[volt-nuts] 732A vs. 3458A stability

[Frank Stellmach]

Hello Randy,

your stability measurements of 732A versus 3458A depend on the 
integration time.
It is not possible that way to distinguish, which of both instruments 
contribute more noise.

For NPLC 100, you may expect around 0.1ppm noise, i.e 1µV/10V, at NPLC 
1000, maybe 0.01ppm.
(mains frequency for NPLC determination normally is measured 
automatically in the 3458A)

For noise figures, see DCV datasheet of 3458A, or have a look at real 
measurements in the legendary LTZ1000 blog here, should be similar what 
you may find with the 732A:

http://www.eevblog.com/forum/projects/ultra-precision-reference-ltz1000/180/
http://www.eevblog.com/forum/projects/ultra-precision-reference-ltz1000/270/

The 732A may also be less noisy, if you measure its outputs running on 
batteries only, and disconnect the mains.


There is no reasonable explanation for the observed drift of the 1V and 
1.018V outputs.

These voltages are derived by simple resistive dividers, and any leakage 
currents should influence the measurement instantaneously and 
statically. And both outputs should behave exactly the same, looking at 
the schematic.

If the 10V range of the 3458A is really stable, i.e. 1µV noise at 10V 
from the 732A, then simply measure the 732As 1V outputs on this same 10V 
range on the 3458A, which will still resolve 100nV.
Maybe, these 1V will also be stable in this configuration, and then your 
3458A has some trouble with its 1V range.

Then, you may check the guard function, if there's a difference with the 
switch set to "open", or to "Low", and connecting guard to the 732A or not .
Maybe you have some odd ground loops in your cabling.

Well, maybe thermoelectric voltages may be a problem in the 1V ranges.. 
hope you use good connections.. blank copper cables may be oxydized, for 
example, giving 500µV/°C vs. copper.

Maybe the 1.000V jack is not ok internally? Some degradation or bad 
solder joint somewhere?

Does this drift stabilize somewhere after some time?

Frank