[time-nuts] VRefs - I'm a bigger Voltnut than a timenut..

wje wje at quackers.net
Sun Nov 30 15:35:47 UTC 2008


Another (sometimes) relatively inexpensive outstanding reference is 
either a Datron 4910 or 4912. These are also LTZ1000 based. The 10 has 
one reference, the 12 has four that can be averaged or used 
independently. You can on rare occasions pick one up for a few hundred. 
I have a pair, and yes, I paid Fluke $600 for traceable certification 
for one.

These use a rather clever way of generating 1V and 1.18 volts without 
significant division error. The 10V reference is sampled by a 
pulse-width modulator whose duty cycle is crystal-controlled and 
programmable. The result is then LP filtered using an active filter and 
tracks the primary reference to better than 1ppm.

My primary one was hot-shipped from Fluke after certification (it has 
battery backup) and hasn't ever been turned off. I have turned the 
secondary one on and off a few times, and the resultant error is less 
than 0.3 ppm. Both of these units had thousands of hours of burn-in 
before I got them, and I've added another 30 thousand or so.

I also have a Solartron 7081 8.5 digit meter, which is a marvelous unit. 
It uses a selected ultra-low TC Zener with a transistor for temp 
compensation, a sort-of discrete implementation of the LTZ1000. The 
Zener current is programmable, and is used to set the zero point for the 
particular zener/transistor pair. I can turn mine off for months at a 
time, turn it back on, and after a 24 h warmup, usually have it within 
1ppm(!) of my primary ref. These are somewhat rare, but can be picked up 
for a few hundreds. I paid $300 for mine.

As for why, I'm surprised that no one has stated the obvious reason - 
because you can. It's the same reason time-nuts collect 10e-13 frequency 
standards. I was into precision voltage and resistance metrology well 
before I got into precision frequency. I find it fascinating to be able 
to determine various values to precisions that not all that long ago 
were limited to national standards labs.

Bill Ezell
----------
They said 'Windows or better'
so I used Linux.



Dr. Frank Stellmach wrote:
> Hello,
>
> there's currently no easy way to convert a frequency into Volt, i.e. 
> other than by cryo Josephson junction array or realizing Ohm by a 
> Klitzing quantum hall effect device. Otherwise, FLUKE would have 
> implemented that already in their Calibrators/Standards. They still rely 
> on Artefact Calibration, i.e. ovenized longterm stable Zener References 
> and Reference Resistors.
>
> The 5720A uses a stacked double Zener reference on a proprietary hybrid 
> circuitry, giving 1-2 ppm/year stability, the 7001 and 732B use the 
> LTZ1000 (Zener above BE junction), as the 3458A, the 732A uses the 
> Motorola device (Zener below BE junction), and the elder 332D and 335D 
> use similar, but less longterm stable references.
> The LM199, AD587 and I think also the Geller device may only be used as 
> transfer standards, having poor longterm stabilities of 20ppm/yr. or more.
>
> Therefore, the LTZ1000 only may serve as a really affordable standard, 
> i.e. below 8ppm longterm stability, for private use.
>
> I was lucky to get some LTZ1000 samples, and used the standard circuit 
> from the LT datasheet for creating a small PCB. I purchased commercially 
> available wire wound resistors (TK 3, 25ppm/yr) for the 5 necessary 
> reference resistors and two OP07. Those resistors should give less than 
> TK 1 and 1ppm/yr additional drifts.
>  
> Compared to my HP34401A, the circuitry showed quite good short term 
> stability below 1ppm, but shielding still has to be improved, as the 
> LTZ1000 circuit is susceptible to external EMC disturbances. This Volt 
> reference then costs less than 100 Euros/Dollars in total.
>
> I designed further circuitry (with TK 0.3 and TK 0.1 precision metal 
> foil resistors from Alpha and Vishay, chopper amp, precision switches 
> and Teflon insulated wires) to have the possibility for precision 
> transfers of the 7,2V reference voltage to 100mV, 1V, 10V, 100V, 1000V 
> level.
> Only then, a Volt Standard is complete.
>
> All this stuff will emulate or replace the commonly needed 732A (ref), 
> 720A (Kelvin Varley), 752A (Reference Divider) and 834B (Null Detector).
>
> Got no time yet to finish and improve the system, and to do the repair 
> on an old 332B standard, but would like to share my knowledge, if 
> anyone's interested.
>
> Frank Stellmach
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>
>
>   




More information about the Time-nuts_lists.febo.com mailing list