Triaxial connections for nanovoltmeter

[wn1fju]

Picked up a perfectly good Keithley 180 nanovoltmeter (the predecessor to the more modern Keithley 181).  The input connector is essentially a Cannon 16-11 which consists of two pins plus the outer thread.  Thus it basically is a triax connection.  I've managed to buy a suitable male plug, but am wondering exactly how to wire the ground/shield.  In addition to the two-wire High and Low connections, do I run the cable shield to the shell of the male plug?   Do I just leave the shield unconnected?  Do I bring out the other end in a clip for future use?  Keithley shows their pre-made cable just coming out in two alligator clips - they say nothing about the ground shield. 

[billfernandez]

I have no familiarity with that meter so I'm just guessing, but from your description of the connector (two pins inside a shield) I'd say it's not triaxial (which would require a single pin inside two concentric rings) but rather that it has a differential input accompanied by a ground shield.

I would think that the way to wire it is to connect the two pins to two wires in a twisted pair, terminating in alligator clips or whatever else you want to make connections to your device under test, and to have the two wires inside a surrounding shield (hence a shielded, two-wire cable, e.g. as used for audio microphones) that runs down the length of the cable but is not connected to the device under test.

Again, just a guess...

[TiN]

wn1fju

I don't have K180, but on my K182 connector is very similar if not same. It's Amphenol Industrial 97 series, available from Digikey and similar retailers. I had covered this in my K182-M review before. Hope this helps.

I have shield connected to cable shield (only at meter end, DUT end of shield not connected to anything), which also follows original Keithley's cable.

[wn1fju]

Thanks to both that replied.  Yes, it is more of a differential input with shield than a triax.  I will connect the shield at the meter and leave it alone at the far end. 

[TimFox]

The connectors used by Keithley with that input were custom-made with low-thermal-EMF metals.  However, I have had reasonable results with the standard mil-style connector, but that particular one is hard to find.

[wn1fju]

I bought an Amphenol MS3106E16-11P from Mouser for about $30.  I realize it isn't "thermal-grade" like the Keithley jack in the nanovoltmeter.  However, I'm not exactly running a metrology lab in my basement.  And after shorting the two holes in the jack with a short piece of wire (not using the Amphenol plug yet), nulling the meter with the zero-suppress control and then watching the display go up about 1/4 uV merely by shining a cheap flashlight on the jack from 12 inches away (it then returns back to zero when the flashlight is turned off), I know why I'm not running a metrology lab in my basement.

This stuff will drive you nuts!

[guenthert]

The connectors used by Keithley with that input were custom-made with low-thermal-EMF metals.  However, I have had reasonable results with the standard mil-style connector, but that particular one is hard to find.

    Not sure, why I thought they were pure copper connectors, but they're not.  The manual clearly states that they are gold-flashed copper-tellurium (much like e.g. the connectors on a HP419A null meter).

    I didn't get a hold on the original connectors (my instrument came, as so many others, w/o connectors -- who's hording the connectors?), but had to make-do with the Amphenol  MS3106A16-11P, which is apparently meant for military application using high currents and has connectors with a silvery surface (tin?).  One I used as a short to evaluate my Keithley 181 nanovoltmeter (about -130nV offset and ca. 20nVpp noise with filter enabled, well within spec).  A 2nd one I connected to a cable meant for microphones (shield + pair of copper strands) to which I crimped copper fork terminals (from the stash I got from TiN a while back).  When I bring those terminals in contact, I get something like 

    I'm quite happy with the noise (the excursion in the beginning is, I believe, due to range switching when starting data collection, the excursion towards the end is probably due to RFI), but is this ~1uV offset to be expected (it doesn't seem to go down, so it's not some EMF due to initial temperature difference, e.g. due to handling)?

Edit:
    It might not be due to handling, but the 1uV offset seems to be EMF after all.  If I touch the Amphenol connector (which is already quite warm -- seems to be the best outlet for excess heat from the meter, perhaps a design weakness) for some 20s, the offset races some 200nV in about the same time with a little lag.  This particular connector had solder cups at the end of the pins, crimping those didn't seem practical.  For the shortening plug I managed to bend the short sturdy copper wire so that it'll touch both pins before soldering, when soldering the microphone cable I didn't quit manage that it seems.