Author Topic: Keithley 2400 SourceMeter review and teardown  (Read 15489 times)

0 Members and 1 Guest are viewing this topic.

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Keithley 2400 SourceMeter review and teardown
« on: January 07, 2014, 03:43:11 PM »
Followed by my Kei2001 endevaours, i quickly found another level of instruments from this vendor.
SourceMeter series is not just a 6.5 digit DMM, but actually combines three
instruments in one. Keithley have variety of SMUs for different use requirements.

First is fast and precise 6.5 digit DMM, with ability to measure voltages, currents and resistance.
Second - precision power source, with adjustable compliance levels (limits) for both voltage and current.
Third - given it's four-quadrant power unit, it can not only source but also sink current, acting as active load.

All these functions are synced together and operate using same controls, which making use of it as easy as just normal power supply, but with much higher functional
and precision/accuracy level. These kind units are widely used in semiconductor and production testing and covered from "simple" 20W units like this 2400, to advanced
systems with multiple channels and wide ranges from picoamps to tens of amps. And even base 2400 cost significant money, I got this SMU with broken VFD,
and even then had to pay over 1000$. After salvaging display from trashed Model 2001 and replacing it - unit worked like a charm :)



Every photo below is clickable to open high-resolution, so enjoy exclusive photo coverage of Keithley Model 2400.
I was trying to find any internal photos on web before, but failed.

Exterior



It's done in common half-width 19" unit format, just like popular 2000-series DMMs, featuring same bright contrast dot-matrix VFD as Model 2001,2002,7001,7002 has.
Front panel is similar to DMMs, membrane keypad and front inputs. Due to ability of source current/voltage it have button to enable output.



Rear side have inputs as well, adding guard output and guard sense input. This comes handy to perform unique 6-wire measurements, which is useful when one trying
to source tiny voltages/currents into test device or measuring gigaohms level resistance.
Rest items on this side quite usual, GPIB port, RS232 port, mains IEC power input (have versatile 85-250VAC 50/60Hz input) and digital I/O connector.
Also have Keithley's Trigger link to team with other SourceMeter or measurement instruments for syncronised operation.



Model 2400 is fanless, so it's producing zero noise, which is nice for home use :D
Left side have massive heatsink to dissipate heat from output stage.



My unit was abused by previous owner with some sticky ugly blue liquid which is hard to clean and front panel had few cracks here and there.
Will replace front panel plastic later probably, if it's not too expensive.

So, following Dave's guidance, don't turn it on, tear it apart!

Internals

If you have 2000/2001 DMM and disassembled it, 2400 is no different. Two screws from back, two screws on bottom, and we are in.





Top board is analog voodoo section, with all sourcing/measuring magic and dozens of expensive Linear Tech, Analog devices opamps, comparators.
ADC is once again using integrating discrete solution, with digital section implemented in ALTERA MAX7160 CPLD.

Under analog board we can find switching power supply and digital brains board. Thanks to SMPS there is no bloody mains switching curcuitry like in Model 2001/2002. :)



Analog board shielded from digital and SMPS section by another PCB without any components, but only with copper polygons.
That board covers most of analog bottom side and fixed on standoffs:



Analog voodoo





Of course, there is no schematics available, just like with every recent Keithley instrument. Pity, it could be pretty educating to learn thru different solutions
used on this level of instruments.

Output stage



All bottom side of PCB on photo is output stage, which can source power with high precision. All power devices installed on aluminum heatsink
for cooling. During operation that heatsink getting warm, but not hot (+50-55C max), which is just normal for 20W capable unit.

Left corner have optical isolators for communication between digital PCB from below and analog world.

I could not spot voltage reference on this board, it's not using LM399 like Keithley DMMs, but probably use zener or integral reference. There is TP232 test point
just near serial number sticker marked as VREF tho. Its connected from LT1097 opamp output (U226)



Output stage power source, ZMA Magnetics transformer to provide -225 and 225 voltage source, driven by two transistors on right left.
All output stage section have separation and HV PCB cutouts from rest of analog stuff, nice attention to details.





ADC



Big beefy EPM7160 in PLCC84 package, just like Model 2002. It's configuration have label 2400-801A01. Clock is delivered from FOX 12.000000MHz CMOS generator nearby.
There are two 16-bit DACs, Analog Devices AD7849BR near it, and bunch of opamps.
R609 is TF245 resistor network, which was spotted before in Model 2000 DMM :)

I'll quote service manual here:

Quote
The SourceMeter unit uses a multi-slope charge balance A/D converter with a single-slope rundown. The converter is controlled by gate array U610. Commands are issued by the MPU on
the digital board through communications opto-isolators to U610, and U610 sends A/D reading data back through opto-isolators to the digital board for calibration and processing.

Rest of analog board photos:











Digital side board:





Service manual again:

Quote
The core digital circuitry uses a Motorola 68332 microcontroller running at 16.78MHz.
The memory configuration includes:

2 x two 256Kx8-bit EEPROMS
2 x 128Kx8-bit RAMs

They are used in parallel to utilize the 16-bit data bus of the MPU. The RAM is battery backed-up, providing
continued storage of data buffer information during power-down cycles.
All calibration constants and system setups are stored in a separate serial EEPROM.

So we are safe here and 2400 will not brick itself when battery go bad after dozen of years.



No much component magic here, it's similar to other Keithley products of that era.



Some LDO power for digital domain.



Another SMPS with ZMA Magnetics transformer.



Battery for RAM and two configuration jumpers. Based on silkscreen labeling, same digital board is used for both Model 2400 and Model 2410 SMUs (2410 can source up to 1100V)



Interesting find - PCB capacitors (?). Maybe it's used to read mains frequency due to AC coupling, but that's really just my guess :) Would be interesting to know
what are those square pads connected to mains input actually do :)

Mains power supply

Not much interesting here, it's KTH-7063 model power supply, probably made by some OEM for Keithley.



Voltage output is 5V 1A and +12V 2.9A.



And, front panel

Membrane keypad with rubber keys, just like good old 200x DMM series.



This is how i got it, VFD glass broken, magic vacuum is gone.



Back side.



Desoldered VFD, using solder wack to suck all solder. Thanks to little bigger pads, i could manage to free display from PCB relatively easy.
I had one of 2001's with lots of components missing, so used VFD from it. New VFD ready prior to soldering:



There is minor difference - 2001 VFD had straight row of pins, while 2400 PCB had checker formed pins. Had to do some creative forming to solder new one on 2400 PCB :)



All done, exploded view of all parts before assembly:



Assembled, first power on:



It works! I already updated firmware to C21, downloaded from Keithley website.



Last cal was 13 september 2007, which is recent, compared my 2001's 1993 and 1995 years, haha.

Some simple checks, 87V and my first 2001 measuring 1.23455V output from 2400.



2V output:



210V output:



Hope you enjoyed reading :)
If anybody have idea what to try or show more on this SourceMeter, feel free to shoot :)

I think volt-nut horde got one more follower, just ordered today bunch of Linear tech opamps, couple DAC and ADCs and first LTZ1000A...
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline chickenHeadKnob

  • Frequent Contributor
  • **
  • Posts: 440
  • Country: ca
Re: Keithley 2400 SourceMeter review and teardown
« Reply #1 on: January 07, 2014, 04:13:24 PM »
@TiN: Thank you for so thoughtfully documenting your efforts.
 I am going to read it multiple times. In many respects a teardown of an old Keithley is more relevant to us than robrenz new SMU as most of us won't buy one of those and the used market is nill for the same. One thing that disturbs me about Keithleys, they do seem to get sick and die a lot, at least much more than the equivalent HP/Agilent. That might just be a false impression as the repaired units get talked about on the EEvblog and working units don't. Looking at all the  "my 2000x  multimeter ist borken" threads I imagine the ratio of living to dead Keithleys is inverse to that of HP3440x which appear to be tanks.
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #2 on: January 07, 2014, 04:40:43 PM »
I'd like to see teardown of 2450 either, but we got what we got here :)

As for repairs, i think it's wrong impression, maybe because of age of those units.
So far i did not found on web anybody saying about faulty keithley's earlier than 2005 year or so.

I bought myself 7 2001's (5 dated 1992, 2 dated 1994) and this 2400 and two 7001's.
And all this was possible only because i bought them all broken and did not mind to take risk
of fixing them :)
2001 did have issue with voltage drive circuitry in old units, which was fixed on later ones.
None of instuments of 20 year old age safe from caps bulging and leakage either. And broken VFD in this case?
Can happen to any meter, and I still prefer VFD over anything else for usability reasons.
Hell, even space satellites fail :)

But i think positive side - due to age and abuses from owners, normal hobbyists (almost like me, lol) can buy
broken precision instruments, spend some time fixing them up, and get a handy and highly useful tool for
almost any kind of use. As everyone starting in EE, i had choice to buy some cheap bench meters and sources,
from rigol or other brands (which still do job in most of cases) or get some not so fancy but reliable and trusty
tools which i can just use, not worry about knobs and fancy menus? Well, i voted my bucks on Keithley, Tek and HP years ago already.
Did not regret a second still by today.
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline chickenHeadKnob

  • Frequent Contributor
  • **
  • Posts: 440
  • Country: ca
Re: Keithley 2400 SourceMeter review and teardown
« Reply #3 on: January 07, 2014, 05:01:49 PM »

As for repairs, i think it's wrong impression, maybe because of age of those units.
So far i did not found on web anybody saying about faulty keithley's earlier than 2005 year or so.

I bought myself 7 2001's (5 dated 1992, 2 dated 1994) and this 2400 and two 7001's.
And all this was possible only because i bought them all broken and did not mind to take risk
of fixing them :)
...rest elided


 I know,  the image of your stacked "keithley reserve cache" is what gives me the wrong idea, I am a visual animal. Even if Keithleys are less reliable, like a high maintenance girlfriend that can still be fun.

As far as the picture of those square capacitor pads purpose - I am thinking isolated zero cross detect.



 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #4 on: January 07, 2014, 05:16:30 PM »
What I meant, is fact of having bunch of dead keithleys in one hands and couple repair threads on popular forum does not mean that they are more or less reliable.
Maybe that person just a zombie addict and was looking for dead meters on purpose, for collection. I got couple dead HP gear too, 33120 and now 4263B (just arrive today, still in post office),
but it says nothing about reliability. That's it, back to topic :)

Yes, that's what i thought about zero-cross or frequency readout as well, as that's all my knowledge in analog stuff can cover for, as for today  :-//

Actually this 2400 already on my desk for a while, almost a year, use it on almost daily basis. Very handy for curcuit probing, as current/voltage levels can be set flexible.
Assembled small buzzer with NFET for one of digital outputs, and setup it to buzz on compliance FAIL condition, so turned 2400 into simple diode-check buzzer  :-DD

Only difference that current and voltage levels can be set accurately, so i can easy distinguish dead short from 1 ohm resistor, for example.
Also useful to test current sources and zeners. Just connect zener, source voltage mode, set compliance current to 10mA and you see zener voltage on screen right away.
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline robrenz

  • Super Contributor
  • ***
  • Posts: 3035
  • Country: us
  • Real Machinist, Wannabe EE
Re: Keithley 2400 SourceMeter review and teardown
« Reply #5 on: January 07, 2014, 11:32:21 PM »
Very, very, nice teardown and review  :clap: :clap:  Excellent pictures and content. :-+

I wont be tearing down my 2450 any time soon.

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 6555
  • Country: us
    • SiliconValleyGarage
Re: Keithley 2400 SourceMeter review and teardown
« Reply #6 on: January 08, 2014, 01:48:59 AM »
Drool... I always wanted a smu but they are so bloody expensive. At 1000$ this was a steal !

As for bad keithley threads.
Keithley had massive problems with ther 2001 2002 machines due to leaky capacitors eating traces on the pcb. This was solved in later revisions.

Agilent then has certain machines that all die because of leaking backup batteries . The 8112 or 8116 are notorious. So is the 34970a...

Everybody has a few black sheep

The agilent 346x supplies frequently fry theor a/d logic and analog. This is because of a spike on the 15 and -15 rail at powerup ..
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline TopLoser

  • Supporter
  • ****
  • Posts: 1341
  • Country: gb
Re: Keithley 2400 SourceMeter review and teardown
« Reply #7 on: January 08, 2014, 02:07:43 AM »
Drool... I always wanted a smu but they are so bloody expensive.

Couple of new in box ones cheap if you fancy one  ;)

Keithley 2636A dual channel sourcemeter
http://uk.rs-online.com/web/p/bench-power-supplies/7588881/
Boxed, unused, with all unopened accessories and manuals. Calibrated a few months ago.
List price £10,200+VAT, I'm asking £4,500+VAT (no VAT outside EU)

Keithley 2410
http://uk.farnell.com/keithley/2410/meter-sourcemeter-1100v/dp/2074738?Ntt=keithley+2410
Boxed, repaired, meter only. Calibrated by Keithley last month. Looks perfect, my guess it was bought and returned - cheap equipment hire.
List price £4,800+VAT, I'm asking £2,100+VAT (no VAT outside EU)

Free shipping worldwide on either item.
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 6555
  • Country: us
    • SiliconValleyGarage
Re: Keithley 2400 SourceMeter review and teardown
« Reply #8 on: January 08, 2014, 02:26:44 AM »
For 500$ maybe. A ything above that is no-go
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline Carrington

  • Super Contributor
  • ***
  • Posts: 1138
  • Country: es
Re: Keithley 2400 SourceMeter review and teardown
« Reply #9 on: January 08, 2014, 03:15:47 AM »
@ TiN: I love it, and the images are stunning.
My English can be pretty bad, so suggestions are welcome. ;)
Space Weather.
Lightning & Thunderstorms in Real Time.
 

Offline uoficowboy

  • Regular Contributor
  • *
  • Posts: 186
  • Country: us
Re: Keithley 2400 SourceMeter review and teardown
« Reply #10 on: January 09, 2014, 04:05:12 PM »
Output stage


Anybody care to venture an explanation as to how this power stage works? I've never seen anything like it!
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 13611
  • Country: za
Re: Keithley 2400 SourceMeter review and teardown
« Reply #11 on: January 09, 2014, 05:00:51 PM »
It is designed to reduce dissipation in the devices, as the output voltage rises various devices turn on and off so as to use the various supply rails so as to not dissipate high power at low voltage output. The mosfets are held off until the output is close to the supply rail for the section then they start turning on and then the diodes block current from flowing back into the various supplies.
 

Online EEVblog

  • Administrator
  • *****
  • Posts: 22816
  • Country: au
    • EEVblog
Re: Keithley 2400 SourceMeter review and teardown
« Reply #12 on: January 09, 2014, 05:12:53 PM »
Awesome work on the teardown and photos.
I so want one of these, or more precisely (pun intended), the newer model.
 

Online EEVblog

  • Administrator
  • *****
  • Posts: 22816
  • Country: au
    • EEVblog
Re: Keithley 2400 SourceMeter review and teardown
« Reply #13 on: January 09, 2014, 05:14:27 PM »
It is designed to reduce dissipation in the devices, as the output voltage rises various devices turn on and off so as to use the various supply rails so as to not dissipate high power at low voltage output. The mosfets are held off until the output is close to the supply rail for the section then they start turning on and then the diodes block current from flowing back into the various supplies.

I recall Doug Ford explaining to me once how he used something similar in his Jands audio amp designs.
 

Offline nowlan

  • Frequent Contributor
  • **
  • Posts: 602
  • Country: au
Re: Keithley 2400 SourceMeter review and teardown
« Reply #14 on: January 09, 2014, 05:29:19 PM »
Sounds like Class-G amplifier.
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #15 on: January 22, 2014, 04:02:19 PM »
Seems this story have a happy ending after all :)

Got my unit calibrated by official Tek/Kei service.

Data before calibration
Data after calibration

So this will become a reference point/source from now on on my volt-nutting explorations.  :bullshit:

Btw, if anyone want - there is an broken 2400LV (no 200V range) poped up on bay recently: eBay auction: #201021188476.
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline robrenz

  • Super Contributor
  • ***
  • Posts: 3035
  • Country: us
  • Real Machinist, Wannabe EE
Re: Keithley 2400 SourceMeter review and teardown
« Reply #16 on: January 22, 2014, 11:26:38 PM »
Nice!  :-+

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #17 on: February 05, 2014, 12:50:18 PM »
Just got my bad boy back :)
Now have some reference point for volt-nut practice.



xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #18 on: March 07, 2014, 12:37:05 AM »
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #19 on: July 09, 2015, 02:25:27 AM »
Yey, Keithley still cares about 20 year old models...

New firmware released recently, in case someone with 2400 series SMU from Japan need it.
Why Japan?

Release notes:

Version C33 fixes the issue of powering up with a low line-voltage (90 volts in Japan) for models 2420, 2425, 2430, 2440. Upgrading to this version of firmware will add a 5-second delay to power-up for all models running it.

Models affected: 2400, 2400-C, 2410, 2410-C, 2420, 2420-C, 2425, 2425-C, 2430, 2430-C, 2440, 2440-C
Symptom: The instrument will power up, but the digital circuitry will not work properly. Therefore, there is voltage on the output and no readings are accurate.
Resolution: The issue has been corrected.
Enhancements: There were no enhancements included in this release.

Remirrored it on on docsite, as usual, in case you don't want login into Keithley's site.
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline rksaripalli

  • Newbie
  • Posts: 1
  • Country: in
Re: Keithley 2400 SourceMeter review and teardown
« Reply #20 on: September 11, 2015, 03:41:08 PM »
Dear Sir/Madam,

Myself Ravi Kiran Saripalli, PhD student in India. I have read the excellent review article by Tin regarding Keithley 2400 Source meter. I have been using the source meter regularly for poling my dielectric materials, by applying a voltage, and varying the temperature externally. I had heated my sample to 380K and connected the leads to the 2400, and just when I was about to increase the voltage, the 2400 just switched off, and is not powering on again (Input volate was 0V). Could you tell me what could have gone wrong and how to fix it? Thanks a lot.
Ragrads,
rksaripalli
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #21 on: September 12, 2015, 12:04:26 AM »
Hi Ravi,

Did you check mains fuse on unit? Voltage was applied to your test sample by 2400 itself or from other source?
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline K. Kumon

  • Newbie
  • Posts: 2
Re: Keithley 2400 SourceMeter review and teardown
« Reply #22 on: November 28, 2015, 08:56:29 AM »
Why Japan?

 (I'm living in Japan. :))

By the way, I recently obtained used 2400

The source part (voltage and current) seems OK with tens milli volt offset voltage.
But measurement part doesn't work at all.

The display always shows "-----" for both voltage and current.
When sampling rate is changed with config menu causes complete deadlock.
All button reply no response except power SW.

It seems as if sampling trigger is missing.

Does somebody kindly advice me how to fix it?
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 2437
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Keithley 2400 SourceMeter review and teardown
« Reply #23 on: November 29, 2015, 09:51:46 PM »
What happens if you reset SMU to factory defaults? It should be reading values without any extra settings.
Perhaps stupid question, but does ON light come up when you enable output?

Take the cover off meter amd check if any cables or parts are damaged.
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline K. Kumon

  • Newbie
  • Posts: 2
Re: Keithley 2400 SourceMeter review and teardown
« Reply #24 on: November 30, 2015, 01:00:27 PM »
What happens if you reset SMU to factory defaults? It should be reading values without any extra settings.
Perhaps stupid question, but does ON light come up when you enable output?

Take the cover off meter amd check if any cables or parts are damaged.
Hi, TiN,

Thanks for suggestion.

I did resetting to the factory default, but nothing has changed.
As the output on/off button pushing, LED (blue) is lit and output voltage comes up

The apparent of the unit is rather good condition, and VFD is still bright.
For this reason,  I've been hesitating to open, but I'm going to open it.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf