On 5/9/2012 11:53 AM, Bob Smither wrote:

On 05/08/2012 12:52 PM, Charles Black wrote:

Hi Bob,
What a great job you are doing by organizing this cooperative field
comparison! It sounds like a lot of work though!
Looking at the first lab photo brought back memories of where I started
with my new (to me) 3458A three years ago! I hope it is OK to offer
suggestions that will make measuring microvolt levels on a 3458A more
accurate . Similar techniques can be used on any meter though.
I have been periodically monitoring my Datron 4910 and 4912 10V
standards for about 3 years looking at measurement  drifts mainly
caused by temperature. I have many hours of recorded data looking at
0.1 PPM resolution drifts in my system. In order to improve data
stability I  had to change my measuring technique from a BNC cable with
dual banana adapters (what is shown in the photo) to twisted pair
copper wires or shielded twisted pair wires (22 or 24 gauge) to
minimize input thermal offsets and settling time. Tin plating on the
wires measures about the same offset voltage as copper so it hasn't
been a problem to use it so far. If I put a huge thermal mass on my
input terminals (plastic dual banana plug) it can take many minutes to
stabilize and even then air currents will keep my meter from
stabilizing as completely as it can. Today my meter has about -240nv
offset voltage from a copper wire short but I see -1.2 microvolt with a
shorted dual banana plug, even a gold plated one. I have a shielded
dual banana to BNC adapter that's even worse.
My 3458A is very temperature sensitive so I constantly monitor the
temperature right in front of the input jacks to 0.1 degree C. I try to
keep mine at 23 degrees and do frequent ACAL's and repeat the ACAL if
more the terminal temperature varies much over a degree from 23C. The
TEMP? is also very important to watch and could be used to predict
temperature offset error. The TEMP? creeps up over a few months due to
dust in the filter but it is easy to clean with a little tap water.
The 3458A has a variable resolution feature that can be a problem at
high resolutions. Full DC resolution is acheived at NPLC = 1000. Lower
NPLC levels can be used to speed up measurements if loss of accuracy is
not a concern.
My shop has distinct thermal stratification so I try to keep the test
unit  I am measuring at the same height as the meter and  close to the
meter when doing accurate work.
As far as the portable standard goes - it would be great if it had gold
plated dual banana spaced terminal posts similar to the 3458A posts for
best repeatability and flexibility but, of course, any jacks will work
if they have enough time to stabilize.
Charlie

Thanks Charlie - great suggestions.

My goal has always been to know the values from the two references in the TS to
with 10 ppm.  For the two refs (nominal 7 Volt and 5 Volt) this is 70uV and 50 u
V.

My goal is to be to measure DC voltages between 10V and 1V with a 1 ppm
accuracy. I have my Datron 10V standard back from Fluke with each cell
at 0.6 uV/V measurement uncertainty (and data to 0.1 uV for 4 cells)
and a 3458A which has a 0.1 ppm linearity. The temperature drift of the
Datron 4910 is 0.05 ppm/DEGC. I checked the linearity of the 3458A with
my calibrated Fluke 720A and they agreed well within 0.1 ppm so my
3458A linearity uncertainty is 0.2 ppm or better. So far I have a total
uncertainty of 0.8 ppm and have to add in the transfer uncertainty
which can be as little as 0.02 ppm at 10V using my and an 845 similar
to yours. I can measure the Datron with the 3458A to see what the
temperature drift is and use that as a correction factor but I have
seen very little 3458A drift that is not temperature related.  And I am
sure that there are other errors that  need to be included as well.

I have one meter here that I can easily see thermal effects on (an old Fluke
845AR - 1uV full scale on most sensitive setting) so I appreciate your notes
about the use of copper wires.

Your first division is at 2uV on the 845's 100uV scale so you can
compare references within +/-100uV to 2uV. My 845AB wanders around
about +/-0.2uV so I usually use my HP419A which is stable. My 845 uses
the early input amp design and I am not sure how much drift is normal.

Even though only trying for 10 ppm I appreciate that any source of error that
can be eliminated needs to be considered.

With your permission I would like to add your notes to the page here:

[1]http://c-c-i.com/ts

so I refer folks that are taking measurements on the TS to them.

That's fine with me. You have my permission.

Best regards,

volt-nuts mailing list -- [2]volt-nuts@febo.com
To unsubscribe, go to [3]https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.

Charlie

References

1. http://c-c-i.com/ts
2. mailto:volt-nuts@febo.com
3. https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts

The third volunteer to measure the TS has reported back his measurements and I
have posted them on the project page:

http://www.c-c-i.com/ts

Thanks for the interest shown in this project!  I have two more volunteers in
line to test the TS and will get it mailed to the next one this week.

A fellow volt-nut across the Atlantic has expressed interest, but we have not
found an economical way to ship it.

Best regards,

Bob Smither

I just dropped off the Volt-Nuts TS at my shipper, headed to the fourth
volunteer for this project.  I will report his measurements as they become
available.

The prior readings are here:

http://c-c-i.com/ts

--
Bob Smither, PhD                                  Circuit Concepts, Inc.

---=======
Foreign aid might be defined as a transfer from poor people in
rich countries to rich people in poor countries.
--Douglas Casey (1992)

---=======
Smither@C-C-I.Com  http://www.C-C-I.Com  281-331-2744(office)  -4616(fax)

I have added more plots of the data taken by our fourth volunteer - so far over
a 3 week period.  The plots are posted here:

http://c-c-i.com/ts

Somewhat unexpected, at least by me, was the drift observed the first week and a
half.  The two references (LM199A and MAX6350A) both drifted down several PPM.

The TS1 has been powered on for several thousands of hours - between being in my
lab and in the labs of other volunteers.  It was on 24/7 for almost 2 months
here before being shipped to the fourth volunteer.

The TS1 is shipped un-powered.

There is a large humidity difference between my location (high humidity gulf
coast of Texas) and the location of the volunteer (high altitude, low humidity).

There is also a significant pressure difference between the two labs.

The LM199A is hermetically sealed.

Although a small sample, the two references appear to be similarly affected by
whatever caused the drift - similar range of drift, similar time constant.

Details (schematic, etc.) about the TS are available at the above link.

Any ideas about what could cause the drift we are seeing?

Thanks!

--
Bob Smither                                            smither@c-c-i.com

---=======
In all criminal prosecutions, the accused shall enjoy the right to a
speedy and public trial, by an impartial jury of the State and district
wherein the crime shall have been committed; which district shall have
been previously ascertained by law, and to be informed of the nature and
cause of the accusation; to be confronted with the witnesses against him;
to have compulsory process for obtaining witnesses in his favor, and to
have the assistance of counsel for his defence.
-- Sixth amendment to the Constitution of the united States

---=======
Circuit Concepts, Inc.                                      281-331-2744