[volt-nuts] LTZ1000 project build

[Frank Stellmach]

Hello,

I monitor two DIY LTZ1000 references, standard circuit from LT 
datasheet, since about 7 years.

These are running on 45°C, and are using 5 precision wirewound resistors 
(PWW), T.C. ~0.2 .. 4ppm/K.
BOM cost are currently 42.85$ for the LTZ1000 (non A!!) directly from LT 
(recommended, beware of chinese fakes), and about 7-10$ for each resistor.
So each reference may be built for about 100$ in total.

By avoiding some pitfalls, like excessive heating, causing hysteresis 
effects, these reference will have an annual stability of typ. 
-0.8ppm/year, without any special treatment, like "ageing", slots in the 
PCB, or usage of extremely sophisticated and expensive metal foil resistors.
One only has to monitor the drift of each reference if it's having this 
typical drift rate, or not.

Andreas Jahn currently makes a great job in investigating on the overall 
T.C. performance of his LTZ references, in terms of selecting these PWW 
resistors, trimming by 400k resistor, and LTZ lead length.

I'm doing that drift monitoring by comparison of a total of 4 
references, that's an additional LTZ1000A in a 3458A, pimped to 65°C, 
and not running continuously (practically no drift), and two stacked 
SZA263 in a Fluke 5442A, which have typ. +1ppm/yr, but this instrument 
also is not powered continuously.

The differences within this group are still within +1/-2ppm after 7 years.

The absolute uncertainty (w/o traceability) is estimated to 1ppm, as 
given by the initial calibration of the 5442A in 2009, one check vs. a 
freshly calibrated 8508A from another volt-nut in 2013 (+0.5ppm 
difference), and vs. two new 34465A last year (<0.5 ppm difference on 
10V range).

So it's really less complicated and less expensive than expected, to 
build and qualify LTZ references.

"Dr. Frank"