not LM339 (it is a comparator) but the LM399 is a heated reference.

really LM119AH? or LM129AH (I cannot find a reference when looking for
LM119)

And the LM129AH has a tempco of around 6 ppm/K.
so you will have either to heat it or to monitor the temperature exactly.

But consider: when you maintain only one reference you are the happy man
with one clock.
And you will never mention when the reference will drift by shipping.
I would maintain at least 4 equal references. Shipping only one at a time.
And comparing to the other 3 local references before and after shipment.

With best regards

Andreas

"Andreas Jahn" Andreas_-_Jahn@t-online.de writes:

Hi Andreas,

In view of your comments, shouldn't we be looking at a LTZ1000 for the
"traveller"? It's relatively expensive but not out of reach I would
think.

John

Shouldn't

--

John Devereux

Hello John,

a LTZ1000 is a fine thing no doubt.
But I personally would not use it as a travelling reference.
It might get lost or damaged during transportation.
And for this its too expensive.

Further you will need to place a short cirquit proof
output buffer to the reference. Otherwise the Output
Voltage will be shifted at least for several days when you
accidently short the output. A large capacitor may be
enough for a short cirquit. When the output is shorted
the temperature regulator goes to maximum temperature
and introduces some hysteresis on the chip.
Its not very easy to get the "hand adjustment" off the
LTZ1000 cirquit since the temperature setppoint
transistor is very sensitive.
(see my cirquit from November 6th the last year).

A LM399 is stocked at Digikey for some bucks. And you
will have to add only a voltage regulator and a constant
current source (resistor, LM334 temperature compensated,
or a simple LT1013 Opamp as Current source).
With some blocking capacitors you get a robust cirquit
with a low output impedance. (around 0.5 Ohms against
around 5 Ohms at unbuffered LTZ).

With best regards

Andreas

Andreas Jahn wrote:

My bad - I should never do e-mails after midnight!

I meant, of course, your reference to the lm399.  I have 3 lm199ah parts
that I was referring to.  The spec sheet shows max TC of .5ppm/C.

:-)

Excellent idea.  Well I have three lm199ah and three lm399ah (not quite
as good).  My construction project just expanded.

Thanks!

--
Bob Smither, PhD                                  Circuit Concepts, Inc.

---=======
"Our enemies are innovative and resourceful, and so are we. They never
stop thinking about new ways to harm our country and our people, and
neither do we."
--George W. Bush, August 5,2004
http://www.whitehouse.gov/news/releases/2004/08/20040805-3.html

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

So the LM199AH and LM399AH will have the same chip.
The only difference will be temperatues in the final test.
In my opinion you have 6 identical references.
So you can sort out the 4 best of them.

With best regards

Andreas

Bob,

You've heard from the experts now here's some thoughts from a mere hobbyist.

First, there was discussion, on this list, in Nov. 2008 on this issue. I
would also be happy to be able to rotate a homebuilt transfer standard
through a kind hearted soul's home lab ( don't worry Charlie, I don't intend
to piggyback on your good hearted offer to Bob) equipped with 732A's and
access to an employer's NIST traceable lab gear. In fact, a master
metrologist did take pity on me, and actually gave away a little standard
which I had to use as soon as I got it at exactly 63.5F to perform the
transfer.  That little standard "moved away"  300ppm in 90 days from the HP
3456 and ancient HP735A transfer standard I applied it to (the 3456 and 735A
were only about 30ppm apart by then). Take a look at the Nov 2008 archive.
Some were happy with the cost/performance of the Geller product
http://www.gellerlabs.com/SVR%20Series.htm, Some others considered its long
term stability suspect. If that's the case, in time, you'll just have
another suspect clock in your collection. As my knowledge, and appetite for
metrological perfection falls quite short in comparison to the scholars and
master designers who dominate this list, a 10ppm transfer capability, from a
transfer standard with a long term drift of less than 5ppm/year would be a
joy. In that regard, my own preference would be to build the clone of the HP
3458A reference board described by Max Carter
http://www.maxmcarter.com/vref/. The use of metal film resistors on the
board as shown I will generously presume to have been placed there after the
builder ran out of funds for Vishay wirewounds, or thin films. In any event,
I can't afford precision, low tempco resistors either. I would also use
metal films, but I would ovenize as per your off-the-cuff, and HOPE the heat
blows off the humidity. I would build an aluminum plate "coffin" around all
components except the LTZ1000 (is it also position sensitive like LM399s?),
build a precision heater control circuit to heat either resistor arrays or
pilot lamps sprinkled around the board, fill the "coffin" with insulating
foam, and seal. Maybe this arrangement would have better long term stability
than the Geller, or maybe it's DOA. Comments from scholars and master
designers welcome. I realize, with the whole board ovenized, shipping under
battery power may be a bit impractical.

As far as relying on the kindness of strangers, that's ok for a one off.  To
regularly (e.g. yearly) calibrate whatever homebuilt transfer standard you
come up with, is a different matter. I realize the list members with the
highest level of standards control, are also the same ones immersed in
highly complex lab project on their own precious time. They can't be
expected to waste time tearing open scores of packages, and then repackaging
standards for return. May I ask if there's someone on this list who has a
wife, child, nephew, niece, or cousin, willing to run a small business from
dad's/grandpa's home lab? While Geller charges $10.00 to recalibrate your
Geller "transfer reference" by throwing it across an HP3458A, a volt-nut
Grandpa Lab may just be OCD enough to demand to have the little family
business do a right and proper transfer right from grandpa's bank of 732A's.
In which case, you aren't going to get away with $10.00 cal+ship. Grandpa's
lab would have to charge at least $50.00. I would certainly be willing to
pay $50.00 vs. $500 to a commercial lab. What a bargain!

Regards,
Stan

On Thu, Aug 25, 2011 at 1:14 PM, Bob Smither smither@c-c-i.com wrote:

In message CAPmQ2J8qtg85nhTQG4EDTRN4qJyr_cb+N562OON+-NW=_++8xQ@mail.gmail.com
, Stan Katz writes:

Copper-tubing is easier to work with and has better heat-conductivity.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Hello,

I do not think that you can get good results with standard metal film
resistors.
Perhaps you can use them for the 70K-values but the current setting
and temperature setting resistors should be at least precision
wire wound types. (UPW50 or Riedon USR2).

the LTZ is much less position sensitive.
Frank measured around 1ppm for the LTZ in September 2010 in this list.
My measurement was around 0.3ppm in November 2011.
(tilting LTZ1000)
But the LTZ1000 is much more sensitive to common mode noise and RF.
So you have to take much more effort to shield and filter on a LTZ1000.

With best regards

Andreas

I am thinking about doing a bit calibrating as a sort of service and to earn some money to spend on calibrating my stuff ( p.e. the 731A transfer standard) with our national standard buro. so not to make money on it, just to cover expenses.

But he problem is I, have no clue what to ask and if there is. Market. I have a mint condition guildline cabinet with known documented history and from the first user. So I can, after the return of the transfer, compare it with my guildline and have a longtime stable reference. But a calibration here is not as cheap as in the states I think. A friend had a tek current probe calibrated, that was over 500 dollar. But i think most " customers" will have a 4.5 or 5.5 DMM and in that case my cells and calibrators are allready good enough.

I bought a new calibrated keithley 2000. As soon as It arrived I warmed it up and measured my cells and calibrators , that I allready had warm up, and documented that. The cells where less then 3 uV away from the last documented calibration and still spaced appart like they where then. So for a 5,5 digit i will be close enough. An old secondhand bought 6,5 digit will be closer then it was.
I have a the 4 cell guildline cabinet, 731A, 332, 750, 760, 720, philips DC volt/current calibrator, and. Fluke 5xx AC calibrator. Besides that a GPS controlled thunderbold Oscillator, a HP 5 MHz standard, counters connected to the 10 MHz ref upto 18 GHz so I can do counters too. Also have timemark en pulsgenerators for scope adjustment but that really is on scopes like tek 7000 or 500 series very much labour if you do not have the calibration plugins. A normal scope is still about 10 hours work.

The problem is I have plenty of time but i am a bit disabled so I can not work long periods and that means two hours warming up, work an hour on the instrument and rest for an hour so it would take me days. The calibration of my 7704 , witch is rather complex took me two weeks. A simple 547 3 days. A letter plugin about 2 to 4 hours. A DMM like a fluke 8000 or HP3535 is done in about 2 hours, a keithley 199 doing the memory calibration is done in an hour or so but a HP3490 is many hours of work. (and there is the problem of shipping and risk invalved. There can be people trying to missuse your service by sending not functioning gear and telling you broke it or it gets dammaged or lost in the mail.)

Are there voltnuts how do such project to earn back a calibration. Are there other Dutch volt-nuts here ?

I do not know there is market for this in the Netherlands. Not much voltnuts around and we Dutch are famous for our cheapiness. I am very active on a very big electronics forum, there are about 4 other volt nuts there, who probably would spent 50 euro or so on a calibration. The rest ( about 80 % i think) use there holy 5 tot 30 bucks multimeter. A 50 euro model is considerd to be rreal expenive and overkill. Cheap is holy around here and being cheap made to an art.
Lot of topics about Rigol, atten and owen scopes asking if it is wise to buy because it is a lot of money and or if  someone knows a cheaper brand or model  :-(
But a calibrion is looking what the deviation is, most think a calibration is the same as adjusting it whitin specs. I do that with my own gear and for a lot of gear that is a hell of a job. And to do that for 50 euro does not sound tempting.
Fred PA4TIM

Op 1 sep. 2011 om 07:45 heeft Stan Katz stan.katz.hk@gmail.com het volgende geschreven: