Keithley 2400 SourceMeter review and teardown

[plesa]

I will take picture with infrared camera,  this was helpful during my repair.
If you cannot find bad transistor, desolder transistors and test them within operating conditions.
Firmware recommended is C32, C33 has delay to work properly at 90V in mains I needs to downgrade, the startup delay was annoying.

BTW is anyone interested in purchasing new VFD DD-51 for Keithley gear, send me PM.

[z01z]

Yes, that startup delay is very annoying, I also went back to C32 because of that. It should be made selectable somehow.

[plesa]

Christmass for Keithley gear 

[plesa]

Time to replace few VFD on various Keithley gear. New VFD are little bit smaller with contacts in one line.
VFD holds on PCB 3M VHB double side tape ( reel 33m/150 EUR).
Contacts are soft and easy to bend. By tweezers I select pins row and later align all in two rows by PCB.
After that started from one side and with tweezers I put each pin into hole by tweezers.
After two hours is repair of K2400 finished ( forgot to connect analog board  - Over temperature )

[ivonenand]

Hi guys,
Does anyone know, how the output current sense is implemented in this SMU (or any other for that matter)? The thing is, you need extremely high CMRR, since the output is anywhere in the +-200V range. Second, the SMU offers better than 1uA resolution, so there must be little or no leakage on the amplifier connected to the output shunt. You cannot do that with a standard differential amplifier.

Any ideas on how the amplifier connected to the output shunt is implemented?

Regards,
Ivo

[VintageNut]

Hi guys,
Does anyone know, how the output current sense is implemented in this SMU (or any other for that matter)? The thing is, you need extremely high CMRR, since the output is anywhere in the +-200V range. Second, the SMU offers better than 1uA resolution, so there must be little or no leakage on the amplifier connected to the output shunt. You cannot do that with a standard differential amplifier.

Any ideas on how the amplifier connected to the output shunt is implemented?

Regards,
Ivo

I do not have a schematic for the 2400. My assumption is that the 2400 is not far from the 238 as far as architecture. The 238 schematic exists as a printed document.

[plesa]

Hi guys,
Does anyone know, how the output current sense is implemented in this SMU (or any other for that matter)? The thing is, you need extremely high CMRR, since the output is anywhere in the +-200V range. Second, the SMU offers better than 1uA resolution, so there must be little or no leakage on the amplifier connected to the output shunt. You cannot do that with a standard differential amplifier.

Any ideas on how the amplifier connected to the output shunt is implemented?

Regards,
Ivo

Check this thread, where is link for schematic of one of Keithley SMU.
https://www.eevblog.com/forum/testgear/keitley-236-teardown-and-review/
2400 is quite similar ( at least output stage, which use the same transistors.

[plesa]

Unbelievable repair 

[julian1]

I did a repair on a 2420 which had burnt power mosfets driving a secondary DC-DC SMPS. The interesting difference with the 2400, is that the power-board and digital board are combined. In contrast, the 2400 uses a third-party non-Keithley power board.

The same pcb is used for the 2430 as well (the silkscreen has a tickbox)- and possibly other high-current output models. Here's a pic of the board,

[z01z]

Unbelievable repair 

It is
Then bumped into a video on youtube from , about a secret menu in K2400.

[TheSteve]

Used TiN's script to calibrate a 2400 from a 3458A - worked fine but I did make one change so it didn't turn my screen background to yellow during the DCV calibration.
Changed sys.stdout.write ("\033[0;43m*") to sys.stdout.write ("\033[0;40m*") to maintain a black background.


Thank you TiN for the script!

[TiN]

Proper would be \033[0;49m ANSI escape sequence, this means default background color.

[eeviking]

Hi
I just scored a Keithley 2400 with a vague "powers off randomly" fault description. It was pulled from a test rack, so was probably running hot for years.
From the description I would suspect bad caps in the psu. Haven't seen the problem yet, but only did some quick tests.
Top board looks clean, haven't torn it down to look at the bottom psu boards yet. It's from around 2001 so a year younger than TiN's.
Question is: Should I just replace all caps?
Anything to look out for? What will be lost if the battery is removed?

[Jens01]

As an addition to the information already shared here on the forum, here my $0.20.

Bought an defect 2400, bought as broken/for parts. Unit came from a automated test set-up so visually in good condition. Symptoms:
- -240V on banana jacks when output was disabled
- When enabled, the current reading was always off by 9.8mA. When sourcing 9.8mA, the current measured with the output shorted with a DMM in current range was zero. The leaking current was independent of the selected current range.
- The voltage reading was always wrong. Also, when sourcing 50V, the output went up to +240V.

There are quitte some usable test pads on the board. My 2400 differs quite a bit from the photo's from TiN, but here are some hints based on the testpoint labels on TiNs 2400:
- TP201 'NVDAC' and TP203 'NIDAC' referenced against FCOM: respectively DAC voltage and current setpoint. Range dependent, but should linear change when giving a new setpoint.
- TP214 ' EAMP' referenced against FCOM: difference of voltage between pin 2 and 3 of U500. Zero volt means that the setpoint voltage matches the feedback voltage, indicating that the amplifier stage is probably working OK. When reading other than zero volt: output stage clips or is defect.
- TP221 '+si' referenced to TP222 '-si': feedback current. Range dependent, should linear rise with measured output current.
- TP227 '+sv' referenced to TP228 '-sv' feedback voltage. Rang dependent, should linear rise with measured output voltage.

I started with reverse-engineering the output stage, as they seem to fail very often and reverse engineering is easily done as the components are mostly on the top side. The list of designators in K2400 service manual did not match the numbers on the PCB, but it did match with the designator list in the K2410 service manual. Obviously the high-voltage part will differ, but the < 30V part seemed to match well.

Eventually i found out that my unit had a fully working output stage. if i knew about that EAMP testpad before i started..
The setpoint dacs seemed to work so the feedback path was the next step to debug. The current feedback path is located near the blue resistor on the front side of the PCB and the voltage feedback path is located next to the black relay on the right side.

 


This component, Q233 on TiNs photo, became very hot when driving the output to a high voltage. As the voltage reading was always wrong, i eventually found out that op-amp U219 (OPA124) had a large voltage difference between the inverting and non-inverting input. As this op-amp is a simple high-impedance voltage follower, i tested it out of circuit to make sure the output was not externally clamped. With success, the output was stuck at -15V

Replaced the opamp and now she's fine again.

[Kjelt]

Bought an defect 2400, bought as broken/for parts. Unit came from a automated test set-up so visually in good condition.

Troostwijk solar factory auction ? Missed those, the prices skyrocketed 

[Jens01]

Bought an defect 2400, bought as broken/for parts. Unit came from a automated test set-up so visually in good condition.

Troostwijk solar factory auction ? Missed those, the prices skyrocketed 

No, Ebay. The name 'Keithley' was misspelled, so i was just lucky to be at the right time at the right place.

[kawal]

All,
I got a 2400 (199X vintage  shipped with C06 firmware and currently running C11) with output of -240V with or without the output on.  I have looked for ideas of what the issue could be on this forum but could not find the same exact issue replicated.
Reading the forum did help to  get a general understanding of the possible issues.  too bad there is no schematic for these available as it would make fixing these a lot easier.

So far what i have checked :
Output stage transistors  - all good.  Pulled the Bipolar Darlington power transistors to check.
Power supply 12V ok  5V  oK  +5VF, +15VF, -15VF, +30VF, -30VF all good
 

NIDAC, NVDAC  signals adjust with output adjust.
EAMP is -7V which is maybe ok for the output of -240V ?

The board must be an early revision as the AD847 power supply clamping zener diodes are soldered direct to pins of the op amp and there is no room for them on the board.

The positive and negative supply of the AD847  was very low so i pulled it to check it.
Checked Q528, Q529 and 100 ohm resistors in the circuit to make sure these are good. ad they are.

With the AD847JN  out it seems this might be the issue. While power consumption of the amp is normal (5.5ma)  The amp will not work as  a normal op amp.  The output is not following the input and there is a 1.2V difference between the input + and input -. Maybe input stage is broken ( when setup at unity gain) To test i put the amp in a socket with bypass caps on socket. Amp configured in voltage follower mode with gain on +1. Supply was +/-10V and applied zero volts to positive input.  Output was 1.2V and same difference on - Input pin.
Still  not sure why the voltage was low on the power pins for the AD847.  Waiting for replacement chip

[openloop]

Have you checked output stage voltages?
They are separate from analog board sources (the ones you listed).
Output stage power comes through T500 transformer and two bridge rectifiers: CR500-CR503, CR504-CR507.
Their ground is at TP501

U500 (AD847) is bootstrapped so its power must be straddling positive output terminal. Thus, when talking about a voltage, you should mention to what "ground" it's relative to.

Additionally, while testing you probably can use any op-amp you've got - U500's pins 1,8 and 5(?) are not connected.

[kawal]

Yes I did check the output stage voltages and they seems fine correct.
In regards to Op amp voltages I checked versus FCOM. 

I am ready to put the output stage back in without driver amp. Without the driver the output should be putting out 0V since its a B class + H class hybrid.

I did find a little difference in the schematic and i will post here as it could help someone in the future.  Basically the op amp should be working at +/-5.6V power supply  and transients no bigger that +/-9V if the zener diods are right.  Still waiting on proper amp but will try with AD845 with i got at local shop.
This is not the right amp as its a Jfet but should be ok for testing.


Ok  Without Op Amp the output is 10V  and 0V measured to FCOM   which is strange  with op amp its -240V so same as before .

Power is +/-250V and +/- 42V
Op amp power rails without op amp  installed is +/- 5.6v
With op amp installed and -240 on output its +4.2V -2V to FCOM

Must be feedback loop issue ?
 

[openloop]

Without Op Amp the output is 10V  and 0V measured to FCOM

FCOM is the positive output. So it's always 0V.

With op amp installed and -240 on output its +4.2V -2V to FCOM

-2V is explainable: there are simply no more voltage headroom on the negative side when output is pulled all the way down.

+4.2v is suspect: it should be +5.5 or some such. It is set by VR500 6v zener and Q528 follower. That means that there might be not enough voltage at Q518 collector.

Just for clarity sake (when SMU's output is at -240v):
What are the voltages (FCOM relative) on all pins (2,3,4,6,7) of U500?
(I'm kind of confused by the way you listed them in different places)

What are the voltages on collectors and emitters of Q518, Q521?

Are any of transistors in the output stage warm/hot?

What is the voltage drop across R562?  And R563 too? (just to see what kind of current we're dealing with)

[kawal]

Openloop thanks for helping :

OP AMP voltages :
Pin 2 0V = FCOM
pin 3 -2.4V Removed the diodes from the schematic  - there is no input diodes
pin 4 -1.95V  Negative supply
pin 6 -1.2V same as output transistor bases
pin 7 4.2V  Positive supply 
R562 0.1mV . 2mA OK
R563 4.1mV !!!! 8.2mA Hmmm  So where is the current going ? this is with output disabled
 BTW why is the -240V on the output when the output is disabled ? there should be a relay for the output correct ?

Meter displays ---.--- when output is disabled and -215.something V  when output enabled. on high range and -21.something V in low rage

[openloop]

pin 3 -2.4V

Damn, I thought it was +

It changes things. Apparently output stage does exactly what it's told to.
It's control (a.k.a. "maindrive") that, ahem, driving it down! 

OK, I need a minute to think about it. 

[kawal]

Open loop
 Your comment about FCOM being the output is helpful

I measure from FCOM to power amp stage output and that is 0V
I measured from FCOM to output positive terminal that is -5V HMMMMM something not right i would think

 

[kawal]

Ok things just got worse .

U660 AD7849 is hot and not changing the current setting.
U661 is warm but not hot.

NIDAC is at 40mv

[openloop]

I measured from FCOM to output positive terminal that is -5V

God knows what kind of protective junk they have there. Smells like a rabbit hole.

Much more interesting is to see why maindrive is bad.

Can you check the voltage at TP213? It's straight down from U500, closer to the middle of the board.
That's combined control point driven by +/- Voltage control and +/- Current control.
They are merged together through 4 diodes: Q209, Q211, Q213, Q215. They kinda form a rectangle with TP213 in the center.

Can you measure voltages on them too to see which one is conducting?