Binding posts/banana sockets cleaning on DMMs/metrology lab gear?

[TiN]

We all know importance of clean copper-copper connection to reduce TEMFs and contact resistance, especially doing sensitive low-voltage measurements, but how you guys keep contacts clean?
I saw different methods, but rarely see anyone shared techniques they used to keep connector clean.

My main question is regarding copper non-plated terminals, as they oxidize really quick.
It's much less of a problem with gold-plated terminals, but still could be an issue there.

Here's some real data how badly this can screw measurements:



Setup : 3458A sampling on VHP101 95K 0.005% resistor with PMO, specified at 0.3ppm/K in window +20 to +40C.
Brown bold line on graph is temperature from BME280 sensor, blue stagged line is TEMP?
Green bold line - ppm deviation (scale on most left) from median
Green dots - samples.

As you can see resistance go crazy with span almost 50 ppms! So before I doubt my 3458A proper operation, let's hook same resistor, using same connection wires to gold-plated bananas on K2002:



Much less deviation, even though reading noise is way larger on Keithley 2002 (you can see sample dots all over the place). But what is more important - averaged resistance reading closely matches temperature graph, as it should be. In both cases connection done is using copper wire in 4-wire mode at DMM end and TRX connectors on resistor box end.

I cleaned connection wire and reconnected it back to 3458 posts to get 2nd set of data. Will see how it goes tomorrow.

So I'd like to hear more about ways to keep sensitive gear terminals in good order, if you have something to share. Non-destructive methods are welcome. 

[ap]

I use Deoxit on my 34420A from time to time and nothing on my 3458As, never had an issue though. Not sure what you see is an oxid problem, we will see.

[quarks]

a few years ago I made this list


and found a good app note from Fluke http://support.fluke.com/find-sales/download/asset/2548277_6001_eng_a_w.pdf
with


and found an interesting article at a German university (unfortunately it is written in German)
http://www.uni-magdeburg.de/exph/messtechnik1/Parameter_Thermoelemente.pdf
In there is a nice table with "Seebeck-Koeffizienten"

[ap]

Nice list indeed. The CuBe value seems high though (even higher than brass) since there is only a few percent Be included. Was this measured or is there a reference somewhere? I have not seen that with my K2002 (no binding posts, so i sometimes use selfmade CuBe cables besides Pomona cables), with brass I have seen highr values.

[wiss]

So you just subtract the Kupfer 6.5 from the values and you have the copper coefficients

[wiss]

And a mix of Silver and Cadmium will most likely not have a coefficient of 0.5 against Copper, it is far more complicated than that,
 don't ask me since I only know it as a fact

[quarks]

Nice list indeed. The CuBe value seems high though (even higher than brass) since there is only a few percent Be included. Was this measured or is there a reference somewhere? I have not seen that with my K2002 (no binding posts, so i sometimes use selfmade CuBe cables besides Pomona cables), with brass I have seen highr values.

In my list I have collected and compared data from many sources.

I.e. CuBe Data is in Agilent Application Note AN 1389-1
"Digital Multimeter Measurement Errors Series System Cabling Errors and DC Voltage Measurement Errors in Digital Multimeters Application Note AN 1389-1" (Figure 2 see att.)

[alanambrose]

Ummm, maybe we should run some real world checks on some actual set-ups? What's the best way of doing that?

Alan

[TiN]

Stable nanovoltmeter and spot welder with set of metals?

[zlymex]

That's right, the real world test is the best, on meter end and on leads or DUT end.
I actually made these tests 8 years ago(when I got my 34420A) and here is roughly what I have done:
At meter end, use various wires lugs and plugs, different ways(five-way binding posts) connecting my 3458A, try to alter the environment around the binding post by fan, by touching or holding my hand or even by blowing hot air.
Also tried similar thing at the other end but use my 34420A nano-voltmeter.
The thermal EMF results I found out is less or much less than what have said in the text book, largely owning to the temperature difference is much less than 1 degree C.

As a result, here is what I do since:
1. bought a lot of binding post made of brass and iron together with Pomona 3770
2. never use my two set of fluke 5440A-7002 again
3. use twist pair(of Cat 5, 6) the most(but apply stronger than average force)
4. wind shield the binding posts by tissue for voltages <=1V

[Kleinstein]

Something like a gold coating does not have a large effect on the thermal EMF, as there will not be a large temperature difference across the thin coating. The same is true with copper oxide. The thermal EMF data for the semiconductor materials like the copper oxide, tellurium and silicon depend very much on the doping / purity. They values can even change sign: p-doped is usually positive and n doped negative.

For the wire material one has to take some extra care: some materials can change there thermal emf from mechanical deformation. So just bending a wire a few times can cause is thermal EMF if there is a temperature gradient in just that place.