Although that's a recommendation if you follow standard procedure from NIST and other metrological sources [every 1-2 years] what folks at volt-nuts have found experimentally is cheapo standards may hold calibration for much longer periods of time.
If you have an older Fluke 80s series, or older reference DVM like the HP3456a, its surprising these 10-30 year old rarely need adjustment, a formal calibration will simply reveal if they are in spec; assuming you measure them in a controlled environment.
The actual research papers, and this is a reason to join the IEEE if you are inclined, describe in detail the factors that cause calibrations to go off that you can correct for, however, these correction factors are never part of routine calibration procedures given to front line folks in industry.
If you were to take voltagestandard.com board or a Geller SVR board as is, without added procedures, you would send it back to them for recal periodically.
I've reconfirmed experimentally what research papers show, is the absolute ratings down to uV: 10.000 000 will vary with seasons due to ambient temperature, humidity and less so atmospheric pressure: so if you make a cal measurement that is not as precise as the day it was originally calibrated, it will be off in uVs, making you think its time for a recal.
You can control ambient temp and humidity in a lab easily, but its difficult to control for atmospheric pressure changes. You will have to put the voltage reference is a pressure controlled pot, with the internal pressure adjusted to maintain 1 ATA against the environmental swings, something not easy to make and I don't have. So, I log measurements at the same weather mBar my original measurements were made, controlling only temp and humidity in the lab.
All microvolt measurements assumes you know how to eliminate and control uV sources such as thermal, galvanic, triboelectric, ionic and static sources of voltage. Simple things like using a copper test clip on a stainless steel measurement lug will create a galvanic source independent of other issues. If test leads are not as the same ambient temperature as the DVM leads or the reference leads, you'll create thermal effects; if you touch the leads during measurement, it will create tribolectric or statics charges into the leads.
If you collect daily data on the reference, you can watch the uV levels swing in sync with the weather very linearly, you can actually predict how off it will be as a combo of temp, pressure and humidity.
Once you control for the environment and uV sources, the final source of drift is internal to the reference itself; if you let the unit age for 200 hrs+ before you send it out as a reference, it minimizes internal drift, but experimentally only references 1+ years old are stable will ne'er vary, except for those caused by environmental changes.
My solution was to buy 4 HP3456a: one 3456a is calibrate against a NIST standard, then calibrate the others against it as specific intervals using voltage and resistance sources that are accurate to ppm. These measurements are then spot checked against the Geller SVR. Each calibration was done at the start of each season for each DVM, but with the environmental conditions controlled, so one 3456a has been calibrated during the summer, fall, winter and spring.
The Geller SVR measurements are made to similar ambient temperature, RH and atmospheric pressure, and when rechecked for voltage, the same environmental conditions must be identical, before I log the reading.
9 months later, the SVR is still reading 10. 000 000 measured at the same conditions, although it has swung as high as 10.000 030V with weather variations. The 4 3456a show no variations with weather, if the original calibration was done as precisely the same ambient conditions. As these DVM are all over 15 years old, their internal references are aged, and should eliminate their internal drift; however, skeptical as I am, I am still tracking their responses controlling for weather, and just short of a year the internal reference is stable as well as the support electronics of the 3456a.
of course in theory you need to have your "reference" calibrated at least once every two years, or buy another one if it's cheaper 
Home
Help
Search
Login
Register
