Tellurium Copper is usually not used for a device's terminal posts but used as the lead wire due, as you say, for the malleability to crimp well and flexibility. The point I was making is to use the same interconnect test lead material throughout as the DUT terminal posts. The 3458A and the 732A both use Beryllium Copper alloy making that type interconnect lug or plug the best choice to minimize the dissimilar metal EMF or Seebeck voltage. The 34420A uses pure copper rather than an alloy terminal and for the same reason, minimal Seebeck voltage is realized with a pure copper interconnect. Any type of Silver or Gold plating on the terminal or wire will introduce the undesirable  dissimilar metal properties, both at the plating junction and at the plating metal to DUT terminal.

The NI website had this chart that quantifies the Seebeck voltage best:
"When two, dissimilar metals are joined a voltage is created. This voltage is known as the thermal electromotive force (EMF) or the Seebeck voltage. The Seebeck voltage is dependent on the temperature of the junction and the composition of the metals joined. The specific metal-to-metal junctions result in specific temperature coefficients (µV/°C), also known as Seebeck coefficients. The following table lists the most common metals and their respective Seebeck coefficients."

Junction µV/°C
Copper-Copper <0.3
Copper-Gold 0.5
Copper-Silver 0.5
Copper-Brass 3
Copper-Nickel 10
Copper-Lead-Tin Solder 1-3
Copper-Aluminum 5
Copper-Kovar 40
Copper-Copper Oxide >500

Granted, Gold and Silver are the next best choice, and is certainly why they are satisfactory, but using either warrants a more critical temperature gradient issue. If your measurements were satisfactorily convincing, than you probably had no appreciable junction temperature differences.

Don Johnson

-----Original Message-----
From: volt-nuts [mailto:volt-nuts-bounces@febo.com] On Behalf Of acbern@gmx.de
Sent: Monday, August 25, 2014 5:37 AM
To: Discussion of precise voltage measurement
Subject: Re: [volt-nuts] 732A and Prologix received

I have used the pomona spades, mainly to interface the low emf pomona banana cables to binding posts. I have stopped this, reasons being, they are large and worse, that the pomona spring loaded insulation tube that covers the banana plug conductor uses such a strong spring that slowly the plug works its way out of the spade. this btw also happend to me when I used the pomona low emf binding posts together with the pomona low emf banana cables. overall I m not happy with these.
so, due to lack of options, I changed to self-made twisted shielded pair of high grade teflon/kapton silver plated copper cable with gold plated copper spades (crimped). I use them not only with the 3458a but also with nanovolt meters. these have higher resolution and accuracy in low level measurements than the 3458a. emf voltages were never an issue with these cables if properly used. I have posted some results doing 34420a stabilty measurements on the pmel forum, and the results are convincing (purpose was actually not to test the cables but the stability of the 34420a, but the emf issue is a part of this of course. we use the 34420a to do low voltage precision measurements on thermal converters where the full scale signal sometimes is 1mV).
that btw also relates to don's statements below, I do not concurr with his comments about copper telurium as cable and spade material and so on. this material, as stated here many times, is used because it is machinable, for copper spades one would not use it. the 34420a factory cable uses copper cable and copper spades, not telurium-copper. if there was a problem, it would be worse with the 34420a than with the 3458a because of its low level ranges. and again, I have not seen any problems in a chain of (output to input):
1.copper-tellurium post from e.g. 8 digit calibrator 2.crimped copper spade, gold plated 3.silver plated tsp copper cable 4a.crimped copper spade to copper-tellurium post or 4b.soldered copper connector(34420) my consistent results over more than a year using them.

volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.

well, your last point is the issue, how can you have a temperature difference within a few microns of material in said connections. theory is one thing, but in reality it does not happen due to the givens of the setup.
therefore in practice it is irrelevant if the wire is silver or gold plated or pure copper. otherwise the gold plated spades and tellurium copper posts from pomona and others would be nonsense. and other than the mysterious fluke wire I have never seen a tellurium-copper wire from any wire manufacturer.

On 8/25/2014 11:02 AM, Don@True-Cal wrote:

Why?

Any Seebeck effect is immediately offset in the opposite direction,
since both junctions are (under normal conditions) at essentially the
same temperature (e.g. there's a copper-gold thermocouple, the minimal
thermal resistance of a micron of gold on the contact(s), then a
gold-copper thermocouple). It seems to me that the improved consistency
of the contact outweighs any loss from the thermocouples.

A more typical contact would be copper-nickel plate-gold plate, but the
concept is the same. Unless there is heat flowing through the entire
assembly so one thermocouple is warmer than the offsetting one (e.g.
shortly after plugging in a banana plug warmed by body heat), they
simply cancel.

Even if connecting gold plated to nickel plated contacts, it works out
the same - a copper-nickel-gold-nickel-copper connection is completely
offset. It's when the offsetting thermocouples occur across a
temperature gradient that you have problems.

--
Mike

Hi Adrian,

Although I have not used the Pomona spade lugs or a nano-voltmeter, my
experience is consistent with yours otherwise.

Anyone who has calibrated (CAL 0 anyway) a 3458A has enough information
to deduce that lowest input short voltage is going to be a copper wire
since the meter is set to zero volts during calibration using a heavy
gauge (14 to 16) copper wire to short the top four input terminals. This
is very convenient since it is so easy to duplicate in the field and
makes simple inexpensive test leads best for high precision
measurements. In order for the 3458A to make full accuracy measurements
(8 digit) NPLC must be set to 1000 (according to the User's Guide). If
you use a lower NPLC value there is a table in the User's Guide that can
be used to determine how accurate your measurements are going to be.

Just for fun I ran several shorts for NPLC 1000 on my 3458A. It has been
about a year since I clid my last CAL 0 so it was going to be
interesting at least for me. The first shunt was my test "U' shaped
shunt that I used for my last CAL 0. Note: STP = Shielded Twisted Pair.

Calibration shunt                        -0.00021mv +/- 10
pV            Equilibration time 5 minutes. 14 gauge per Calibration
Manual.
"U" heavy wire                            -0.00021mv +/-
10pV              Equil. time 5 minutes. Used the through holes in the
Input banana posts only.
Copper wire                                -0.00019mv +/-
10pV              NAPA PVC covered automobile wire at same contact
points as CAL shunt
Standard Ground Plate              -0.00040mv +/- 10pV Equil. time 5
minutes. Gold plated ground plate from my Datron 4910
Copper wire                                -0.00019mv +/-
10Pv              Equil. time 2 seconds. Used the banana through holes.
STP 2 meter test lead                -0.00021mv +/- 10pV
Equil. time 2 seconds. M27500 24 gauge STP Tefzel insulation.
Banana plugs                              -0.00021mv +/-
30pV              Equil. time 20 minutes. My best "Perfect" gold plated
plugs with copper wire.

Charlie

On 8/25/2014 3:36 AM, acbern@gmx.de wrote:

Well, you sort of answered your own question.  The equipment is called a
Thermal Transfer Standard, but instead of thermistors, it uses a
thermocouple.  Look at the manual for the Fluke 540B
(http://bama.edebris.com/manuals/fluke/540b/) and you'll see how it's done.
Basically, the AC source is input into the transfer standard, and the
standard's internal reference voltage is adjusted for a null on the
galvanometer.  Leaving the reference voltage setting alone, a DC voltage is
input into the unit, and the DC source is adjusted for a null on the
galvanometer.  At that point, the AC voltage source is equal to that of the
DC voltage source.

Ther are thermocouple-type thermal converters used for RF voltage
measurements with the transfer standard.  They aren't cheap, and you have to
have a converter for each range of voltages that you need to measure.  The
thermal converters used with this type of transfer standard isn't great (50
MHz or so typical), but their accuracy far surpasses that of the thermistor
type sensors.

There are other brands and models of thermal transfer standards, but I have
a Fluke model 540 and a few thermal converters.  That's why I referred you
to the manual for it.

Cheers,
Dave M

pa4tim@gmail.com wrote:

fred,
generally you raise a good point, I had the same issue of calibrating an ac voltage to a high level of accuracy. you need this e.g. to validate the self.cal of a 3458a or other precison stuff like the 8506a0.

what i would recommend to do if you want to keep costs down is:
in a nutshell, get a thermal converter in the lowest range you need and a second one on range above. build a set of resistor range extenders (rf type with appropriate connectors and housings) to expand the range to where you need to be max. get one of the thermal converter calibrated (the higher one usually, and you need to havr  good cal lab, should be <10ppm accuracy) and use it to calibrate the rest. generally, up to 20khz, the accuracy is some 20 ppm anyway for thermal converters! at higher frequencies, due to reflections and stray capacitance/inductance influences, the accuracy decreases. the resistor range extenders though, if build up correctly, only have a few ppm impact (there is a paper from nist on that, but this is only typical). you can calibrate all converters to the one you got externally calibrated. do some research in the web, when you do the calibration, you need to determine the so-called constant N. then do an ac, dc+, ac, dc-, ac measurement between the the two and establish the deviation, also establish the error propagation. the end result will be a set of highly precise (low inaccuracies9 thermal converters good enough to calibrate a 3458a an better devices. if you want to spend the money, you could also buy a set of converters/range resistors (with/without a 540), that typically is a few k altogether, while a single device sometimes is available for below 100 bucks. you need to have a stable 7.5 digit nanovoltmeter though for the measurements of the tvcs (34420a or 2182 typically ) and precision (stable) dc and ac sources. but in the end, all you need is a single calibrated thermal converter.

adrian

Look up the phrase "AC thermal transfer standard".

Using a heater/thermocouple element in a vacuum is the tried
and true way of linking an AC voltage to a DC reference.

-Chuck Harris

OBTW, trimming your quoted posts is considered friendly.

pa4tim@gmail.com wrote:

I just realized that all my measurement tolerances are  nV not pV. Sorry
about not checking units first.

Charlie

On 8/25/2014 9:50 AM, Charles Black wrote:

Adrian,
Do you have a link or title for the NIST paper that you mentioned?

Dave M

acbern@gmx.de wrote:

When the people fear the government, there is tyranny. When the
government fears the people, there is liberty -- Thomas Jefferson

Dave M

There is one more thing that enters into this discussion and that is
"reversal errors" on the DC.  The complicates the transfer somewhat.  AC is
always going "+" and "-".  DC is in one direction so you have to then
reverse the voltage to the Thermal Transfer Standard and then take the
average of the two readings.  That is why there is a "Reversal" switch on
the 540B.  When you are using a fixed voltage High Frequency Thermal
Converter you need an external DC reversal switch in addition to other
equipment.  You also need an AC/DC transfer switch so that you don't have to
disconnect the AC source and then hook up the DC source manually.  See the
540B again.

All Thermal Transfer Standards have some "reversal" error.  This is
controlled by the internal construction of the unit and exactly where the
glass isolation bead is located on the heating element.  The thermocouple
converter used in the 540B is selected to have a very low reversal error,
but always will have some error.  The error is fixed so you can approximate
a DC measurement once you have characterized the particular converter.  I
can't remember now but I think there can be up to around .05% reversal
errors on some converters, while the ones selected for the 540B are under
.01%.

Read the FLUKE "Calibration: Philosophy in Practice" for further
information.

Bill

----- Original Message -----
From: "Dave M" dgminala@mediacombb.net
To: "Discussion of precise voltage measurement" volt-nuts@febo.com
Sent: Monday, August 25, 2014 11:38 AM
Subject: Re: [volt-nuts] AC calibration

done.

is

the

to

(50

thermistor

have

https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts