In my opinion, the 0.43 ppm/day drift rate is a rather weak criterion, because a good ADC definitely drifts less 0.1 ppm/day. The SN-18 unfortunately is a bit superficial on this issue. As was KS, when I nagged them about the issue.
The problem is, that in order to properly determine a drift you need
- a failry stable lab, so you don't see any 5 or 10 K excursions, have humidity going on a rampage when the weather turns and so on
- to to have to instrument turned on 24/7 for at least a month or two
- to continuously monitor both temperature and the calibration constants
- apply a 2d fit to determine both tempco and drift
Using the method suggested by HP/Agilent to determine the drift reduces the problem to taking two samples 7 days apart, without proper temperature monitoring, therefore the error margins are fairly high. A 1 K drift is typicall deemed acceptable, and T_cal ± 5K is the normal working range. On the other hand continuously logging will take the instrument out of service, which might not be acceptable. From the style of writing I would also assume, that HP did not anticipate, that all users are capable or willing to properly extract the drift of their instrument. So my guess would be, that 0.43 ppm/d drift is a dying DMM slapping in your face and not a marginal situation, but it can be easily measured, hence not many false positives.
This 0.43ppm/d criterion is just the interpolation of the timely drift specification of about 8ppm/year, down to 24h.
If you apply a square root function for the ageing of the LTZ1000A reference: dU/U
ref ~ 8.2ppm * SQRT(t[h] / 8760 hr), then for 24h you get 0.43ppm drift, 90days would be 4.1ppm, and 2 years 11.6ppm, about the values given in the specification.
There's no official specification for the short term stability of this instrument, neither for the T.C. of the A/D converter w/o ACAL.
Therefore no wonder that KS is close-lipped about these (internal quality) criterion.
'Normal' or 'healthy' values are a vanishing timely drift for CAL?72, practically meaning you can't barely distinguish it from fluctuations, or T.C. effects.
And a T.C. of about +/- 0.35 ppm/K for U180, and +/- 0.5ppm/K in total, i.e. ~ +/- 0.15ppm/K for the LTZ1000A, regarding these T.C.s, see specification.
My 3458A behaves exactly like that, therefore an absolutely stable room temperature dT << 1K is required.
Frank