[volt-nuts] Keysight 3458A vs Keithley 2002 8.5 digit multimeters

[Frank Stellmach]

Hello David,


Illya already made a complete and competent summary of the 3458A vs. 
2002 features. So I'd just like to emphasize that the ACAL function of 
the 3458A really makes the difference.

Basically, this might bring the 3458A back to its 24h specification for 
each mode and range at any time, provided, its internal references 
(LTZ1000 and VHP101) do not drift as much as specified.


The specifiaction parameters of all other (non-autocal) DMMs diverge 
over time, as they are not able to compensate for the drift of their 
internal resistors.

The 8508A still beats the 3458A, because it's full of Vishay hermetical 
BMF resistors (which drives its price into insane heights), and because 
it has the metrology grade 45°C LTFLU reference inside. And it's got a 
much better Ohm circuitry, also providing an OHM Transfer Accuracy 
specification, over the 3458A.

So, the 3458A could either be more stable, if you pimp its LTZ reference 
from 90 to 65°C, or if you pick an old, more stable instrument, or an 
opt. 002 instrument.

And you might want to have an instrument which is built after 1995, as 
these have the better VHP101 40k resistor (0.3ppm/K, 2ppm/6yrs.). This 
was never updated in the specs of the 3458A, anyhow.

Then, you should avoid the ser. numbers of AN18.

Regarding the DCV specification, the stability is not defined correctly. 
Are these parameters achieved only, when it's running 24/7/365? What 
happens, if you save energy AND instrument lifetime, by switching it off 
as much as you can?

If the LTZ1000 would run on 45°C, it would drift at typ. -0.8ppm/yr. 
only, and to near zero, when unpowered. Hysteresis effects would also be 
small at this oven temperature. Same goes for 65°C.

Beyond that, the LTZ1000 may show big hysteresis and would therefore be 
better off, when powered continuously, especially for opt. 002. 
Reference for that behaviour is AN18, and Pickerings hysteresis removal 
patent.

So one also should avoid too old instruments, or ones with dim display, 
as this indicates a long run-time. Many parts inside the 3458A are 
already obsolete. Especially the 3..5 fast comparators EL... very 
probably degrade over power-on-time, and cannot be replaced.

I would always prefer the 3458A over the 2002, as the latter is good for 
7 1/2 digits only, due to its worse noise, linearity and stability 
perfomance. Additionally, the 3458A AC capability is much better.

The 3458A is a great (the best) instrument for ratio, related to range 
and mode transfers.

But it's not a real "standard", therefore external 10V and 10kOhm 
references are recommended.


Frank