[volt-nuts] "averaging reference" / "Perfect Volt"

[Frank Stellmach]

Joel,

your requirement about a "Perfect Volt" is very diffuse.
What shall that be?
"Perfect" in terms of tolerance, stability over temperature, stability 
over time?

REF102 is far from either of these parameters, i.e. 1000ppm tolerance, 
2.5ppm/K, and in best case 5ppm/1000hr, that may be 40ppm/year only.

Later you mention 20ppm "longterm" (??) stability..

1000ppm tolerance are 50 times worse than these 20ppm/year (?) stability 
you tend to achieve.. and that REF102 is NOT easy to calibrate .. you 
need a precise DVM for that, anyhow .. or how many hundreds of REF102 do 
you want to average to reach reasonable statistical tolerance and stability?

There may be also better references with about 0.01% = 100ppm tolerance, 
but that's probably still not sufficient, compared to these 20ppm 
"longterm"..

Perhaps you first clearly define your requirements in those three 
technical parameters given above..

Then you will for sure recognize, that you will only achieve that goal 
with buried zener references, like LM399 or LTZ1000, stable external 
components and trimming the output according to a standard. On EEVBLOG, 
you may find two good and long threads about building your own ones.

Everything else is far away from being "The Perfect Volt", or you may 
buy a JVS.

Frank