EEVblog #718 - Keithley 2400 SMU Teardown

[EEVblog]

What's inside the Keithley 2400 Source Measure Unit?

Service Manual: http://www.keithley.com/knowledgecenter/knowledgecenter_pdf/2400_902_01D.pdf

Datasheets:
http://www.analog.com/media/en/technical-documentation/data-sheets/AD7849.pdf
AD847 opamp: http://www.analog.com/media/en/technical-documentation/data-sheets/AD847.pdf
Photomos Relay: http://www.alldatasheet.com/datasheet-pdf/pdf/87873/ETC/AQV214.html
Max7000 PLD http://www.altera.com/literature/ds/m7000.pdf
LT1097 http://www.linear.com/product/LT1097
LT1007 http://www.linear.com/product/LT1007X
Differential Amp AMP03 http://www.analog.com/media/en/technical-documentation/data-sheets/AMP03.pdf
AD712 http://www.analog.com/media/en/technical-documentation/data-sheets/AD712.pdf

[AdShea]

Seems like there's an error in that simplified schematic.  On the negative side there's a resistor between the high voltage transistor stack and the low voltage stack.  On the positive side there's a cap.  I don't think there's a DC path from +250V to the output in that schematic for the signal.

[gardner]

Below and to the right of the Altera CPLD (looking at it with markings right way around) are a collection of parts silk-screened VR600, VR602, VR603.  The diode one could be a low tempco zener reference or something like that.  VRxxx could be a matched set of hand-selected components.  I think the voltage reference in my Keithley 195A is built that way.

[Halfdead]

Below and to the right of the Altera CPLD (looking at it with markings right way around) are a collection of parts silk-screened VR600, VR602, VR603.  The diode one could be a low tempco zener reference or something like that.  VRxxx could be a matched set of hand-selected components.  I think the voltage reference in my Keithley 195A is built that way.

I was going to point this out as well.

 

[eV1Te]

I'm guessing the reason they use DIP packages for some amps is so that they can have guard traces going around the input legs.

The SMD packages would have too small pitch to enable guarding?

[EEVblog]

I'm guessing the reason they use DIP packages for some amps is so that they can have guard traces going around the input legs.
The SMD packages would have too small pitch to enable guarding?

SO SMD packages are easy to add guard traces.

[JohnnyBerg]

Nice video, however, I'd love to see this puppy in action. A short 2 min. demo before the teardown?

[flabbergasted]
IMHO its very brave to say on camera that this device perhaps uses the voltage rail as a reference!
[/flabbergasted]

[Dr. Frank]

In their BOM and schematics, Keithley names general purpose diodes as CR1, CR2,.., whereas diodes with higher requirements were named VR1, VR2,..

In the Keithley 199, they use an 1N4579 as the reference, it's a 6.4V / 5ppm/K compensated zener type, see schematic attached, it's VR2.
They specify for this 5 1/2digit DMM: 30ppm for 24h 23 +/-1°C and 60ppm for 90d / 23+/-5°C, which fits well into the zener spec.

Therefore, also in the 2400 SMU they may use a compensated zener reference, presumably VR600 in conjunction with  VR 602, 603.. but I 'm missing resistors nearby in that part of the circuit.. And there is no other leaded glass diode around the A/D, therefore VR600 for sure is something like an 1N4579.

Maybe, if Dave would measure TP232 (Vref), and then the voltage across these VR diodes, maybe this would match.

In the end, the stability / accuracy specifications of the 2400 SMU are nearly an order of magnitude worse in DMM mode, than the DMM 199, for example.. Therefore, the usage of a simple zener reference is very probable.

Frank


PS: I'm so blind, today! It's all in the Keithley SMU 2400 manual!

page 6-9: replaceable parts: VR600 = 1N4579, ZD 6.4V
page 4-11: analog circuitry check: TP232 = 6.4V +/-6V  (? ? ?)

So it's exactly, as assumed, and Keithley is very conservative with their components, over the years.

[TiN]

I can measure as well.

There is also secret menu, as was mentioned  in my 2001 thread.
To see it hold "Arrow up (round button)" and SWEEP together when power on, and navigate into MENU>GENERAL>SECRET

There will be SECRET MENU with next items:

MODEL# = allow you to change Model ID between 2400,2410,2420,xxxx(I think this item takes based on HW),2430,2426,2440,2425.
FP-BURN-IN = Just lits all segments on VFD
ANALOG-REV = Allow to set revision ID (mine was H)
DIGITAL-REV = Allow to set revision ID (mine was H)
CONTACT = Allow to enable/disable contact check option

Caution, if you change MODEL# - calibration will be lost. Yes, i did it 

[JackOfVA]

I was going to complain about the battery as well. Soldered in batteries buried in annoying places are just evil.

Agreed 100%. The service manual describes the battery as held in a clip, not soldered, so there must have been a running change, albeit not necessarily for the better.

From Page 4-12:

To replace the battery, first locate its holder. Use a small, non-metallic tool to lift the battery so
that it can be slid out from under the retainer spring clip.
The new battery should be reinstalled with the “+” terminal facing up. Lift up on the retaining
clip and place the edge of the battery under the clip. Slide the battery full into the holder.

At least the battery backed up memory does not hold calibration constants - those are in non-volatile memory.

[mmarks3141]

I see TiN has already posted details about the secret menu and found the good stuff.  I'll just add for the benefit of those who aren't reading the Keithley 2001 restoration forum that almost all Keithley products from that era with the same form factor and similar keypad layouts will have a secret menu usually accessed in the same way (hold 2 keys at power-up; go into Menu -> General and then navigate to the far right).  Some were Easter eggs for fun such as showing the design team; others were for real testing or helping the analog guys with their development and debugging.

I was the firmware engineer at Keithley responsible for most of the front panel menus on those products between 1991 and 2004.
--
Mike

[cypherpunks]

I'm trying to follow it, but there are three different grounds, four "VF" signals (F=force?  feedback?), and nothing obviously an output.

Also the "output transistors Q518 and Q521" don't seem to have their bases connected to anything except each other.

The signal connections to the unlabeled buffer are interesting, but I know two ways to make that work, and I'm not sure which is being used:

• An amplifier with limited voltage range can be bootstrapped by tracking power supplies to handle a larger voltage swing without losing precision, or,
• An op-amp's supply pins can be used as extra outputs which connect to separate high-side and low-side circuits.  Usually the "output" pin is grounded, and the two supply currents driver external transistors.

Dave said he didn't want to go into the details, but this sort of circuit design is far more interesting than lists of parts in use.  Is someone else willing?

[uwezi]

Funny enough I have one of these 2400 standing on the shelf here at work with a problem  . Inspired by the teardown I will try to locate the problem again (tried about a year ago or so) in the coming weeks .

I just went and had a look - ours is an even earlier revision than the one Dave opened. I'll take some pictures, because there some interesting tweaks inside.

And actually at least in hours the battery is not soooo badly placed: it is mounted on top of the circuit board and could be unsoldered from the side without actually taking out the upper circuit board. Given that our particular 2400 appears to be from 1996 (revision of the circuit board and date codes) I guess a battery change cannot be the worst idea...

[uwezi]

Yesterday I found a 2400 in our dumpster. It was completely disassembled with the VFD board lying loosely around - but luckily the vacuum hadn't escaped from it. As we say here "don't take it apart, turn it on" I put everything together and switched it on. It definitely looked inside the plastic back were all the parts were, that someone had tried and failed to repair it.

But, yes, it works. I haven't check any accuracy yet, but it seems to be ok with regard to its functionality. In our group we have 2 working ones and other equipment which I can use for more accurate checks and a home-made calibration which will be good enough for me.

I cannot find a date code on the board, like on Dave's, but the newest date code on any chip I found is week 12 1996. I don't know when Keithley started the series. On the board there are two botches and I wonder if someone else has seen anything like this or if these were traces of the attempted repair.


Also the service manual which I found oat electrotanya doesn't contain a full schematic of the unit. Is there any schematic available out there somewhere?