In our lab, we have had sometimes strange behaving on 432A+478A-H76 mount. We measuring Vcomp and Vref voltages backside of the 432A using 3458A. Normaly when mount zeroing is fine etc, 3458A should show around 85mV when power meter DUT Ref-Out power is ON. Time after time even zeroing is perfect (less than +/- 200uV) and turn
Ref-Out ON voltage staying 82-83mV and it meas that Ref-Out is out of spec. Problem can disappear itself and all working perfectly long time. Same issue has happened with another 432A meter, mount, cables etc.
All used equipments are annualy re-calibrated. What can cause this strange behaving?
This is an old question which may have already been resolved, this delayed input is for anyone else that may run into this problem.
[The Standard] It's very difficult to isolate an issue like this without a long history of Standards Lab measurements of the thermistor used for calibration. In the past, our experience has been that when the 478-H76 has been calibrated at a higher level standards lab such as NPL, we notice a slight change over time. Sometimes the reading goes up, sometimes down. This is not to say the unit is changing value, but the measurement conditions can change. Even in a tightly controlled environment, the reading can vary. So without specific information such as the last three years of measurement values, serial number of the unit, where calibrated and the uncertainty in that measurement, the lab that finds the "issue" environment and uncertainty in the measurement conditions, the changes in the measurement cannot be verified initially. When a reading varies from any given time in the measurement environment, all parameters affecting that measurement must be analyzed. This includes cables, environmental thermal changes, the placement of the thermistor as it is "warming up" (stabilizing), the connection method to the appropriate test points, voltmeter stability, how long the unit is allowed to rise to the reading in question and the time it is allowed to "cool off", (referring to cooling off; this is a thermal device that changes as it has power applied and power removed. Temperature change and the rate it changes is a major factor in the H76 measurement.) And of course, the 432 stability can be a major player in the measurements as it can also change, particularly if it is not used often and the silver on the switch contacts tarnish with age. One other factor that we use, a shielded screen room.
[The Device Under Test] To determine where a measurement "varies" in subsequent measurements, all of the above needs to also apply to the DUT as well. The DUT has to be in good condition, not subjected to thermal (power) overload, dropped, have dirty connections, and a history of stable measurements. Two of the primary causes of thermistor instability is excess power applied at some point or dropped. Once either of these happen, the thermistor will tend to drift as a "normal" power is applied, even if the zero is "good". Another rare condition is that if the "balance adjustments" in the thermal converter had to be changed any time after initial adjustment, it could indicate the aforementioned issue.
[Resolve] To ensure that your measurement system is proper and accurate, you need to vary more than one element. Substitution of another DUT is the ideal way to test your measurement system. In many cases, when we suspect a DUT or system, we use a second comparison device (typically calibrated at NPL) to differentiate the system as the source of error, or the DUT.
Hope this helps,
Gary