I would like to know more about your setup. Which source(s) are you using
for the input and which nanovoltmeter(s) are you using to transfer the 10V
TVC to your other converters? I remember you asking on PMEL forum about the
accuracy of using a 34420A nanovoltmeter. I did not see a response as to
whether you opted for an alternative like Keithley 2182(A) or a low noise
preamplifier connected to a 3458A.

I have done some reading about how NIST transfers their calibrations using
two TVC's in parallel and I am guessing that is what you are doing.

http://www.nist.gov/pml/div684/acdc/tcs.cfm

To transfer the accuracy up/down to other TVC's at different rated voltages
appears to be a difficult task since they typically need at least half the
rated max voltage to be within spec. It would be similar to starting with a
SR104 standard and transferring its value through a set of SR1010 and
SR1050 resistors using an ESI 242.

I have a few AC sources, and I would like to be able to verify my TVC's
without sending all of them out for cal. Ballantine quoted $600+ per TVC
and I haven't checked what Fluke would charge for each A55.

Todd

I personally did the following: I got a Ballantine 1605A transfer

Todd and anyone else would is interested,

Measuring a Thermal Converter against another Thermal Converter is a bit of
a black art. The main problem is thermal converters are a square law device
that is if you change the input voltage by a factor of 2 the output voltage
will change by a factor of approximately 4. Now I say approximately because
for most thermal converters like the Fluke 540B, A55, Ballantine and Holts
the actual factor can be anywhere from 1.4 to 1.8 due to losses in the
thermal converter. In the literature you often see this factor refer to as
the N factors. Each thermal converter will have its own N factor which must
be measured to make sense of the measurements.

It even becomes more difficult in that the AC-DC difference of a thermal
converter is defined as (Vac-Vdc)/Vdc where Vac and Vdc are the inputs to a
thermal converter which give an equal output from the converter. Also Vdc is
the mean of the forward and reverse DC voltages. The problem is that when
you have two converters connected in parallel you cannot balance both
converters AC and DC inputs to produce equal voltage out of the converters
at the same time because each converter has its own AC-DC difference and its
own N factor.

It anyone is interested I can send them a technical paper that describes
this process and the appropriate math to use but I cannot send it to the
list due to copyright issues.

Kind Regards,

Stephen Grady
Sydney Australia

-----Original Message-----
From: volt-nuts-bounces@febo.com [mailto:volt-nuts-bounces@febo.com] On
Behalf Of Todd Micallef
Sent: Thursday, 10 July 2014 2:36 AM
To: Discussion of precise voltage measurement
Subject: Re: [volt-nuts] AC Voltage Measurement Standards

I would like to know more about your setup. Which source(s) are you using
for the input and which nanovoltmeter(s) are you using to transfer the 10V
TVC to your other converters? I remember you asking on PMEL forum about the
accuracy of using a 34420A nanovoltmeter. I did not see a response as to
whether you opted for an alternative like Keithley 2182(A) or a low noise
preamplifier connected to a 3458A.

I have done some reading about how NIST transfers their calibrations using
two TVC's in parallel and I am guessing that is what you are doing.

http://www.nist.gov/pml/div684/acdc/tcs.cfm

To transfer the accuracy up/down to other TVC's at different rated voltages
appears to be a difficult task since they typically need at least half the
rated max voltage to be within spec. It would be similar to starting with a
SR104 standard and transferring its value through a set of SR1010 and
SR1050 resistors using an ESI 242.

I have a few AC sources, and I would like to be able to verify my TVC's
without sending all of them out for cal. Ballantine quoted $600+ per TVC and
I haven't checked what Fluke would charge for each A55.

Todd

I personally did the following: I got a Ballantine 1605A transfer

converters (0.5V to 100V).

speed is good.

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.

This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com

I have a set of 7 TVCs, from 1 to 100V. having them all calibrated externally is just too expensive. There is no doubt that having all 7 TVCs calibrated at NIST or PTB, wherever you are, is much more precise, I will just not spend it and live with the accuracy I get.
As source I am using a Datron 4808, as nanovolt meter an Agilent 34420. Problem with this is, you pointed it out, Agilent does not specify transfer accuracy, I am in contact with them, so far no outcome though, not sure there will be any. So I have not yet completed my error calculation for the TVC cal. I will probably end up doing some calculations based on data sheet an then vaildate by measurement. Need that on the 10mv range only, so effort is limited. additionally I also have a ratio transformer, which is good to about 10kHz, that also allows for validating the TVC results by comparison in lower frequency ranges and narrow down the tolerances.
I do have a 182-M, but do not trust it, I saw a lot of drift doing some other measurements although it passes performance verification, and I just got a 2182, which has an issue that needs to be fixed first. I need to say the 34420a is very stable with the digital filter switched on and you can see the nanovolts walk until the TVC stabilizes. vice versa, if you have an issue in the setup (thermal drifts...), you also seethat very well, so the filter is not hiding this.
as far as your comment re. calibrating a set of resistor cal from one calibrated one, whats wrong with it? if you do that with a stable voltage and by voltage divider measurement using the linearity of the 3458A, you end up with pretty decent accuracies. I cannot comment on the 242, just did some quick checks some time ago and my quick assessment then was it is worse/not better than the 3458a method.

in a nutshell, what i am doing is that I first establish the dc (+/-) output of the ref. TVC at nominal and then determine the ac voltage from a precision, highly linear (datron 4808) ac source that generates that default output voltage at the key frequencies. that establishes a set of ac voltage settings. for these the deviations of the TVC are known from the calibration. then I do the same with the TVC to be calibrated. (factor of 2 nom. voltage is important to stay within allowable range). that way you can link both.
sure, sending all to cal is more precise, if you get the right lab at least, but also very expensive.

would sure be interested in your tech paper

The ultimate AC standard would be NIST's AC Josephson junction array. Sam Benz and his team continue to make advances, pushing the accuracy and output level. They have also substantially reduced the needed electronics placing much of the peripheral electronics in a custom built two-U enclosure. NIST has also recently introduced a fully programmable JJA that automates most of the set up and operation. It won't be long before we need 10 1/2 digit DMM's
Cheers;
Thomas Knox

The remark about calibrating the resistors was merely an example of how I
thought you may have been transferring the calibration of the 10V TVC. My
first thought was that you compared the 10V TVC to the next closest model
at fixed points. The inter-comparison would then continue from that model
to the next value down (or up).
I had a hard time trying to understand how that could be accurate, and the
thought of using a ratio transformer didn't cross my mind. Thanks for the
explanation.

I can see how the costs of calibrating a TVC can be so high. I am guessing
few labs possess the equipment to calibrate them and places like Ballantine
probably use NIST for their cals.

Todd

Fellow Volt Nuts,

The ultimate standard of AC Voltage is still primary sets of Thermal
Converters that are characterised in terms of their thermal models and the
modelled lumped elements ( inductive and capacitance effects). The
programmable AC JJA's are very stable and reproducible AC sources cannot be
primary references in themselves because their frequency response (although
modelled) is still calibrated against other standards mostly TVC's.

To those who have requested a copy of the paper on calibrating thermal
converters against other converters please be patient and I will try and get
my response over the weekend.

Kind Regards,

Stephen Grady
Sydney Australia

-----Original Message-----
From: volt-nuts-bounces@febo.com [mailto:volt-nuts-bounces@febo.com] On
Behalf Of Tom Knox
Sent: Friday, 11 July 2014 9:47 AM
To: Discussion of precise voltage measurement
Subject: Re: [volt-nuts] AC Voltage Measurement Standards

The ultimate AC standard would be NIST's AC Josephson junction array. Sam
Benz and his team continue to make advances, pushing the accuracy and output
level. They have also substantially reduced the needed electronics placing
much of the peripheral electronics in a custom built two-U enclosure. NIST
has also recently introduced a fully programmable JJA that automates most of
the set up and operation. It won't be long before we need 10 1/2 digit DMM's
Cheers; Thomas Knox

output of the ref. TVC at nominal and then determine the ac voltage from a
precision, highly linear (datron 4808) ac source that generates that default
output voltage at the key frequencies. that establishes a set of ac voltage
settings. for these the deviations of the TVC are known from the
calibration. then I do the same with the TVC to be calibrated. (factor of 2
nom. voltage is important to stay within allowable range). that way you can
link both.

least, but also very expensive.

of

device

voltage

because

Holts

as

must

to a

Vdc is

when

converters

its

using

10V

the

noise

using

voltages

the

with a

and

protection is active.

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

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

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.

This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com