Keithley 2010 Repair

[KungFuJosh]

I've jumped in and I'm going to try and get this 2010 working properly. It mostly seems to work, but it has a couple fails.

Consistent fails:

Bank 100 A/D Converter
101.2 TestCal

Bank 400 VAC
400.2 Non Inv Path

Only showed up once after hours of warmup:

Bank 300 DC/OHM
305.2

Bank 600 AMP/OHM
600.1 Ohm/Amp
600.2 Ohm/Amp

Voltages:
R271 TF-245
1: +12.4207
2: +6.21061 (too low, right?)
3: -12.4196
4: +12.4208
5: -12.4193
10: +/- 3V (unstable)
13: -24.85mV
14: -4.1384
16: +11.7987

U179 79M15CT
1: -21.8342 (too high?)
2: -15.8288

U180 L78M15CV
2: +21.8369 (too high?)
3: +15.1885

CR102 WM18 - is this bad? 18V outputs ~22V?

The electrolytics look fine without any signs of leakage, but I'll replace them all anyway to be safe. In case anybody else plans to do the same, here's what was installed in mine:

Electrolytics:
C104 100μF     63V 10mm⌀ 5mmLS
C131 1mF    50V 13mm⌀ 5mmLS
C146 2.2mF    16V 13mm⌀ 5mmLS
C148 1mF    50V 13mm⌀ 5mmLS
C156 6.8mF    16V 18mm⌀ 7.5mmLS
C276 470μF    63V 13mm⌀ 5mmLS
C277 470μF    63V 13mm⌀ 5mmLS
C281 470μF    63V 13mm⌀ 5mmLS
C284 470μF    63V 13mm⌀ 5mmLS
C296 47μF    35V 6mm⌀ 2.5mmLS
C297 47μF    35V 6mm⌀ 2.5mmLS
C104,C156,C296,C297 VX(M). All others VZ(M).

Visual inspection didn't find any crazy standouts. There was a little bit of an odd color on the PCB around U179 & U180, and that's why I tested those.

I'm not sure what voltage is supposed to be at TP105 (I assumed around 7V?) but I get unstable voltage there, and it looks very weird on my scope.

Any help is appreciated.

Thanks,
Josh

ETA: Corrected capacitor value for C104. Oops.

[BillyO]

I'll be following this.  Thanks for sharing your journey, Josh.

[Cyclotron]

In my experience, it's a good idea to back up your constants before going too far.
I've repaired 2-3 K2001Ms but not K2010, so I'm not much help, but I'm interested in following along.

[KungFuJosh]

Thanks, that's probably a good idea. It already has the latest firmware for that hardware version, so I wasn't planning on messing with the chips...but you're right; I should back them up anyway.

[Kleinstein]

AFAIK the K2010 ADC part is very similar to the K2182. It uses an extra low noise zener diode for the actual ADC and the LM399 reference is only used for the longer time scale. The ADC part is still similar to the K2000, where one can find at least a reverse engeniered schematics.
The actual ADC of the K2001 is quite different, but it also uses a similar system with extra reference for the ADC.

So the 6.2 V at pin2 of the TF245 looks still OK (though my guess would have been more 6.4 V), as the low noise zener is lower voltage than the LM399.
Some 22 V in for the +-15 V regulators are sensible.

TP105 looks a bit like the ADC input / main amplifier output. So one may see the steps in the AZ cycle (with reference readings in some modes).

[zepto]

I'm not sure how valid these tests are since they appear in no official 2010 documentation.

You can step through the tests to see some of the values, but the k2000 descriptions won't match exactly. You can also use them to freeze certain configurations before probing. Different meters will have different values, but some differences/sums/ratios should be within certain tolerance.

Some example values:

101.1 (step can't fail) value : 0155081261
101.2 (pass) value: 0224570428
difference: 69489167 -- k2000 manual says should be between 74475970 and 78075970, but as Kleinstein mentioned 2010 uses a different reference with a lower value.
101.3 value 1.002744 volts -- k2000 manual says should be between .97 and 1.09

another run:
101.1 0155081180
101.2 0224570417
difference:  69489237
101.3 value 1.002743 volts

400.2 seems to fail if below 4.5 volts -- K2000 manual says 5.6 +/- .6V.

[KungFuJosh]

Some example values:

Where/how did you measure to get those values?

Thanks,
Josh

[zepto]

You have to manually select and step through the test. It pauses after each one and you can press shift.

[KungFuJosh]

You have to manually select and step through the test. It pauses after each one and you can press shift.

I assume you're measuring with a DMM at a test point? Or are you getting values from the screen?

Thanks,
Josh

[zepto]

It will display the value on the screen when you hold shift. I don't remember the exact wording to get it to pause after each step. bit/manual/step, something like that.

[KungFuJosh]

For anybody playing at home:
Hold "Open" button when powering on K2010.
Press Shift then Test to enter test mode.
Choose Built-In test
Bit: Manual (then choose which numbered section to test)
Mode: Step (so you can press Shift at each spot to see the value)
Fail summary listed at end (press Shift at each fail for description)

I tested everything, but these are the relevant values:

100.1 0155467966
100.2 0155466648 (failed on first round "SIGNAL NOISY" - no fail later after warmup)

101.1 0155466349
101.2 0232244055 (Fail: "NO 7V AT A/D")
101.3 1.004392

400.2 5.172274 (Fail: "NON INV PATH")
400.3 5.172274

Thanks,
Josh

[KungFuJosh]

I made two BOMs for recapping the 2010. One for Nichicon, and one for Panasonic.

Nichicon: https://www.mouser.com/Tools/Project/Share?AccessID=d1992fae1e
Panasonic: https://www.mouser.com/Tools/Project/Share?AccessID=bd368c9362

I'll probably get the Panasonics. Any thoughts on those choices? They seem like they should last forever if they all fit.

Thanks,
Josh

[GigaJoe]

10K hours should be fine,  ( personally i prefer Nichicon ,  his chemistry more refined ,but it my personal preference )
as they pack quite closely, you tolerance per diameter are fixed, but nothing bad to go taller and a bit more capacity ...
there thing, you can measure an actual voltage on 63V caps, an actual may be 38V ( really not remember ) so 50V would be OK too...

[KungFuJosh]

Never mind about the weird diameters/LS. Now that I removed the caps, I confirmed what they were supposed to be, and corrected the BOMs. There's two 1mF 50V and one 100μF 63V, and they're all the correct diameters/LS now.

OP updated with correct cap also.



Thanks,
Josh

[Cyclotron]

I made two BOMs for recapping the 2010. One for Nichicon, and one for Panasonic.

Nichicon: https://www.mouser.com/Tools/Project/Share?AccessID=d1992fae1e
Panasonic: https://www.mouser.com/Tools/Project/Share?AccessID=bd368c9362

I'll probably get the Panasonics. Any thoughts on those choices? They seem like they should last forever if they all fit.

Thanks,
Josh

I really like the Panasonic FR series for SMPS. I still use Nichicon and Rubicon too from time to time.

[J-R]

So did any caps test bad or no?

[KungFuJosh]

So did any caps test bad or no?

I haven't tested them yet, but I probably will. The failure behavior seems maybe directly related to AC filtering, and it's a good idea to replace the electrolytics anyway. I guess in that case testing them is secondary.  

[KungFuJosh]

Original cap tests...

[KungFuJosh]

More...

[KungFuJosh]

The 470s...

[Uunoctium]

tl;dr
The orange electrolytic capacitors from EPCOS were of very high quality. Some of them were branded SIKOREL, a name they had adopted from their predecessors, the black ‘Siemens LL’ (Long Life) capacitors. They were automotive-specified, rated at 135°C. Last year, I was able to buy some of them, 6,800µ  from DK in this configuration and was particularly impressed by the MTBF in the datasheet: 20,000 hours. I used them for a K2000 or 6485. Unfortunately, this brand has disappeared since the fusion/takeover by TDK. I don’t know whether similar grades can still be ordered from TDK.
Another brand that is obviously also of very good quality are the electrolytic capacitors from Wurth.

And by the way, provided they fit, stick a couple of clip-on heat sinks on the two TO220 voltage regulators. They run quite hot – something I like to avoid with semiconductors. In the K2000, this will fit without any problems

[KungFuJosh]

I've used some high value caps before from EPCOS with no issues. Good stuff, and they were reasonably priced too.

For this, I've already ordered the Panasonics. I've never had issues with those either, and 10000 hours is plenty. The VZ caps that were in it were only rated for 1000 hours.

I don't think there's room in the K2010 for more heatsinks, but I'll check again when I put the new caps in.

Thanks,
Josh

[BillyO]

Does that list include the tantalums?

[KungFuJosh]

Does that list include the tantalums?

No. I didn't find any shorts on the tantalums, but I didn't check values either. I'm going to see if changing the e-caps makes a difference, and if not, then I'll dive a little deeper.

Thanks,
Josh

[KungFuJosh]

And by the way, provided they fit, stick a couple of clip-on heat sinks on the two TO220 voltage regulators. They run quite hot – something I like to avoid with semiconductors. In the K2000, this will fit without any problems

I had some heatsinks that fit, but I don't love how big the screw heads are. Might be fine- I'll see how it looks after I install the new caps. I might order some shorter flush mount screws; the ones I have are too log.

Thanks,
Josh