Little update on my two primary DMMs. They are still happy and at daily service near other ppm-friends.
Cooked some plots from my two main 3458A's. Unit "xDevs-3" is bit more noisy but more stable. Other one "xDevs-2" is about 40% less noise on DCV but drifts a bit more.
Both have modified A9 reference for lower oven temperature and calibrated externally from time to time.
Meters are always powered except few interruptions to move them to other bench. They run without fan dust filters and non-factory faster Delta fans to reduce internal TEMP? temperatures a bit.
I run collection on some (not all) constants from time to time, with ACAL ALL processed before constants capture.
Dates below are not uniformly matched, but I try to run captures every week or more often when meters are available.
Both meters were never calibrated externally, since I see no benefit in such calibrations in my case. To adjust both 3458 I use LTZ1000A-based
rugged FX reference that is periodically annually checked with U < 0.5ppm accredited standards. As resistance standard 3458s adjusted to
glorious ESI SR104 with value verified to ~0.2ppm uncertainty. Other ranges and functions are checked with help of 5720A calibrator from time to time, to confirm ACAL correct operation.
First data set on xDevs-3 unit, 800 days after sea shipment Taiwan - USASame data set but without 1kOhm data and rescaled Y axis :
My observation on this meter:
* A9 reference drift = -0.9ppm/800 day = -0.41 ppm/year linear
* ADC 10VDC CAL?72 drift = -0.657ppm/800 day = -0.3ppm/year = -0.0247ppm/month
* 1 VDC drift = +2.47 ppm/800 day = +1.127ppm/year = ~+0.1 ppm/month
* 40kOhm VHP101 drift = -1.03 ppm/800 day = -0.47ppm/year
* Significant 1kOhm gain seasonal correlation (to humidity, but I don't have humidity sensor reading at time of ACAL points
).
Longer second data set on xDevs-2 unit, 1195 days, including before sea shipment and after.Observations on 3458B:
* A9 reference drift = -2.95ppm/1195 day = -0.9 ppm/year linear
* ADC CAL?72 drift = insignificant down in noise
* 10V drift (ADC drift + A9 drift total) = 3 ppm/1195 days = 0.92ppm/year
* 40kOhm VHP101 drift = -1.4 ppm/1195 day = -0.43 ppm/year
It looks like second unit started to drift slightly at a rate 1ppm/year after 2 month sea shipment and power restore. But key point is good ADC stability, which is down in the noise and just random walk around 0.5 ppm-ish zero value. So little 1ppm/year drift can be easily accounted for or corrected by performing fresh external calibration at any time.
Also traceability to SI obtained thru two calibrations from freshly JVS-measured solid-state zener references (Datron 4910 & Fluke 732B). They are marked at June 2019 and November 2020 dates. First xDevs-3 meter was adjusted same time as well, actually.
As a bonus, INL was also verified of both meters relative to another freshly calibrated good Agilent 3458A DMM:
Still within ±0.1 ppm on -10V to +10V
Keep in mind, even thought these two meters show great stability better then the specifications they are
NOT reference standards and cannot be used as such for serious metrology work.
These plots are only useful more to judge if instruments are healthy and can be used to perform good quality voltage measurements and transfers. Questions?