[volt-nuts] HP 3458A and DC current measurement

[Frank Stellmach]

Tony,

obviously, there is a basic problem with DC current, already beginning 
at the definition of the SI Ampère.

This is the worst realized electrical unit, i.e. the 'mise en pratique' 
is difficult to an error level of about 1e-7 only. DCV and OHM can be 
realized to a much higher degree of stabilty, therefore the upcoming 
redefinition of the SI system will improve the electrical units by 
orders of magnitude.

In a classical DMM setup for DCI, there are two basic problem, both 
trivial, but to be reminded here:

The first one, the measurement of the Ampere is an indirect one, 
calculated from two units, i.e. a standard resistor and a standard voltage.

The second problem is the measurement setup, in series with the 
circuitry under test.
This serial path can not be interrupted, so this prevents measuring 
thermal offsets periodically, as it is the case with DCV (OCOMP ON). 
Also, it is not possible to measure gain drifts.


But anyhow, it is possible to increase stability by chosing better 
components:

There exist Zeranin (< 1ppm/K) based shunt resistors from 
Isabellenhuette or the metal foil technology of Vishay (< 2ppm/k) 
compared to the mediocre 10ppm/k in the 3458A (Manganin alloy?). They 
also have smaller  thermo electrical force.

The drift effects of T.C. and thermo electricity could be cancelled by 
heating the shunt to a constant temperature, of say 50°C, like in the 
Fluke 5440 calibrator.
Additional heat dissipation due to the current to be measured would not 
cause any temperature related drifts any more.



There are several publications from CERN, how to calibrate currents up 
to 20kA to the level of a few ppm, simply search for Fernqvist, CERN, 
current, as this one:
http://www.researchgate.net/publication/3090510_A_novel_current_calibration_system_up_to_20_kA

The 10mA current standard, built by Pickering is also very instructive:

http://cdsweb.cern.ch/record/643294/files/cer-002399331.pdf

Frank