410.1 TRMS
411.1 TRMS filter
411.2 TRMS filter
500.7 Amps/LO ohms
600.2 Ohms sense

We have gotten rid of some of the errors Tomorrow I when I have the thermal imager lend out, I will measure out the pre-regulator again an oversee the temp. of the regulators. When I am assured of all the voltage regulation, I will start tackling other problems 

To be fully hones, I tried to measure it out again with my Fluke DMM and when measuring between pin 4 and 6 where the output is, the transformator brummed again. This time though it probably didnt break again 

Thank you very much and I wish you a nice evening/morning/midday 

[Per Hansson]

Just a note on the MOSFET, the most important specification in this implementation is not RDS(on) or the max Drain Current (Id) you can go with much better values than the original and it will not cause problems.
What is important is that the Gate Charge in Columbs (nC) is similar, because that is how much drive is required for the device to switch, which is of course matched to the U100/U101 photovotaic opto-coupler's drive strength.
And following that the important parameters are the actual switching voltage (Vgs th).

Thank you very much, thats exactly what my seniors told me too. The Gate Charge and Vgs are very important. I have had a look at your MOSFETS, the latter one to be specific is the one, which I will do a little overview with:

Value		Vishay SIHF30N60E-GE3		TSMC TSM13ND50CI		Original 2SK1451
Vgs th.		2.8V		3.8V		2V-3V
Gate Charge		85nC @??V		39nC @10V		???nC @?V

The table above shows a little comparison between the one you recommended, I bought and the best I could find for the original FET from left to right (pdfs attached).
The Vgs voltage of the one you recommended seems to be good, would you say mine wouldnt work as well/is out of range? When buying I compared with the output of the opto-coupler, which seemed to be able to drive even much larger FETs perfectly fine. Should I stick with my FET or change it?

I used the MOSFET from the Keithley manual part list: Motorola MTG9N50E (I attached the datasheet) when looking for replacements.
This is quite allot better characterized than the (what I assume is a Sanyo) 2SK1451 that your unit apparently had installed.
That said the most important characteristic is missing from it, but it really seems quite similar if you eyeball it from the graphs:

[Kleinstein]

The gate charge has 2 components:
1) the gate capacitance times voltage change at the gate  (e.g. 5 or 10 V)
2) the gate to drain capacitance times the drain soure voltage.
With higher voltage (e.g. > 50 V)  the 2nd part is often the dominating part. For the charge specs on also has to look at the conditions: one gets a higher gate charge with more drain voltage. Especially with modern parts the tricky part is that the drain capacitance can be quite nonlinear. So Crss changes with voltage and gets quite a bit smaller with higher voltage. Modern parts got better be reducing Vrss at higher votlage.

If there are no gate charge specs, one could still use Crss as a crude alternative parameter, but whatch out for the conditions. If the curve is given one could integrate the area to get the charge.

With some 10 µA of current from the coupler and some 50 nC the switching speed would be some 5 ms, so rather slow, even compared to the rather low frequency (100/120 Hz).

[SelectLOL]

Hello everyone, I hope you all are doing fine  It sadly took a little longer than I wanted to post a new update. 

But dont worry, I was not doing nothing! First of all thank you Kleinstein and Per Hansson for the replies related to the Pre-Regulator FET. If it really seems to be a problem, I will switch it out 

In the meanwhile I bought a thermal imager that has a little bit more resolution that the Fluke I was using. And I was in the search of an isolating transformator for my measurements.


When I got my hand on both I started measuring with my Scope. I used my 4-CH Hantek with 1GS/s and wasnt pleased with the resolution in roll mode over >10 seconds.

So I tried fixing my LeCroy, which was out of service since I tried to switch from an HDD to an SSD. I reverted all my changes and got it back running. 

I will post both the oscillograms of the Hantek and LeCroy scope.

I am still trying to figure out the Pre-Regulator, because it still doesnt seem to operate correctly as far as I can tell.


Lets start with the oscillogramms from the Hantek. In the picture (Ke2002_Pre_Reg_Meas.png) one can see what Channel and Color corresponds to where I measured. All referenced from the Source of the Control FET (Q608 Ke2001 schematic)

Since I dont have a differential probe to seperate my ground I couldnt measure the Control FET out together with the Pre-Regulator FET. Thats why I put 2 oscillogramms together in one picture. The blue signal is the Gate Voltage of the Pre-Regulator FET referenced from its Source.


And now the same measurement with the LeCroy. Since it has a higer sample rate of 2.5GS/s and more memory, I was able to save the 3 Signals and measure out the fourth in one oscillogramm. I even was able to zoom pretty hard into the Gate signal of the Pre-Regulator FET as to be seen in the oscillogramm below. The measurement points are the same and are labeled directly on the LeCroy.

As far as I can see, after a while the Pre-Regulator seems to be getting its 100Hz like signal that Kleinstein mentioned. The problem is, that I really can't see the signal coming from the FET that controls the CONT line. How can that be? Is it working correctly?

If anyone wants me something to be measured out differently, I can do that without a problem now with the isolating transformer   


For the thermal imaging, one part is getting pretty toasty. It is labelled U108 and is located behind U107 and U109 (-15V and +15V) regulators. The U108 is an +5V regulator, same as the U110, why are there 2 of them? (page 19 Ke 2002 repair manual)

As far as I can tell the the similarities to the Ke2001 end there, since it doesnt have the same part in the same place as the Ke2002. Does anybody know what it supplies voltage to?

The thermal images show its temperature, it gets over 85°C in about 2 minutes.  

Thank you all very much and greetings

[saturnin]

I am still trying to figure out the Pre-Regulator, because it still doesnt seem to operate correctly as far as I can tell.

The pre-regulator in K2002 works in such way that it maintains voltage on C611 (big axial capacitor on the digital board) vey close to 7.50 V (the voltage is quite stable due to large capacitance of C611). If the voltage on C611 is correct, it is likely the pre-regulator works fine.

You previously reported +/-15V rails on the analog are in specs, is it also true for +/-BS voltages?

[IanJ]

If you look at the last 5 minutes of Part 5 of my Keithley 2001 repair video, and the Part 6 video I explain how the pre-regulator works including how the CONT line operates and demonstrate how to test the various parts with an LED, multimeter, DC PSU and variable transformer.

Ian.

[SelectLOL]

Thank for the reply saturnin and nice observation of the 7.5V!

I measured the BS voltage from -BS to +BS and it shows around 38.7V which is a little over the maximum and over where it should be  I physically measured here (first pic).

With the multimeter I measured some little jumps of the voltage rail, so I hooked up the scope and the voltage can be seen on the little gif I made. And of course a static image for a quick view.

The average voltage of the highlighted window can be seen under the Measure tab. It is weird that it tries to ramp up and then down again. Weird behaviour, something isnt right 

[SelectLOL]

Hello Ian Thank you for your reply. I watched your videos only until the 5. video and sadly only to where you explain the circuit and didnt even know that there is a 6. video

Tomorrow I will hopefully set up the circuit the way you did and try to measure out the Gate of the FET for CONT and apply some voltages to where you also did to see how it reacts

You bridged some relays for your device, how is it holding up? I am quite temped to just do the same 

[IanJ]

You bridged some relays for your device, how is it holding up? I am quite temped to just do the same 

The circuit is a clever design, but sadly it's not a great nor a particularly robust design. One small issue with the 120/240vac switching or even the pre-reg and it's goodbye just about every IC......hence why my digital board was a complete mess!

My 2001 is working great with the relays by-passed. However, I don't think you are at that stage yet.......I would test the CONT signal per my video first, easy and safe to do with no AC voltages involved........and move on from there.

Ian.

[Kleinstein]

Bypassing the 110/220 V relay may be a good idea. It is a clever design to automaticly decect, but if the detection fails and uses the 110 V setting with 230 V this is quite some stress.  Bridgeing that relay is more reliable than the automatic detection and thus more like a safe option and you only loose the 110 V use option that is not relevant in Europe.