Big differences when I short the input of my Prema 5017

[bastl_r]

Hi
Can someone explain why there are so big differences between a old goldplated MC-connector an a cheap pomona clone with a wire as a shorting bridge.
In particular, I would be interested to know why the deviation from zero is not approximately the same after turning over.

[bastl_r]

Now, when I tried again, I only had a difference of +800nV to -50nV after turning around and waiting.
Hence a guess:
Could it be that different temperatures of the test sockets, the lower one could be a tenth of a degree cooler, in combination with different materials of the plug already cause so much EMF?
It takes a relatively long time, approx. 5 minutes, for the displayed value to stabilize after each reversal.
Hmmm. I'll put on some thermal insulation and then see what happens.

[bastl_r]

Physics is fascinating, but also relentless!
With an insulating cover, I still have a difference of -1.42mV to -0.4mV.
So both values are still about 1mV shifted to the negative side.
What could be the reason for this? This is about the same deviation as in the first post.
The zero point with the MC connector is still about the same.

[Vgkid]

Yes , Thermal EMF reeks havoc when measuring low voltages(especially below 10's of UV , even more at the nV level) . Which is why input terminals to nanovolt/microvolt meters are either fully shielded , or in a shielded box. Though not always, sometimes they are on the back of the instrument.
Here is the input to a Keithley 260 nanovolt source. Top image is the door closed , bottom is open.
 

[branadic]

Physics is fascinating, but also relentless!
With an insulating cover, I still have a difference of -1.42mV to -0.4mV...

What I see on the image are -1.42 µV not mV. 

-branadic-

[Dr. Frank]

Hi
Can someone explain why there are so big differences between a old goldplated MC-connector an a cheap pomona clone with a wire as a shorting bridge.
In particular, I would be interested to know why the deviation from zero is not approximately the same after turning over.

Hello,
The mc short is quite fine, I also use them on my 3458A for shorting the input between measurements.
Usually, the e.m.f. is well below 200nV.

The Cu wire obviously is strongly oxidized, because its color is dark orange, instead of bright orange.
I can't recognize, which material this double jack has, but it's probably also not a low-e.m.f. alloy.

Therefore, you have built a nice thermocouple, which explains the big e.mf. of 1µV, as well the turnover effect.

Any small temperature difference, especially w/o shield, will give a huge e.m.f., which is not the case for properly designed shorts.

For zero calibration of the 3458A, a fresh, cleaned (polished) copper wire is required, which brings down the e.m.f. to around 100nV.   
Frank

[bastl_r]

What I see on the image are -1.42 µV not mV. 

I forgot the Alt GR key. 
I just realised this because I expected that turning the "shorting plug" would change the displayed polarity.
I wouldn't have said anything if there were only one or two digits at the end. But there are three digits.
How else do you do this in practice? Do you just ignore it?
After all, such an error sometimes adds up in a self-made measuring cable. I first noticed it on the 3457a after reversing a plug. The deviation was even greater with a second cable, which is why I tried it with one of the plugs alone.
Initially, I only used the Pomona clone and the MC short-circuit plug with which no zero (I accept 100 - 200nV as zero for the MC plug in this case) could be reached.
For use with a zero voltmeter, as discussed in the neighbouring thread, a plug like the Pomona clone would be disastrous.

Ahh, while i was writing, Dr Frank answered.
Thank you for your description. Your 200nV are  the same value ithat i would had tolereated without asking.
I will do another Check with a new cleaned copper wire later.

Regards Armin

[TUMEMBER]

Physics is fascinating, but also relentless!
With an insulating cover, I still have a difference of -1.42mV to -0.4mV...

What I see on the image are -1.42 µV not mV. 

-branadic-

I think I have visual hallucinations too.
I don't know if our colleague knows that he has a resolution of 10 nV on the display in the 300 mV range.

[bastl_r]

I think I have visual hallucinations too.

All good. Nobody was injured. 
As already written, I had not pressed the ALT GR button to the "M".
Then the "µ" is written by the German keyboard layout.

[DVM-James]

Attached a little bit to read ...

[bastl_r]

Thanks for the document. I already knew that.

However, it is not entirely clear to me how the one-sided deviation comes about.
With the short-circuit plug I built myself, I basically have a symmetrical structure in which the thermoelectric voltages should cancel each other out. Normally...
Yes, OK. This is probably due to inaccuracies in the design, so that a thermoelectric voltage occurs on one side and is not balanced on the other.
This is where the difference between Pomona and a clone becomes apparent.
I have now examined several plugs. Some are better, some are worse, but none are error-free.

[DVM-James]

Have you tried out to remove that "dark brown" copperwire and turned it arround (let the plug in the same direction)?
You may have a kind of Copper-Oxide-Rectifier (in german: "Kupferoxidulgleichrichter).
Try out a new wire with not that large diameter - 0,4 mm² only, or so. The thermal adaption my be a little bit more fast.
I have just tried out some bridges I found (selfmade ones also) and on my very old DATRON1062, Solartron7066 and
Solartron7060 the difference is at least not more than 200 to 300 nV (are 6 digit DMM only
and I didn´t wait for the regular warmup of 2 hours!)

regards
James

[bastl_r]

The experiment with the slightly oxidised copper wire was the first to test only the plug as a short-circuit element. In the meantime, I tried the wire with a freshly stripped bare 2.5mm² wire. Same result.
Previously, I had inserted the well-functioning MC shorting plug into the socket openings on the same plug and did not get a satisfactory result either. Only then did I use the (brown) copper wire to get the short-circuit bridge even closer to the measuring device.
I also made three different (test) cables. All of them showed differences in the millivolt range. The worst was with a supposedly good cable with PUR insulation and shiny silver wires as conductors.
I'll have to try a series of measurements to see which plugs fit together best.

Regards
Armin

[TUMEMBER]

The experiment with the slightly oxidised copper wire was the first to test only the plug as a short-circuit element. In the meantime, I tried the wire with a freshly stripped bare 2.5mm² wire. Same result.
Previously, I had inserted the well-functioning MC shorting plug into the socket openings on the same plug and did not get a satisfactory result either. Only then did I use the (brown) copper wire to get the short-circuit bridge even closer to the measuring device.
I also made three different (test) cables. All of them showed differences in the millivolt range. The worst was with a supposedly good cable with PUR insulation and shiny silver wires as conductors.
I'll have to try a series of measurements to see which plugs fit together best.

Regards
Armin

Sometimes it's not the wire's fault. The photo shows the effect of short-circuit measurement of pure Cu. No, it's not Seebeck tension. This is crap in the lab. calibration "supervised" by the Polish Center for Accreditation. The DMM was obviously damaged and was consciously used for calibrating thermocouples (approx. 3,000 pieces), the head of this lab. he was aware of this and deliberately hid it. The important thing is that you could get a lot of money quickly "for a commission". The calibration of this laboratory is marked with the ILAC-MRA mark EA-MLA.

[bastl_r]

Now I have more experience.
In my case, it's definitely the cables or the plugs. I would tolerate a deviation of the last two digits, but not if it is three digits.
I have now fed in 20mV from the calibrator and measured with both Premas at the same time. The right one was connected in the same way throughout. In the case of the one on the left, I took the comparative measurements with the test leads by switching the leads from top to bottom and waiting a few minutes.
I have documented various cables in the pictures below. (coming soon)

Regards Armin

[bastl_r]

More pictures