Keithley 2636 Teardown, Battery replacement, Test fixture

[ch_scr]

So here it is, glorious Keithley 2636. Came shipped in original Keithley box, very nice.
I took the cover off the long instrument, noticing what looks like a missing cal sticker.
And man, is the magic smoke squeezed in tight there. Tight enough to only take a sneak-peek until i have a replacement battery, at least. (Also, im have to admit, i'm a bit scared to mess up the high resistance of the output boards.)

So, without further ado, here are the peek pictures.















I tried Sourcing voltage on both Channels, held multimeter probes to the appropriate part of the triax connectors.
Voltage measured ok and and close enough to the 5 3/4 digit HMC8012 at a first glance. I also measured the "High Z >10G" input impedance of the HMC8012, which showed a plausible ~15Gohm on the Keithley.
I also noticed at first turn on the described "Battery low, Userscript lost" Errors, but not on second turn on. Possibly just a flat (or maybe weak) battery? Only time will tell.

I saw the mentioned Trompeter "Wrench Crimp" connectors, with .200" or .250" cable inlet diameter. The only low noise cable I found cheap-ish and in short length (min order is 10m @ 4,5€ per m) "G_02330_HT" from "Huber & Suhner" which looks good, but. Its 0,2087 outer diameter (a hair over 2 thenths of a millimeter more), so i would have to drill out the "clamp nut" a bit to fit it trough. My only fear is the "cone" might slip trough the "clamp nut" then, can anyone with experience chime in, if there is enough overlap between the two to do this?

[alm]

Nice score! Though not that cheap .

No experience with that particular cable/connector combination. I would indeed be very careful touching anything in the high impedance circuit. A finger print is low resistance compared to the TOhm input resistance this unit is capable off. The usual recommendation after any servicing of the high impedance parts is a complete wash in an alcohol (methanol or isopropanol), so I would avoid touching it.

I guess you could compare the two channels if you had the correct cables. Keep in mind that the inner shield is likely a driven guard, so be careful to not connect the guard from two channels together (i.e. no straight triaxial cable between the channel outputs).

[HighVoltage]

You received it fast.

Nice to see the first pictures of the inside.
I would probably wear gloves for protection so that you don't leave any fingerprints.

Can you measure the battery voltage easily?

[alm]

Does it pass the self tests? Or do those fail on low battery?

I sometimes use the HP 16058A fixture for SMU measurements. It was designed for the HP 4145A semiconductor parameter analyzer (basically a combination of multiple SMUs with some software to do semiconductor tests), which does not go down to the fA level like this SMU. I have not tested the leakage. Unlike the Keithley 8002 (?) fixture, the connections inside the box are not guarded, but it is shielded and probably close to light tight (I have not verified this). I think I got one for a little over $50 with jumper leads (they are nothing special, but there are two sizes of pin connector in there), and another $50 to get most of the inserts.

The thing has four triax inputs and two BNC inputs, plus some signals on the D-sub connector. Internally it breaks them all out in pin jacks. You can use jumpers to connect to the insert with the socket, or just air-wire something. Even four of those triax sockets might run you more than what I paid.

[ch_scr]

@HighVoltage: The Battery is insulated nicely, and the connector is unreachable. Not sure i want to mess with the wire insulation while it's in there.
@alm: That HP Test fixture is nice Inspiration, although i (foolishly?) believe i don't need all that functionality it provides. I had a bigger diecast hammond case in mind with 8 triax jacks, and a smaller, el. insulated hammond case inside, connected to the lo guard of one of the channels. Inside shall be 4 small Kelvin clips (835-502001 ~7€ each @ Mouser) connected to force and sense accordingly. The DUT could be connected, mech. supported by the wires only, hanging in free air. Only thing to mind should be where on the DUT the inner case guard potential makes sense (also when measuring a resistor or another 1 ch job). Priciest thing would be the triax jacks. Should be a lot cleaner than the usual aluminium food tray method, but sure falls short when trying to measure OPA input current and the like. Back to the drawing board, and having a closer look on that HP thing.

[alm]

As I stated, I do not claim that the HP solution is ideal down to the fA. The Keithley 800x fixtures are probably better. I seem to recall they maintain the guard until right at the DUT. The argument in favor of the HP is if you can get it cheaply, and that it works good enough. The Keithley ones always seem to go for well over $200. You could probably design something better for low current measurements. A double shielded box: one layer of ground for safety and one layer of guard for leakage would be optimal. Having extra connections to feed in e.g. an extra power supply or to connect a DMM or scope can be a useful feature. You might want to look at this video from forum member robrenz about how a professional machinist might solve the problem:


I used the 16058A as a breakout box for triax connectors before I had sufficient cables with alligator clips and triax-to-BNC adapter with guard removed or shield removed so I could connect to other instruments (e.g. DMM) easily. I would just connect the triax cable on the back, connect a jumper to the corresponding jacks and clip some banana or BNC pigtails to the jumper.