[volt-nuts] Ye Olde HP3458A

[John Devereux]

John Devereux <john at devereux.me.uk> writes:

> "Poul-Henning Kamp" <phk at phk.freebsd.dk> writes:
>
>> In message <87y5z97jwe.fsf at devereux.me.uk>, John Devereux writes:
>>
>>>What are peoples opinions on the likelihood of a newer model coming out
>>>soon?
>>
>> p=0.
>>
>> Read the HP Journal article carefully, and you will realize that if
>> you want to build a DVM an order of magnitude better than the 3458
>> it will have connectors for both "power" and "helium", because such
>> a instrument will crash directly into the thermal noise.
>
>  :)
>
> I have read that article before and was suitably impressed. But it was
> not clear to me where their 100nV/rtHz "theoretical limit" came from. It
> seems arbitrary rather than fundamental - it is the noise of a 600k
> resistor. 
>
> There are plenty of opamps (and transistor front ends) with <1nV/rtHz
> noise.

(replying to myself)

Actually not so many at ~DC :)



> And plenty of resistors less than 600k:).
>
> Also while digging around I found some patents that appear to improve on
> the architecture. For example 4951053 describes a summing node switching
> scheme that seems obviously superior to that illustrated in the journal
> article. (In fact it predates it, maybe the article is not quite right?)
>
>> The only place where you could drastically improve on the Hp3458,
>> is the two internal references, and this is where much of the
>> "value add" of Flukes 8508a comes from.
>>
>> I am somewhat dubious if we will ever see compact high-temperature
>> superconductor based josephson voltage references and quantum resistance
>> standards, but if they appear on the market, it would be pretty trivial
>> to retrofit a HP3458 with them, provided they can fit in the box.
>>
>> The other argument for buying an HP3458A is that it is a much more
>> interesting instrument to play with, up to and including downloading
>> and running your own M68k programs on it.
>
> Hey that sounds fun :) I wire wrapped a homebrew 68k board around then
> too.

-- 

John Devereux