Charlie wrote:

My understanding of Bob's original proposal was that he wants to
build a stable reference of uncertain voltage, and send that to
someone (or several someones) who will render an opinion on its
actual voltage and send it back.  A well-calibrated 5440B and a null
voltmeter (Fluke 845A/AB/AR, HP419A) would work, as would a 732A or
equivalent and a precision voltage divider (Fluke 720A or equivalent).

Bob did mention having the recipient power the traveling reference up
-- if possible, it should be shipped "hot" under battery power so
there is no power cycle/retrace to worry about.

Best regards,

Charles

Charles P. Steinmetz wrote:

Correct.

I can see the advantage of this, but as an example the lm199ah draws
about 15 mA at 25C.  If exposed to low temperatures during shipment it
will draw more.  Perhaps this could be reduced if the part is put in a
thermally insulated box.  A one week shipment (out and back), at 15 mA,
would require a 2600 mA-hr battery.  Certain not impossible, but
somewhat heavy.

With the goal of 10ppm overall uncertainty, will the thermal hysteresis
be a limiting factor if the unit is shipped un-powered?  How long should
it be powered at the remote lab before a measurement is made?

Best regards,

Bob Smither, Ph.D.                                Smither@C-C-I.com

---==
But when a long Train of Abuses and Usurpations, pursuing invariably
the same Object, evinces a Design to reduce them under absolute
Despotism, it is their Right, it is their Duty, to throw off such
Government, and to provide new Guards for their future Security.
-- Thomas Jefferson, Declaration of Independence, July 4, 1776

---==

Practically I can power a LM399 with 8 NiMH cells (9,6V 2000mAH)
for about 2 Days before I need to recharge.
I tried better thermal insulation but this has no significant influence.

On my LM399 with a 6,5 Digit instrument (10uV resolution)
you will not see any thermal hysteresis.
(except when comparing 2 of them in a higher resolution range).

To the diagram:
Y-axis is the output of my 5V  24 Bit ADC in mV. X-axis is time in minutes.
(integration time 1 minute so measurement noise is around 1uV at ADC-input)
The reference is attached to the ADC by a precision 2:1 divider (LTC1043)
Voltage before power off 3431.174 * 2 mV
1 minute power off
1 minute value out of diagram
about 15 minutes warm-up time
about 1.5 uV * 2 = 3uV Hysteresis.
2nd power off for one minute off gives additional shift of 0.5 * 2 = 1uV.

So I would recommend a power-up time of at least 1 hour or so.
The instrument which is used to measure the value will need this too.

With best regards

Andreas

Andreas Jahn wrote:

Too short to mail it out and return.  I think a line powered unit makes
the most sense (especially in light of your following data).

Haven't tried it yet - thanks for the input.

This is what I gather from the data sheet.

Thanks for the data.  Looks encouraging.

Noted. Thanks.

Here are my back of the envelope preliminary design / calculations:

Power transformer / rectifier / capacitor => ~35 Volts DC.
LM7815A => 15 Volts (regulated).

LM199A: Vout =~ 7 V.
Set Id =~ 1 mA with an Rs = 8.06 K 1% resistor from 15 V to diode.
Power on-chip heater from 15 V.
Rout =~ 1 ohm => delta Is of 7 uA causes 1ppm shift.

Require Zin of measuring VM > 10M => 0.7 uA => 0.1 ppm error.

LM7815 TC =~ 1 mV/C => 0.125 uA/C => 0.018 ppm/C.
LM7815 line regulation =~ 13 mV/V.
For 35V +/- 10% => 46 mV => 5.7 uA => 0.813 ppm.

Rs TC =~ 100 ppm/C => dIs/dT =~ 100 ppm/C => 0.1 uA/C => 0.014 ppm/C.
Rs stability =~ .5% / 1000 hrs => 5.0 uA => 0.714 ppm.

Comments - feedback?

The two largest errors (other than 20 ppm stability of the lm199ah) are
both under 1 ppm.  They could be reduce further but is it worth it given
the stability of the lm199a?

Best regards,

Bob Smither

Bob wrote:

When one is trying to achieve repeatability in the single-digit ppm
range, I would not leave anything to chance -- and retrace
(hysteresis) is always a matter of chance, particularly if the device
has been powered off for a long time (several days) and may have been
exposed to significant temperature shifts during that time.  In my
view, the best approach in this instance is to use something like a
5Ah gel cell (to ensure relatively little voltage droop during the
transit time), with at least one stage of preregulation.  The total
package should be under 2 kg and should easily fit into a USPS
medium-size flat-rate box that ships for $11 (this assumes that you
and the recipient are both in the US).  UPS should be similar.  (If
$22 for a round trip is a deal-breaker, we might conclude that one
wasn't very serious about the project.)

You also might include a charger, so that the recipient can top off
the battery before measurement (if necessary) and before return
shipment.  All measurements would be made with the charger disconnected.
Best regards,

Charles

Hi:

Batteries are considered hazardous material and can not be shipped
without special Hazmat paper work which complicates things.  I think
there is an exception for some carriers, but not others, when the energy
is below a couple hundred watt hours (amp hours times volts).  The
battery pack  mentioned (2 AH * 9.6V = 19.2 WH).

Have Fun,

Brooke Clarke
http://www.PRC68.com
http://www.End2PartyGovernment.com/

Bob Smither wrote:

Brooke Clarke wrote:

I wondered about this.  Since 9/11 there is precious little we can do
without getting someone's permission 8^(.

Andreas's results imply that, at least at the 10 ppm level, that
hysteresis may not be an issue.  It is stretching my measurement
capabilities, but I will try some experiments after I have some data on
several potential Traveling Standards.

Thanks,

---=====
Bob Smither, PhD                                Circuit Concepts, Inc.
"Every election is a sort of advance auction sale of stolen goods."
-- Libertarian H.L. Mencken
Smither@C-C-I.Com          http://www.C-C-I.Com          281-331-2744

---=====

"Charles P. Steinmetz" charles_steinmetz@lavabit.com writes:

[...]

Hi Charles,

I think you will end up on Fox news and the Reference will end up
"defused" with a controlled explosion :)

I am still thinking LTZ1000 is the way to go on balance. It has all the
complications Andreas pointed out. But I think:

• stable without keeping it always powered up

• stable with RH

• stable with ambient temperature (compared to others)

• stable with respect to orientation, less likely to be affected by
  handling(?)

Does anyone know if the ebay / Chinese LTZ1000 are the real thing?
E.g. 250823975393.

Regards

John

--

John Devereux

Bob wrote:

With 20 volts of overhead, why not use two stages of pre-regulation
(say, LM7824/LM7815) to improve the line regulation even
further?  You might also use separate LM7815s for the reference and
heater supplies, to minimize the reference supply changes due to
varying heater current and the associated load regulation/wiring drop.

It is good design practice to reduce all sources of error to the
degree that they can can be reduced for little cost.  In this
context, I consider two additional three-terminal regulators "little
cost."  (If two additional regulators are just too much to bear, the
heater could be run from the 24 V supply.)

BTW, here is a standard similar to what you are
proposing:  http://gellerlabs.com/LNVR%20Series.htm  It uses an LM399
and the scaling amplifier of an AD587 to produce 10 V (I have
previously criticized this design for using the 587 instead of a
discrete op-amp that would give better performance, but the 587 used
this way appears to meet your design goals).  Geller also offers a
$40 bare board based on the 587 alone, for which he claims short-term
transfer accuracy of +/- 5
ppm:  http://gellerlabs.com/SVR%20Series.htm    Here is another one
(+/- 25 ppm at 5
V):  http://www.voltagestandard.com/New_Products.html    I believe
that both vendors will recalibrate them for a nominal fee.

Best regards,

Charles

John wrote:

I recently ordered several of these, and I also quite often order
larger SLA batteries for UPS replacements.  They all arrive by UPS,
USPS, or FedEx with no drama and no hazmat labels that I can
see.  (But I don't think they are shipped by air, if someone is
contemplating that.)

I concur, since "better is always better."  But it does appear that a
399 can, without undue difficulty, meet the design specifications, at
least if it is kept powered on.  I still think hysteresis is a
significant worry if the off time extends to days, particularly if it
is subjected to wide temperature changes during shipment.

Best regards,

Charles