question about short term drift test of Agilent 3458A

[analogRF]

I want to do the short term drift test (as per service note 18A) using the CAL? 72 but I cannot keep the unit running 24/7 for 7-10 days.
I can run it for like 5-6 hours per day.
Is it ok to run it say 6 hours a day and get the temp and cal?72 after an ACAL at the end of each day and then turn it off and repeat the next day to
gather enough data? or is it mandatory to run the unit 24/7?

[EC8010]

Unfortunately, 24/7 is required. Many instruments drift because of temperature. But thermal time constants are often very long, so kit can take considerable time to achieve stability. I've found my Fluke 341A needs six hours of warm-up to achieve its best stability. Turning it off after those six hours just sends it back to its starting point. I've a test running at the moment with an ADR1000 voltage reference being monitored. It's been powered for almost two days now and the general trend of its voltage has been gently falling all the time.

[Kleinstein]

Turning the meter off for the nights is not ideal and adds some uncertainty to the test. It can still work, but may need more time for a useful result. A longer time interval to abserve the drift can compensate for the extra warm up related effect. Ideally one would compare data at the same time after turn on (e.g. 2 or 3 hours).  The test data will show how stable the CAL72 data are. If there is large drift it would show relatively fast. To be sure that one has a good unit it may take more time (e.g. 20 days).

[Dr. Frank]

I want to do the short term drift test (as per service note 18A) using the CAL? 72 but I cannot keep the unit running 24/7 for 7-10 days.
I can run it for like 5-6 hours per day.
Is it ok to run it say 6 hours a day and get the temp and cal?72 after an ACAL at the end of each day and then turn it off and repeat the next day to
gather enough data? or is it mandatory to run the unit 24/7?

The "short Term Drift" issue relates to a possible U180 fault, i.e. an excessive drift of its internal gain and A/D conversion resistors.
The test method in SN18A is misleading, because it misses the temperature dependence of these resistors.
Hence, the CAL? 72 DC gain parameter as well depends on the T.C. of U180, which is up to 0.5ppm/K, i.e. on the same order of magnitude for a 1°C room temperature or TEMP? change.

Therefore, your method is possible (read CAL?72 and CAL? 175), and even better to give consistent results.
Anyhow, you need to roughly determine T.C. of U180 first, which might be quick 'n dirty achieved by repeatedly doing ACALs and reading both parameters (72, 175) directly afterwards,  starting from "cold", until 2h later, when the 3458A is finally warmed up, i.e. every 15mins, or better every 3°C internal temperature rise, monitored by TEMP?.

A proper conclusion, how good or bad your U180 is, might take many months or even years, simply monitoring CAL?72 and CAL? 175 for each normal use after an ACAL.
The better your U180 is, the longer it will take.

I have described the background and a better way of measurement in those two threads here:
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg5230803/#msg5230803
Frank