Restoration glory of Keithley 2001 DMM

[Zucca]

Ok, getting ready for my sick puppy.

Tamiya Extra Thin Cement - Ordered
12% Peroxide - Ordered

It will be a long journey... my K2001 was used in Canada to monitor the liquid temperature in a wine production process. It passed all the self checks... but there will be some surprise inside... if only my day would have 48 hours...

I think the display lens can be removed entirely as well.

I´ll try that... like macboy did

https://www.eevblog.com/forum/testgear/restoration-glory-of-keithley-2001-dmm/msg937701/#msg937701

[mimmus78]

So yesterday I was testing my new DB62 decade resistor and after many hours of switching and tests I found it maybe still in spec.
It was a maybe because I wasn't able to test the 100K Ohm and 1M Ohm with my K2001.

After two years of using it I found repair job I made two years ago wasn't complete because it still has a problem with high range resistance measurements. The problem seems affecting all Ohm ranges, but it became significant at 200K Ohm range, getting worse at 2M ohm, culminating with almost not working at 20M, 200M and G Ohm.

I noticed multimeter was out of specs in some Ohm ranges one year ago when I was testing my 100K Ohm 0.005% Vishay "standard resistor", but because I really never used this high Ohm ranges I never had a chance to realise it was so bad.

Initial diagnosis:

  - seems not a leakage problem, in G Ohm I can read OL
  - self test always passes with no error

First things to do is check current source circuit and all relevant analogue switching stuff. Most probably is one switch gone bad.
Actually digging in schematics and self test description ... to get a clue how Ohm measurement circuits works.
Seems I win a new repair for free.

[Zucca]

mimmus78, self tests no problems right?

And with these the calibration failed! Error 385, 20 MOhm 2W gain out of spec. High ohm resistors are a nuisance so I figured out that my soldering spree panic on the analog board left a mess affecting ohms current sources. After cleaning everything under a microscope, every guard trace, resistor cases, with dozens of cotton swabs and isopropanol the calibration worked!  I could even measure 1GOhm resistor.

is your board clean?

[Alex Nikitin]

Initial diagnosis:

  - seems not a leakage problem, in G Ohm I can read OL
  - self test always passes with no error

First things to do is check current source circuit and all relevant analogue switching stuff. Most probably is one switch gone bad.

I would first check the input current on DC Volts (10V and below) - leaky JFETs (protection and switching) would create a problem on DCV and OHM inputs.

Cheers

Alex

[mimmus78]

>> is your board clean?
>> I would first check the input current on DC Volts (10V and below) - leaky JFETs (protection and switching) would create a problem on DCV and OHM inputs.

I did a small repair on this meter, one analogue switch involved in current measurement was half working. The culprit was found on 20 minutes investigation, the component easily replaced and all cleaned with IPA.
I'm sure I didn't contaminated it, but who knows what did previous owner!?

I remember there was also a bodged mod on some JFET (Q550) that probably was made by previous owner. I posted a photo here few years ago of this bodge, but at this time there wasn't public available schematics so I wasn't able to speculate for what it was done.
Maybe it is some goofy attempt to repair the meter made by by previous owner for the same failure I fixed.

I can see some noise starting at 10K and it really show up at 100K, don't think contamination can have this high effect at this "low" Ohm levels.
10V is 11PPM away from another K2000 since two years. Noise at 10V is as specs and comparable with noise of other K2001s owned by other members here.
I used 100mV some time but I do not remember problems there.

Think I have enough stuff to work out for now ... and as usual any suggestion is appreciated.
I'll keep you all "K2001 aficionados" updated.

[mimmus78]

So this the bodge ... I checked with microscope and I can see some glue on the wire. At least it's seems some "definite" modification and not a temporary attempt.

Also 100mV range is clean.



Inviato dal mio Nexus 6P utilizzando Tapatalk

[lukier]

I got my meter warmed up, so I've decided to measure the Ohms current sources. Maybe someone will find that useful or other users with more reliable K2001's than my still-uncertain-recently-repaired one can provide their own data.

I've used Keithley 485 Picoammeter for all ranges except 20 Ohm, where it is out of range of K485 so I've used 3457A in the current range.

Ohms Ranges	Current
20 Ohm	~ 9mA
200 Ohm, 2 KOhm	0.981 mA
20 kOhm	89.3 uA
200 kOhm	7.09 uA
2 MOhm	772.7 nA
20 MOhm	70.72 nA
200 MOhm, 1 GOhm	4.394 nA

Keithley 2001-TCSCAN manual, in the calibration section (page 4-10) mentions that on the 200 kOhm range the sourcing current is 7 uA, so at least in this range my meter seems OK.

I've tried to measure the input bias current on the DC Volts low ranges, but it is well below my K485 sensitivity and measurement setup (BNC cable with a BNC to Banana adapter, no triax). Similary, I've tried connecting 1 GOhm resistor to K2001 on the DC range (200 mV) and the readings are just jumping around between -70 mV to +20 mV. 

[mimmus78]

So situation kinda getting worse.

Now self test always give me errors 304.6 and 306.7.

Sometimes also lower tests up to 304.3 give me some errors, but this depends on humour of the multimeter (maybe some part gets hot).

I figured out I have some leakage somewhere, here is what I found when running test 304.3.



This error give me just time to take a note because now 304.3, 304.4, 304.5 are working again.

So I made some analysing the circuit and I was almost sure my leakage was caused by one of this Q320, Q312 or Q324.

I removed all three fets, replaced Q324 with a 30ohm resistor between source and drain and by running test 304.6 I demonstrated I was wrong and leakage is still there.

Will continue investigation tomorrow ... if anyone have any idea how to solve this puzzle, I will appreciate.

Domenico

Inviato dal mio Nexus 6P utilizzando Tapatalk

[plesa]

I propose replace shift registers 14094 and all analog switches (DG411/DG404/DG211). It is much faster than troubleshooting these failures.

[mimmus78]

Yes plesa: I'm starting thinking to brute force attack too ...

Inviato dal mio Nexus 6P utilizzando Tapatalk

[plesa]

Yes plesa: I'm starting thinking to brute force attack too ...

Inviato dal mio Nexus 6P utilizzando Tapatalk

It is not brute attack, the components are cheap and it is more efficient to replace them all. It is difficult to find where leakage is.
Keithley also repair multimeters this way. They will replace few components and in case it does not help you will receive discount on new meter

[mimmus78]

Gotcha! Test 304.6 was ok without U325 ... little bastard!

Inviato dal mio Nexus 6P utilizzando Tapatalk

[plesa]

Gotcha! Test 304.6 was ok without U325 ... little bastard!

Inviato dal mio Nexus 6P utilizzando Tapatalk

Replace them all, it is really worthwhile. It needs to be calibrate again anyway.

[mimmus78]

Seems nobody in Europe has stock!
Ebay or get from USA?

[plesa]

Seems nobody in Europe has stock!
Ebay or get from USA?

wait what you want to replace?
DG411 and shift register MC14052 Farnell has on stock.
If you needs something special drop me PM, I will check my local inventory for Keithley repairs

[lukier]

Seems nobody in Europe has stock!

What about Farnell?
http://uk.farnell.com/on-semiconductor/mc14052bdg/4000-cmos-smd-4052-soic16-15v/dp/9666419

[mimmus78]

I need MC14052BCP in DIP 16 package.

It was not ROHS part, so nobody use here Europe anymore.

I asked some eBay guy if he is sure parts he sells is genuine ... let see tomorrow what will reply.

Anyway if you have this part, we can arrange something.

Inviato dal mio Nexus 6P utilizzando Tapatalk

[Kleinstein]

You could likely get away with an CD4052B part. It might be slightly different parameters but most of them overlap and speed should not really matter.  This one is available, e.g from tme.eu.

One might have to select for low leakage samples: typical values are good, but worst case performance is likely not good enough.

[plesa]

I need MC14052BCP in DIP 16 package.

It was not ROHS part, so nobody use here Europe anymore.

I asked some eBay guy if he is sure parts he sells is genuine ... let see tomorrow what will reply.

Anyway if you have this part, we can arrange something.

Inviato dal mio Nexus 6P utilizzando Tapatalk

http://se.farnell.com/texas-instruments/cd4052be/ic-mux-demux-4ch-4052/dp/9589538

[2N3055]

RS has CD4052BEE4 ... RS cat no : 121-9243

it might have a bit better leakage... they are cheap, maybe buy few and sort the best ones...

[mimmus78]

Ti CD405xB seems to be OK, but compared to original Motorola part it has is far less flat in R(on) response ... because this IC is part of the resistance measurement I think R(on) flatness if very important.
So or I stick with Motorola part or I need to check if this R(on) flatness is really important there.

[2N3055]

I think they are in high impedance path switching ranges...
 But whatever you do change BOTH 4052 to have same type... Looking at the schematics, it looks like they are balancing them both...... So having two 4052 of the same type, manufacturer might help...

Now you made me curious...  I'm gonna look into circuit a bit more later....

[mimmus78]

Got 15 items from a DE ebay seller I hope they arrive in reasonable time ... I found this part is also classified as military part with export restriction.

[mimmus78]

Got 15 items from a DE ebay seller I hope they arrive in reasonable time ... I found this part is also classified as military part with export restriction.

The waiting for parts is killing me ... really tempted to go to local electronic parts store and slap in a 4052 or whatever compatible one.

Anyway specs are 30ppm for lowest ohm ranges so I think they are calibrated out some way.

There is ohmcal signal going to ADC ... and it's just before switch, I think this is used to calibrate out switch resistance.

[TiN]

Huh, I remember there were some posts talking about LTZ1000 upgrade on K2001, which I wanted to comment on, but they are not here anymore?
Or it was other thread? Yup, it's here! 

I'll be opening one of 2001s soon, could try some stuff! 

K2002 btw using MAX326 ultra-low leakage switches for current source switching instead. They are available in DIP too, but I didn't check if pinout differs.