Is there a "standard" much better than a LTZ1000, but much cheaper than a Josephson Junction Array?
The question about the Josephson Junction Array got me thinking. I wonder
if there are any sort of technologies that can produce a voltage with much
better stability than the LTZ1000, but without the cost of a Josephson
Junction Array. A sort of half-way house.
BTW, I see some rather battered up LTZ1000A's on eBay,
http://www.ebay.com/itm/1x-LTZ1000ACH-Ultra-Precision-Reference-LTZ1000A-/111325506311
that have dents in them. They are more expensive than buying the chip new
from the Linear Technology website.
http://www.linear.com/purchase/LTZ1000
Yet people seem to buy them. Now I know the stability of those improve with
age, but when they have dents in them, its clear they have not been exactly
cared for. Maybe I will buy some new LTZ1000As and stick them on eBay for
$100 and see how many I sell !!
It seems the real cost of an LTZ1000A standard is not the chip, but the
resistors you need.
In message CANX10hCbbrWvtQhybFf-Pm1uaSyZ3agHLEo8p_5iyoyYR-gP8Q@mail.gmail.com
, "Dr. David Kirkby (Kirkby Microwave Ltd)" writes:
The question about the Josephson Junction Array got me thinking. I wonder
if there are any sort of technologies that can produce a voltage with much
better stability than the LTZ1000, but without the cost of a Josephson
Junction Array. A sort of half-way house.
As far as I know there are only two steps between the LTZ1000 and JJAs:
-
Specially cared for LTZ1000's (See: Fluke)
-
Lots of LTZ1000's to get sqrt(N) reductions.
--
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.
Hoi David,
On Wed, 19 Oct 2016 22:10:15 +0100
"Dr. David Kirkby (Kirkby Microwave Ltd)" drkirkby@kirkbymicrowave.co.uk wrote:
The question about the Josephson Junction Array got me thinking. I wonder
if there are any sort of technologies that can produce a voltage with much
better stability than the LTZ1000, but without the cost of a Josephson
Junction Array. A sort of half-way house.
As far as I am aware of, no. Beside the standard cells, the only other
way I am aware of that was used were voltage balances. Ie measure
the size of the electromagnetic force of a capacitor vs a calibrated
weight. Of course the capacitor would need to measured precisely as well,
which would require a capacitance standard (aka calculable capacitor).
To exert any measurable force, you'd need a couple of kV of potential,
which is not exactly nice to handle and to keep stable. And because
everything is very sensitive to the slightest disturbance, you'll have
to put everything into a vacuum chamber. Even though it does not need
to be high vacuum as for atomic clocks, it's still not something you'll
do on a kitchen table.
Maybe one way would be, if you could make use of the quadratic
relation of power dissipated over a impedance and the voltage over
the impedance, then you could work with a group of calculable capacitors
vs a watt balance and build a voltage standard like that...
Sorry it's late on this side of the globe and I'm not sure whether
this makes sense... much less whether the idea even works.
BTW, I see some rather battered up LTZ1000A's on eBay,
http://www.ebay.com/itm/1x-LTZ1000ACH-Ultra-Precision-Reference-LTZ1000A-/111325506311
that have dents in them. They are more expensive than buying the chip new
from the Linear Technology website.
http://www.linear.com/purchase/LTZ1000
Yet people seem to buy them. Now I know the stability of those improve with
age, but when they have dents in them, its clear they have not been exactly
cared for. Maybe I will buy some new LTZ1000As and stick them on eBay for
$100 and see how many I sell !!
Because people have no clue. A lot of people think that ebay = cheap,
and if it's expensive on ebay, how much more expensive is it from the
manufacturer!
It seems the real cost of an LTZ1000A standard is not the chip, but the
resistors you need.
Well, you need to calibrate them. Which you can do using a null-volt
detector and a Kelvin-Varley divider, not something very cheap either,
but at least not difficult to build yourself. I think, with todays technology
it should be possible to build a self-calibrating Kelvin-Varley divider,
even if it would be very tedious. I am not so sure whether a null-volt
detector is easy to build. I have not figured out yet, how to compare to
voltages to nV levels precision.
Attila Kinali
--
Malek's Law:
Any simple idea will be worded in the most complicated way.
On Thu, 20 Oct 2016 01:16:27 +0200
Attila Kinali attila@kinali.ch wrote:
As far as I am aware of, no. Beside the standard cells, the only other
way I am aware of that was used were voltage balances. Ie measure
the size of the electromagnetic force of a capacitor vs a calibrated
weight.
Sorry, that should have read "electrostatic force".
Attila Kinali
--
Malek's Law:
Any simple idea will be worded in the most complicated way.
Is US patent 6091281 just "whistling in the wind" or does it have some
merit?
On Oct 19, 2016 7:59 PM, "Attila Kinali" attila@kinali.ch wrote:
On Thu, 20 Oct 2016 01:16:27 +0200
Attila Kinali attila@kinali.ch wrote:
As far as I am aware of, no. Beside the standard cells, the only other
way I am aware of that was used were voltage balances. Ie measure
the size of the electromagnetic force of a capacitor vs a calibrated
weight.
Sorry, that should have read "electrostatic force".
Attila Kinali
--
Malek's Law:
Any simple idea will be worded in the most complicated way.
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/
mailman/listinfo/volt-nuts
and follow the instructions there.
On Wed, 19 Oct 2016 21:14:42 -0400
Stan Katz stan.katz.hk@gmail.com wrote:
Is US patent 6091281 just "whistling in the wind" or does it have some
merit?
They do use the EFC voltage dependence of the frequency of an TCVCXO to
define a voltage standard. Yes, this can be done, but the variability
of the EFC voltage due to temperature and aging is at least in the low
ppm range. If one would use an OCXO instead, one could go below 1 ppm
temperature dependence, but I am not sure how good the aging behaviour
would be. Unfortunately, I couldn't find any information on the aging
characteristics of varactors.
That said, this would be still only a secondary standard and would need
calibration against a primary. It would also be plagued by all the problems
OCXOs have, like retrace, vibration/shock sensitivity, etc.
So, IMHO it's a nice idea but probably not that much better than an
LTZ1000 and quite a bit more expensive (needs a good frequency reference
to measure the TCXO/OCXO)
Attila Kinali
--
Malek's Law:
Any simple idea will be worded in the most complicated way.
On 20 October 2016 at 00:08, Poul-Henning Kamp phk@phk.freebsd.dk wrote:
In message <CANX10hCbbrWvtQhybFf-Pm1uaSyZ3agHLEo8p_5iyoyYR-
gP8Q@mail.gmail.com>
, "Dr. David Kirkby (Kirkby Microwave Ltd)" writes:
The question about the Josephson Junction Array got me thinking. I wonder
if there are any sort of technologies that can produce a voltage with much
better stability than the LTZ1000, but without the cost of a Josephson
Junction Array. A sort of half-way house.
As far as I know there are only two steps between the LTZ1000 and JJAs:
- Specially cared for LTZ1000's (See: Fluke)
Do you have a specific reference? A google found lots of kits on Fluke and
LTZ1000, but none of the pages were from Fluke.
- Lots of LTZ1000's to get sqrt(N) reductions.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
But that assumes the drifts are completely uncorrelated, which I doubt is
the case - especially if they are all from the same batch. But there are a
limited number of resistors of the quality needed (Vishay is the only one I
can think of), and PCB materials are likely to be the same. It seems to me
that you would not gain sqrt(N) unless N is quite small. How small, is
anyone's guess.
Dave
In message CANX10hCDAXkzLY__JP_pY9tEpSaP-RorCtNAPvuTUbv-OB+XjA@mail.gmail.com
, "Dr. David Kirkby (Kirkby Microwave Ltd)" writes:
- Specially cared for LTZ1000's (See: Fluke)
Do you have a specific reference? A google found lots of kits on Fluke and
LTZ1000, but none of the pages were from Fluke.
They get their own feed of devices from Linear. Nobody outside Linear
and Fluke knows what is special about them.
- Lots of LTZ1000's to get sqrt(N) reductions.
It seems to me
that you would not gain sqrt(N) unless N is quite small. How small, is
anyone's guess.
N=4 is commonly used.
--
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.