[volt-nuts] High precision calibration for the poor man

[Samuel DEMEULEMEESTER]

Thank you so much for all your answers. 

Well, I'm not rich :-P I do not own the 5500A nor the 3458A. I will rent
them for a week in order to keep my other instruments up and running in
good, "calibrated" condition. 

I only own some 6.5 and 7.5 digits meter like 34401A, 34420A & Keithley 2000
as well as some old calibration equipment (Fluke 5101B, 731B and 510A) for
4.5 meters and less.

I just checked the manuals and that's correct : all of them can accept a
large offset (+/- 10%) on the input value for cal purpose, as long as the
source is stable enough. 

I feel a bit stupid :P 

The only exception seems to be the Keithley 2000 while calibrating AC
voltage/current. According to the cal manual, you really need an "absolute"
AC standard :-/



-----Message d'origine-----
De : volt-nuts-bounces at febo.com [mailto:volt-nuts-bounces at febo.com] De la
part de Dr. Frank Stellmach
Envoyé : jeudi 21 octobre 2010 19:20
À : volt-nuts at febo.com
Objet : [volt-nuts] High precision calibration for the poor man

  Hi Samuel,

your question has to be split into two problems.

First one is the transfer of the unit of the Volt from the 3458A to the
DUTs.

For sure it is possible to calibrate your DUTs relative to the 3458A as you
described, provided your DUTs require about 10 times less accuracy, as your
3458A.

Some manuals of DVMs, as the HP34401A, recommend the usage of an appropriate
calibrator as the Fluke 5700/5720A, or alternatively, the 3458A plus
appropriate stable sources.

In the latter case, the 3458A acts twofold, as a volt reference analog to
the 732A, and as a transfer standard for scaling the 5500As output, analog
to the 720A, due to its ultra high linearity. So the accuracy of a
calibrated 3458A can be easily transferred to the DUT, plus taking into
account diverse additional errors, like EMF, but this problem would also
exist with other standards.

Due to its limited DAC resolution, the 5500A will never provide an exact
10.000000V reading on the 3458A.

Trimming the output voltage to an exact value by a resistor divider is also
critical, due to the necessary stability of the divider, or you have
available a Fluke 720A.


Your proposed method, yet being an "accepted metrological practise", it is
nether the less quite cumbersome, and in most cases of DUTs it is not
necessary at all.

Most modern DUTs are easily calibrated electronically, ie they also accept
odd values for the reference voltages.

So you simply measure the 10.000073V output by the HP3458A and type this
reference value into your DUT.

Just read in the DUTs manual, how far off the reference may be from the
standard value, to be accepted by the DUT and to bring it within its
specification limits.

For example, the HP34401A accepts everything between 0.9 to 1.1 of Full
Scale (eg 9.000 - 11.0000V) for most ranges and functions.

In the same manner, if your DUT is analogously calibrated (by a trimmer),
simply trim its reading as near as possible to your reference value, ie
10.000073V, and you are done.



So, now for the 2nd problem, calibration "for the poor man".

Well, your are not a poor man, because you don't have a 5720A or a 5440A,
but only a 5500A Instead you have have the 3458A. In fact, owning this
device, you must be really rich.

But you are in a poor mid- and long term situation, as your 3458A is not a
very good volt standard (8ppm/yr.), and therefore needs very frequent
calibration (which will make you really poor), if you want to maintain a
defined level of uncertainty.

So your real problem is the transfer of the official volt from NIST or an
accredited cal lab to your lab.

To have less frequent recalibrations, a better volt standard, like a 732A/B
or similar would mitigate this 2nd problem.

Frank




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