Author Topic: Restoration glory of Keithley 2001 DMM (Read 326749 times)

wiss · « **Reply #25 on:** December 15, 2013, 05:49:40 pm »

Fluke 8800a, Prema 5001/6001, Agilent 34461 and others do this, at least the fluke is a bipolar input with bias current compensation. You should worry if they don't behave like this.

TiN · « **Reply #26 on:** December 17, 2013, 02:27:58 pm »

Thanks, that's what I thought, but wanted to make sure

Started to recover schematics of analog board input protection area.
Not exactly clear function of everything there

TiN · « **Reply #27 on:** December 23, 2013, 04:18:23 pm »

Was lurking around on the web, and found on bbs.38hot schematics of Keithey 2000 meter. Local forum guy went same way I'm going now with 2001 and
reversed whole schematics except front panel

I managed to register on that forum and download it for history.

All credits for it go to jhzyou

TiN · « **Reply #28 on:** December 28, 2013, 08:45:47 pm »

Started cleaning-up schematics.

Here's PDF for checked parts:

ADC Board - ramp source?
ADC Board - gain amp and zero cross comparator?
ADC Board - Integrator

Can see how ramps are charging, and then discharge.
So if I'm correct - integrator charging cap switching VREF and ADC/IN and toggling bits by either positive or negative comparator into CPLD?

Channel 1 (yellow) is ZCTL2 net, Channel 2 (cyan) - ZCTL1 net.
Sorry, my third active probe for Tek did not arrive yet, so cannot use third channel to capture actual integrator capacitor voltage yet

Here's quick view of measurement outputs of comparators U803 and U804:

TiN · « **Reply #29 on:** December 29, 2013, 05:50:00 pm »

Finished redrawing analog ADC section.
Now need to get knowledge to understand it correctly

PDF Exported

Seem this project not much interested to volt-nuts here?
Or need more suggestions/feedback to be posted? I know there are guys here with K2000/2001's

I also hooked self-test passing unit to EDC MV106 to check stability:

Keithley #5 2001 measurement, 7.5d, REL -6.770493e-06, Adv filter 100, 0.005% RNG, Line-sync and autozero - ON, range 200mV
Meter and EDC powered on constantly, 2-wired connection to rear inputs, 2 sec delay
EDC Setting: 100mV range, Positive polarity, 0-10-0-0-0-0

Date	Measurement min	Measurement max	AC temp	Delta
27.DEC.2013 1:34am	10.01116 mV	10.01153 mV	23 C	0.37 uV
27.DEC.2013 8:49am	10.00971 mV	10.01073 mV	22 C	1.02 uV
27.DEC.2013 23:43	10.00933 mV	10.01079 mV	22 C	1.46 uV
28.DEC.2013 4:15am	10.00904 mV	10.01184 mV	22 C	2.80 uV
28.DEC.2013 16:08	10.00904 mV	10.01206 mV	22 C	3.02 uV
29.DEC.2013 0:11am	10.00781 mV	10.01466 mV	23 C	6.85 uV
29.DEC.2013 17:31	10.00781 mV	10.01466 mV	22 C	6.85 uV
30.DEC.2013 1:45am	10.00781 mV	10.01466 mV	23 C	6.85 uV

Galaxyrise · « **Reply #30 on:** December 30, 2013, 06:16:15 am »

Quote from: TiN on December 29, 2013, 05:50:00 pm

Seem this project not much interested to volt-nuts here?
Or need more suggestions/feedback to be posted? I know there are guys here with K2000/2001's

I've certainly been reading, anyway!

I did manage to track down the comparator pins and such while trying to find the ADC noise on a k2k (I did not figure out the noise.) Could watch the ADC input switch from ground, VRef, and VIn, watch the cap charging and comparator firing. I don't remember where any of those pins were now, except the ADC test point. I would loved to have had a pin out of the ADC!

casinada · « **Reply #31 on:** December 30, 2013, 06:56:56 am »

I follow your posts with great interest. Mine still works fine. I would be interested in the newest firmware........

TiN · « **Reply #32 on:** December 30, 2013, 07:52:43 am »

I posted K2000 schematics few posts ago already

Not done by me, i don't have K2000

I'd be interested in latest firmware too, but all have is A05, A08 and B07 (which I don't like).

Good news - Tektronix replied today that they can calibrate my 2400 and 2001, so i'll try to prep at least these for cal,
and check how much that would cost. Also no need to send to USA, they can do it in Taiwan lab, which is relieving

TiN · « **Reply #33 on:** December 31, 2013, 08:22:06 am »

Got a quote for 2001 and 2400 ISO9000 calibration with before/after data . It's around 600USD for 2001 DMM and 400USD for 2400 SMU.
I think I will do it after get all self-test and issues sorted out at least on two meters

neslekkim · « **Reply #34 on:** December 31, 2013, 03:56:41 pm »

$600 for calibration???, is it very difficult to calibrate such units on your own?

KJDS · « **Reply #35 on:** December 31, 2013, 04:11:30 pm »

Quote from: neslekkim on December 31, 2013, 03:56:41 pm

$600 for calibration???, is it very difficult to calibrate such units on your own?

The procedure for this type of equipment is usually fairly easy. The difficult part is having the very expensive test equipment that has a recent $2000+ cals from the national standard.

Dave · « **Reply #36 on:** December 31, 2013, 04:13:44 pm »

Quote from: TiN on December 31, 2013, 08:22:06 am

Got a quote for 2001 and 2400 ISO9000 calibration with before/after data . It's around 600USD for 2001 DMM and 400USD for 2400 SMU.
I think I will do it after get all self-test and issues sorted out at least on two meters

As far as I am aware, ISO 9000 has nothing to do with calibration. ISO 17025 is what you are looking.

plesa · « **Reply #37 on:** December 31, 2013, 04:16:58 pm »

Quote from: neslekkim on December 31, 2013, 03:56:41 pm

$600 for calibration???, is it very difficult to calibrate such units on your own?

2002 calibration cost about 1k USD. It is impossible to calibrated it at home lab.

TiN · « **Reply #38 on:** January 01, 2014, 05:20:34 pm »

Time to bring some infrared voodoo to fixing one of analog boards (board #4).
Took one meter, two analog boards and got some IR pics with Fluke Ti32, swapping boards.

Board #4 have next self-test failures:

304.3, 304.4, 304.5, 304.6, 304.7, 306.1, 309.1, 407.1, 407.2, 407.3, 408.6, 409.6, 410.1, 411.1, 411.2

Board #1 have few items failing initially, but after 5 min warm-up, it can pass self-test cycling over and over. So used it as a "working" reference.

Here are results of 30 min poking around with IR cam:

Board #4

Board #1

All seem similar here, so I looked over ohms/reference area, and spotted some higher temp parts

Board #4

R286 from CR335 near optocoupler U537

Little bastard resistor heating up +65C, CR536 over 45C, hmmm...

Passing board #1, same area

R286 little over ambient, +38C, CR536 dead cold +33c.

Another spot - Q540 under ADC board, near U324

On left side - passing #1 board, right is faulty #4 board.

Correct Q540 temp - 35C, while on failed board it's over +53c..

Time to draw schematics of these boys above and find out what's wrong...

Bonus:

Money shot, LM199 VREF

.
Taking into judgment that in Model 2001 whole analog section have thick plastic cover from both PCB sides with metalized internal surfaces to keep temperature constant on analog circuitry, probably VREF running over 70C.

More thermal images are on redmine related ticket.

plesa · « **Reply #39 on:** January 01, 2014, 08:43:42 pm »

I made similar measurement on 2002 and the voltage reference seems to be quite cool (50 deg) but the 5V regulator 78L05 U108 has about 90deg.
I noticed that in your 2001 are four regulators and not only three
http://dev.xdevs.com/attachments/714/k1_top1.jpg
On 2002 are only two with heatsing (+/15V) and tiny 78L05 instead four with proper heatsinks.
Could you please check the voltages on regulators? For which voltage is the last one?

TiN · « **Reply #40 on:** January 01, 2014, 09:17:33 pm »

2001 have two regulators only in that area too. Those are +15 and -15 LDOs. Two heatsinks closer to board edge are TO220 darlington's for "BS- BS+" circuitry. I dont understand that section yet, but I guess its something to do with signal conditioning or ground offset. Check my schematics scetches, i have those on it. There are also test points BS+ BS- near AD637. They read +39.8 and -39.85V on current meter I debugging.

2002 prolly have other way to handle that, I would only dream to recover schematic for 2002 and see how its done. 2002 also uses LTZ reference, not LM199 like 2K1.

TiN · « **Reply #41 on:** January 02, 2014, 07:09:55 pm »

Okay, run some more comparsions regarding ohms issue.
There are two testpoints marked +8VF and -8VF near ADC header.

There are some pulse trains when meter in resistance measurement mode, which depends on measurement speed and range.
On failing 304.x selftests board - those pulsetrains non-existant on ranges over 20kOhm.

200kohm range, high-accuracy speed, 7.5digit

-8VF PASS:

+8VF PASS:

-8VF FAIL:

+8VF FAIL:

I draw schematics of that part long ago, here it is in PDF.

BS- reads -39.8VDC, BS+ reads +39.8VDC on both failing and passing analog boards, disregarding of operation mode.
Anyone have clue what that +8VF, -8VF and NPN/PNP pair do?

plesa · « **Reply #42 on:** January 02, 2014, 09:55:51 pm »

I made some pictures of Analog board and ADC board as you requested, sorry for quality it has been taken by smartphone.
Thanks for tip about the current measurement. My unit ohm current source is totally out, instead 7.2mA it has 5.8 and on lower ranges the difference is much more bigger.
I was quite surprissed by the corrosion of the PCB trace under the sticker.

plesa · « **Reply #43 on:** January 02, 2014, 09:57:24 pm »

the rest

TiN · « **Reply #44 on:** January 03, 2014, 03:40:28 am »

Any chance to get camera pics later in future?

Sorry for being picky, but 2002 deserves better photo treatment than phone webcam

Maybe 2002 use dual slope integrating, thus implement two ADCs in bigger CPLD MAX?

Also on photo of 2002 from bbs.38hot forums there are more resistors than on yours.
I cannot tell for sure, but maybe your analog board missing some components.

I'm refering to custom arrays near red reed relay this photo, as there are four packages (R346, R339, R341?, R335?), while yours have only one (R341).
Service manual for 2002 also mentions about R335,R339,R341,R346, so that gives some hints too.

plesa · « **Reply #45 on:** January 03, 2014, 04:13:46 pm »

Quote from: TiN on January 03, 2014, 03:40:28 am

Any chance to get camera pics later in future?
Sorry for being picky, but 2002 deserves better photo treatment than phone webcam

Maybe 2002 use dual slope integrating, thus implement two ADCs in bigger CPLD MAX?

Also on photo of 2002 from bbs.38hot forums there are more resistors than on yours.
I cannot tell for sure, but maybe your analog board missing some components.

I'm refering to custom arrays near red reed relay this photo, as there are four packages (R346, R339, R341?, R335?), while yours have only one (R341).
Service manual for 2002 also mentions about R335,R339,R341,R346, so that gives some hints too.

Yep, I will make better photos next week.
R341 are glued two Caddock resistors together TF050R 9,2MOhm and TF020R 100KOhm and overall type is TF-255, which seems to be custom.
It looks to that they were solving some resistor shortages or obsolescence, I do not know. The resistor pinout does not correspond to PCB layout.

TiN · « **Reply #46 on:** January 03, 2014, 04:28:02 pm »

Great, thanks.

Yes, those CADDOCK resistor networks are custom made for Keithley, and sandwich packaging on them is OK too.
If there are missing ones - replacing them possible, but not easy. As alternative will be get bunch of lowest tempco resistors with correct value,
assembly them together into thermally connected network, test tempco and stability - and install in place of original ones. Couple weeks of precision
voodoo

That's of course if target is getting meter to fully comply with spec.

I was planing to try do so on one of 2001, to improve resistance accuracy. But that's plans for future.

plesa · « **Reply #47 on:** January 03, 2014, 04:42:14 pm »

Quote from: TiN on January 03, 2014, 04:28:02 pm

Great, thanks.

Yes, those CADDOCK resistor networks are custom made for Keithley, and sandwich packaging on them is OK too.
If there are missing ones - replacing them possible, but not easy. As alternative will be get bunch of lowest tempco resistors with correct value,
assembly them together into thermally connected network, test tempco and stability - and install in place of original ones. Couple weeks of precision
voodoo That's of course if target is getting meter to fully comply with spec.

I was planing to try do so on one of 2001, to improve resistance accuracy. But that's plans for future.

I do not think there are any resistors missing in this area.
The PCB layout is prepared for parallel combination of R341 and R346
and next to them is another parallel combination preparation for R335 and R339.

TiN · « **Reply #48 on:** January 04, 2014, 11:01:48 pm »

Seems K2000 use same CADDOCK assembly marked TF-255.

Ok, somebody was asking earlier what one should do for attempt repair 2001.
Here's starting point, as all seven units I got here had this issue more or less.
Due to age (my units were manufactured in 1992-1994), even good electrolytic
capacitors become bad and leak electrolyte on PCB. This can cause major destruction.
Problem is not always obvious, contrary to dodgy caps repairs like in PSU, as
these caps don't blow, but silently leak electrolyte under them.

Not a problem you say, PCB covered with protective soldering mask? Wrong!
Check it out (all photos clickable):

-

That's utter damage done to traces, as it got under mask and eated traces, vias and pads.
Older revision L analog board elevate this problem even more, as near C114,C115,C116,C117 there are diodes and two zeners,
which fail that curcuitry, causing PCB burns.

-

And electrolyte spreads around board, screwing everything around, which is really bad as analog board have all these high impedance signals around
which really sensetive to leakage and coupling. Cleaning this PCB is not an easy job. Not clean board often fails such self-tests like 304.6, 304.7 and 20MOhm,200MOhm,1GOhm ranges,
which require to source proper current levels in nanoAmp range.

-

This is how bad it can be

And notice, its not some cheesy Samwa/noname capacitor, it's well-respected Nichicon VZ cap.

So as a story to learn, if you have 20-year old test equipment, not only Keithley 2001, take it apart and replace all electrolyte caps just to prevent such horrors happening. Capacitors, unlike voltage references, are not getting better with age.

As a bonus today:

Frequency test on passing 2001 #4

Speedup 400% on video, to make it short

Was able to measure almost 25MHz, even while spec only states up to 15MHz.

TiN · « **Reply #49 on:** January 08, 2014, 10:26:48 am »

Redraw today digital section of ADC board. Much better now, can see what goes where

PDF,PNG, 1MB.

Sorry for colorful PDF, exporing from orcad replaces black background to white. I use colors to help defining signals in visual.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Restoration glory of Keithley 2001 DMM (Read 326749 times)

Share me