[Test-Equipment] STD Cells----NBS Monograph 84

[wolfbob]

I have trouble stating things.

NIST uses the average of a group of cells to establish THE ABSOLUTE volt.
You then get your standard cell calibrated against this average and the
value is stamped on the side of the secondary standard cell. If you have
more than one secondary standard, particularly three or more, you can detect
when one is gone bad by differential readings between the ensemble.
Averaging your pack of secondary standard cells will do nothing to improve
any accuracy but will only corrupt the calibrated accuracy.

The misleading part is not the fault of NIST but simply that users have not
thought through what is going on. NIST averages BECAUSE they have NOTHING to
compare to and one is just as good as another and there is a tacit
assumption that the average is MORE STABLE than any one cell. Not more
right, not more accurate (I dunno what accurate means without something to
compare to) but simply more stable.

Your secondary cell has specified stability/aging and other known and
controlled quirks and as there is a standard to compare to, it has a
specified accuracy. Using more than one for a measurement just muddles the
data and serves no purpose.

WBob

----- Original Message ----- 
From: "Tom Dawson" <wb3akd at earthlink.net>
To: "wolfbob" <wolfbob at csnsys.com>; <test-equipment at mailman.qth.net>
Sent: Tuesday, June 01, 2004 10:54 AM
Subject: Re: [Test-Equipment] STD Cells----NBS Monograph 84


>
> Well, Bob, the bad news is that the national standards organizations are
> always going to be providing standard values with uncertainty values based
> on the mean values and standard deviations of a large number of
measurements
> made over time, and usually derived from intercomparison between different
> standards.
>
> From the NIST website:
>
> "The U.S. Representation of the Volt is maintained by monitoring the emfs
of
> several groups of saturated standard cells in ovens on a monthly basis
using
> the ac Josephson effect. "
>
> Most likely they take the mean and then assess the uncertainty based on
the
> standard deviations of the monthly readings.  NIST's uncertainty for
voltage
> measurements in => 0.2 ppm.
>
> With respect to local groups of standard cells, I don't recall anyone
saying
> that they were obtaining an absolute value based solely on the average of
a
> bunch of measurements.  If you read Technical Note 430, the starting point
> is the stated values of the cells, which is established by comparison with
> some other standard, and the Tech Note 430 procedures provide a means of
> monitoring the behavior of the cells over time.  If, using the statistical
> methods described in the text, the resulting cell voltages remain within a
> reasonable range of values, one can have confidence that the cells have
not
> drifted, or otherwise deteriorated since the last comparison with primary
> standards.
>
> Somehow I don't feel misled by NIST's process.
>
> regards,
>
> Tom
>
>
>
>
>
>
>
> ----- Original Message ----- 
> From: "wolfbob" <wolfbob at csnsys.com>
> To: "Tom Dawson" <wb3akd at earthlink.net>; <test-equipment at mailman.qth.net>
> Sent: Tuesday, June 01, 2004 12:22 PM
> Subject: Re: [Test-Equipment] STD Cells----NBS Monograph 84
>
>
> > I did not read the Monograph in detail but scanned it looking
specifically
> > for any indication that the variations in the emf of an individual cell
> was
> > possibly random and if they had any clues as to the distribution of the
> > randomness. I found NOTHING that would indicate anything but a
systematic
> > change in emf due to a number of factors (temperature, age, shock
history,
> > etc).
> >
> > I think the world has been misled by the FACT that the NIIS uses the
> average
> > of a number of cells to DEFINE the STANDARD volt. This is a definition.
It
> > is a good definition as it solves some of the problems germane in
> > determining a fundamental standard. But we poor folk are not trying to
> > establish and maintain an ABSOLUTE standard, but trying to establish and
> > maintain a secondary standard that has a voltage stamped on it side and
> that
> > is that. Averaging a bunch of these is no more accurate than the value
of
> > any one of the calibrated secondary standards. In fact the average is
> worse
> > as it is contaminated by your local differential measurement errors.
Using
> a
> > number of standard cells is useful in determining if one is bad or
changed
> > from its original, but the average is definitely not as accurate as what
> is
> > stamped on the side of a good cell.
> >
> > Either that or I'm wrong...
> >
> > I hope this discussion will bring out the experts (I am definitely
nothing
> > but a cranky old fat man) and this forum can expand into one of
> measurement
> > methodology as well as the interests for the equipment to support that
> > methodology.
> >
> > WBob
> >
> >
> > ----- Original Message ----- 
> > From: "Tom Dawson" <wb3akd at earthlink.net>
> > To: <test-equipment at mailman.qth.net>
> > Sent: Tuesday, June 01, 2004 5:02 AM
> > Subject: [Test-Equipment] STD Cells----NBS Monograph 84
> >
> >
> > >
> > >
> > > Monograph 84 regarding all you ever wanted to know about Weston
standard
> > > cells is at
> > >
> > >
> > > http://www.radio-science.com/std_cells/mono_84.pdf
> > >
> > >
> > > If you want it, download within the next couple of weeks because its
> about
> > > 6MB and consumes quite a bit of that site's storage.
> > >
> > >
> > > regards,
> > >
> > > Tom
> > >
> > >
> > > _______________________________________________
> > > Test-Equipment mailing list
> > > Test-Equipment at mailman.qth.net
> > > http://mailman.qth.net/mailman/listinfo/test-equipment
> > >
> > >
> > >
> >
> >
> > _______________________________________________
> > Test-Equipment mailing list
> > Test-Equipment at mailman.qth.net
> > http://mailman.qth.net/mailman/listinfo/test-equipment
> >
>
>
>
>
>