What do you think of Geller's work?  He uses the AD587 as a voltage
reference, and writes of results > = the AD specs.

Your need seems close to the principle used in his thermal
anemometer,  which he patented.  IIRC, it measures temperature change
as changes in the stable voltage response of a reference, which is
then proportionate to air current.

http://www.gellerlabs.com/MGTA%20kits.htm

From the link, and use his patent number to describe the process better.

PS: This is my first post.  Thank you all for the good information.

At 06:03 PM 8/14/2010, Andrea Baldoni wrote:

Sincerely,

Marv Gozum
Philadelphia, PA

On Sun, Aug 15, 2010 at 03:34:29PM -0400, Marvin E. Gozum wrote:

I think the same of voltagestandard.com. These boards adds nothing to the
original chips they use (the chips are not selected)... they may
eventually just degrade performance with layout/mechanical stresses/passive
components.
The calibration service they offer, because it is applied to this kind of
references, is not useful as a "transfer" to anyone with already decent
instruments (with better references inside).

As far as I can see from the circuit, he uses that poor LM399 just as
heater resistor and RTD. The reference diode is unconnected.
The reference of the wheatstone bridge is the +15V power.

I don't see what kind of principle he could have patented. Hot wire anemometers
are not a new and this is the worst one I could imagine.

Maybe he has tried to use the LM399 to sell voltage references, he has failed
to obtain good results (it's not so easy to use as the AD587) and having
already bought some (and probably having burned the diode...) he recycled them
in this kit...

Best regards,
Andrea Baldoni

At 06:36 AM 8/16/2010, Andrea Baldoni wrote:

Hello Andrea,

Yes, as for transfer reference, if you have access to a calibrated
reference you can adjust your 34401 too, you don't need such.  But, I
recall you mentioned earlier you didn't have a calibrated reference
for your precision DMM.  At least empirically, Geller's reference has
been around since 2004, he now has longevity statistics on his side
that his board is a capable of at least a 1 month transfer of 10. 000
00 Vdc reference from his lab to you, if so inclined. Malone doesn't
detail much on the site beyond the basic specs of the original chip.

Sorry for a reply delay, but I was intrigued enough to get one and
measure it as soon as possible, lest it drift!  Sure enough, under
the same temperature conditions, its accurate to my calibrated 8456A,
reading 10.000 00 Vdc.  I am now gathering stats on its performance.

It is a bit unusual, but in the article he wrote for a hobbyist
magazine, detailing this device, he does describe that he didn't
expect this response either, from the chip.  But it seems the heater
of the 399 responds in a way not well characterized, so he seized the moment.

I'm not so sure how the well it measures cfm, but the setup is so
simple and cheap, one can easily breadboard the device and see of
what value it is.

Its interesting this has come up for his services several times, the
issue of sales.  But I think differently.  I don't know him, but you
can easily find out what he does for a living via google, As an
aside, I don't think anyone who can afford a new HP 3458A to use as a
hobby will unlikely be making a living selling electronic kits.

Cheers!

Sincerely,

Marv Gozum
Philadelphia, PA

Marvin wrote:

Geller has a patent covering a method of improving the temperature
coefficient of an AD58x or similar monolithic reference using an
LM399 or other ovenized reference
(http://www.pat2pdf.org/patents/pat7382179.pdf).  A friend has a
little reference box he bought from Geller that claimed to use a
circuit based on the patent.  However, according to the patent,
Geller didn't use opposite tempcos to offset each other, he
essentially just swamped the internal reference and used the AD58x as
an output scaling amplifier for the ovenized reference.  So why use
an AD58x in the first place?  You could do significantly better with
a precision op amp and the ovenized reference, with a little
attention to details.  So, the patent appears to be what we used to
call a puff patent.

I notice that his current offering is just a PCB with an AD587 and a
voltage trimmer
(http://gellerlabs.com/SVR%20schematic.pdf).  Apparently, he
concluded that his patented circuit didn't improve things over the
current AD parts enough to be worthwhile.

This is not to say that his references aren't any good, or don't do
what he says, or aren't good value.  They are certainly reasonably
priced, and he is very clear about what they are and are not
(http://gellerlabs.com/SVR%20M.html).

Best regards,

Charles

Hello Charles,

Thanks for the links and the insight.  Sorry for delay, but was also
reviewing the old commentary on time-nuts about the Geller
references, his patents, and more on voltage standards not on this
lists' archives.

Overall though, the more important item is taken as a whole is this
box with the patented stabilizing circuit a better reference than
something else, assuming this is what your friend has?

I agree with your criticism of the circuit function.  See the
schematic on the patent.  There is a novel item, notice that the
supply voltage for the LM399 comes from the vref output of the AD587
op amp, rather than V+ as expected?  This provides an ultra stable
supply for the LM399, compared to one from a standard regulator or
battery, while also putting a continuous load on the op amp.  Not
sure what this does to the ppm/oC over time, so I would think Geller
should characterize such as effect better to capitalize on it.

Geller does offer the patented circuit on a stabler and more costly
version of SVR, but you have to ask him for it.  Its near $300
compared to the $35 SVR.

But I think there are more conventional ways to achieve stability
without the uncertainty of this circuit's function.

If others aren't aware, the Time Nuts archives around 2008/9 have
discussions about WarrenS's modification and his experience with
Geller boards and chips, mailed to-fro East and West coast USA, and
his experience with its stability.  There are other posts affirming
Geller and Malone's boards at least for short term stability as
transfer references, no time statement what short means, I presume at
minimum the time it takes to receive US mail from coast to coast.

Thanks a bunch again, for the leads and comments.

At 05:03 PM 8/22/2010, Charles P. Steinmetz wrote:

Best Wishes,

Marv Gozum
Philadelphia

Marv wrote:

I don't know -- he has not characterized it rigorously AFAIK, nor
compared it to an SVR.  I have on several occasions compared it to my
Fluke 732A, and would say that it seems to work fine at the tens of
ppm level for drift and probably less than that for tempco, although
that is anecdotal -- I made no attempt to separate the drift and
tempco effects.

Any proper voltage regulator circuit (not to mention voltage
reference!) does this, doesn't it?  Certainly, everything I've ever
designed does.

I have not observed any effect on tempco using this connection
(assuming equivalent tempcos of the bias circuits for the two connections).

Interesting.  I wonder why he doesn't list it on his web site?

As well as low noise.  There is nothing wrong with the concept -- at
the end of the day, it's just an ovenized reference (LM399 or
similar) with a scaling amplifier.  But the op amp and resistors in
the AD587 are not the most optimal choices, so why pay for a
reference you're not using, only to get an op amp and divider string
that aren't the best-suited to the task?  I suppose if you require
"pretty well optimized" and not "best I can do in an amateur lab," it
makes things easier.

Best regards,

Charles

On Sun, Aug 22, 2010 at 01:37:11PM -0400, Marvin E. Gozum wrote:

Unfortunately, I have it not.

Calibration of 34401A costs around 150EUR here. It's more than the cost of
the Geller's reference, but the full calibration could not be done with
just a 10V voltage source.

Anyway, could I trust that this little board, after a travel overseas,
would bring to me any precision? It has a trimmer on it. Better would have
been if he just measured the untrimmed output of the reference and noted
on paper, like the folks of HP do with their LTZ1000-based board...

But at this point he would just be selling me a measure, because the chip
itself cost around 10$ and it works without any external component...

How could the Geller's board be better than the chip over it?
If the chip itself is good enough to reach alone a stability worth to transfer
a calibration good enough for a 34401A, would be very interesting to simply
measure chips by putting them in a suitable ZIF socket, put a note on them,
and sell them as they are. No trimmer, no PCB, no resistors. If I could
claibrate a 34401A against a 10.00000V, I could calibrate without any
problem against a 10.12345V.

I am just going abroad for a fair for a couple of weeks, so I will also reply
late.

You are doing an interesting work. Interesting also would be
to buy a barebone chip and measure it untrimmed.

I just read the patent. Substantially, what he patented is the fact he's using
the heater of a voltage reference of a LM399 kind as a sensing element, and
that the circuitry is composed by operational amplifiers, resistors, etc.

I never patented it, but once, while developing a low power switching supply,
I used a common house lamp bulb as a load.
Measuring the voltage on the lamp (the switching was a current source) I was
able to determine the wind around the bulb. I am not joking.
The lamp filament was kept just glowing red and bringing away heat from the
bulb made a noticeable change in his resistance.
The simple fact that an anemometer could also be done with a lamp bulb and
a circuit comprised of operational amplifiers, I think, is not the kind of
innovation that deserve a patent. The use of a lamp, or the use of a
heater of a voltage reference, would not make actual hot wind anemometers
any better :)

I think you are right, maybe he's just making kits also for hobby.

Best regards,
Andrea Baldoni

On Tue, Aug 24, 2010 at 01:37:12PM -0400, Marv Gozum @ JHN wrote:

I agree.
In bandgap references, you use the reference-generated voltage to
regulate his current. With buried-zeners you usually do the same.
There's not any reason to use a different reference (of an inferior grade) to
regulate supply to a superior one (even if it may not do any harm, if the
superior reference has some grade of power supply variation indifference),
if not to cancel opposing thermal coefficients or maybe because it's a simple
circuit and surely autostarting... but the classical approach is cheaper.
As my lamp anemometer... you could do it, but why?

Best regards,
Andrea Baldoni

Hello Andrea,

WarrenS has just posted a dataset similar to what I am still
collecting.  It has been done!  But his reflects special selection of
the AD587 chip and a modification to it he describes in an old post
at time-nuts.

http://www.mail-archive.com/time-nuts@febo.com/msg16518.html

FWIW, I am collecting preliminary data before connecting my PC and
doing something similar to WarrenS, so I can get a feel for the
response.  I make several thousand automated measurements daily,
covering 75-80F, 35-50% Relative humidity, over the past
week.  However, I am using a 3456A and have no GPIB adapter so I
cannot download the data, so I am doing it the hard way, allowing the
HP to do statistical analysis, and recording the variance for fixed
time intervals, and recording temp and humidity manually.

My curve is similar to WarrenS but I can only resolve 10uV.  Just
quickly, my stats as I type, my standard deviation is 15uV from a
mean of 10.000 05 Vdc.  Thus, my 95% confidence interval is 10.000 00
to 10.000 10, and this includes the internal error of the HP 3456a
too.  This is a fairly consistent response over 1 week except the
mean will travel as the temp deviates from the 73F when Geller
calibrated the SVR.

It is fairly crude compared to WarrenS's results, but it has a
similar response. I cannot be sure which contributes to drifting
more, the HP 3456a, the Geller, or both!

But overall, as you see, the reference is not shabby, and now 1 week
since purchase is still reading true.

At 06:43 PM 8/24/2010, Andrea Baldoni wrote:

As WarrenS comments, a reason for getting the assembled board is a
need to burn in the assembly and the chip for hours, I forget if
Geller does 10 or 200 hours before calibration, then calibrate
it.  The SVR calibration includes both trimmed and untrimmed
calibration values, so if you think the pot has gone bad, you can
remove a jumper to detach it from the circuit, and measure the native
reference output directly.

Sincerely,

Marv Gozum
Philadelphia, PA