Just to render TiNs statement more precisely, a stable voltage reference is NOT required for the A/D hybrid test (the so called 'short-term drift').

That really stupid 0.43ppm/day criterion is simply derived from the ageing law of the LTZ1000A reference.

All such temperature induced ageing processes (Arrhenius) follow roughly a square-root-law:

dU/dt ~ 8.2ppm/year * sqrt(t[days]/365 days), with 8.2 ppm/year being the standard annual drift rate.

From that, you can calculate 1 day, 90 days, 1 yr and 2 year drift specification of the LTZ1000A alone, 1 day would be 8.2ppm * sqrt(1/365) = 0.43ppm

A good A/D hybrid has to be much more stable, as such high a drift would be a linear effect, i.e. w/o ACAL DCV it would drift 0.43ppm/day * t[days], giving about 39 ppm after 90 days.

The specification btw. contains NO explicit specification w/o using ACAL DCV.

There's an indirect hint only, that specification limits are valid within 24h of last ACAL only... very poor specification method.

And in fact, a fully working instrument will mostly return to or stay at its original reading, at the same operating temperature.

Frank