Reply about the TS

Hello Bob,

Im experimenting for some years with different references.
For a traveling reference with 10ppm I would use a heated reference like a LM399.
You will find it in most of the serious 6,5 Digit Multimeters like HP34401A or Keithley 2000.

I use them battery powered (NIMH-Cells) with a voltage stabilizer.
(the heater voltage has some influence on the output voltage).

Time stability can be determined by building several of them pre-ageing for some 1000 hours
and then sorting out the extremes.
Of my 2 LM399s one is running pretty close to 2 LTZ1000A references
the other (a desoldered part from a old measurement card) is drifting around by ±25uV.

For unheated references like a MAX6250 you will have to determine the temperature
very exactly. I use NTCs mounted directly at the ground pin of the Reference to get
nearly the temperature of the reference with a resolution of 0.1 K.
When selecting some 3rd order compensated LT1027CCN8-5 references
you will be able to find some with a low TC near room temperature.

But all plastic references suffer from 2 things:

first: mechanical stress will shift output voltage.
So I solder only one pin directly to the pcb the others with thin wires
So for a traveling reference you will run into trouble.

2nd: when having calibrated the temperature out humidity will play a role.
On my LT1027 the humidity change is around 0.4ppm for each per cent of relative humidity.
so for a change of 30% you will not get your 10ppm level.
The problem is that the humidity shift has a long time constant. Which is around 4 or 7 days
with my 2 LT1027 references. So you will have to record the history.

With best regards