[volt-nuts] Resistance standard

[Rob Klein]

Bill,

Thanks for your comments, they are appreciated.

WB6BNQ schreef:
> Rob,
>
> Fluke has over 60 years of real experience correctly make highly accurate  stable
> resistors.  Actually they figured it out in the very beginning of their
> business.  Without a serious metallurgic background, plus a few more disciplines,
> I doubt seriously you would even come close.  However, there is nothing wrong
> with trying.
>   
But I am not going to try to actually make a resistor, I 'merely' want 
to build several resistance standards, out of
commercially available ones. And I certainly do not have any intention 
of beating Fluke at their own game.
> First off, what is your purpose for such accuracy ?  What do you intend to do
> with these ?
>   
Purpose? This is the volt-nuts group, isn't it? :-)
I'm a self-employed electronics engineer with 'a thing' for precision 
measurements. I have, over the years, built
up a nice collection of multimeters, calibrators, etc. and a resistance 
standard is going to make a welcome
addition to that.
> I see a number of problems that need to be addressed.  First, it does no good to
> have the highest possible quality resistor if you do not have the proper means to
> compare it to other items.  To maintain that level of quality, for that given
> resistor, you would need to duplicate its’ measurement environment precisely,
> i.e., temperature, humidity, etc.
>   
While my workshop isn't a cal-lab, it is nonetheless pretty stable, 
environmentally. It has to be, in order to keep
my pick-and-place machine happy.
As I'm not aiming for sub-ppm precision, it will more than likely suffice.
> Having a resistor, in of itself, serves no purpose without the proper equipment
> to compare it against other items or to use it in a measurement process.  Items
> like a highly accurate voltage/current source and a very good null meter (Fluke
> 845A/B) are but a few things needed.
>   
Let's see:
- Fluke 720 KVD, check
- Fluke 845, check
- Various high stability voltage sources, check
Add a recently aquired lead compensator (bought "as is" and not yet 
tested, so maybe not) and I think I'm
reasonably well set-up.
> Nonetheless, some of your intended construction ideas need to be reviewed.  You
> talk about wanting to equal or beat Fluke, yet you only intend to buy 0.01%
> instead of 0.001% resistors.  What’s up with that ? 
The VHP202's I'm buying *are* 0.001%, as I stated.
>  Those are not considered standards by any means.  
Really? Then kindly explain how the metrological world was able to 
succesfully maintain the Volt for decades,
using standards that deviated almost 2%.
As long as the value is known and stable to within the required 
uncertainty limits, I'd almost say that anything
could be considered a "standard".
> Just series/paralleling a bunch of resistors is not
> going to help you unless you know, precisely, the temperature coefficients are
> for each resistor.  In order to play that game you would need to buy a whole lot
> (like thousands) of them, not just a couple.  Then you spend an inordinate amount
> of time testing the temperature coefficient for each and then mixing and matching
> in trying to achieve a zero temperature coefficient.  Or at the least the
> smallest variation of resistance verse temperature change.
>   
The RTC of Vishay's Z-foil resistors is as low as it gets, not by 
selection, but by design! The series/paralleling
is mainly to increase the long term stability
> On the one hand you talk about buying expensive resistors then decide to get the
> non-hermetically sealed models because they are cheaper.  
I intend to use *both*, so that I can build several different types, in 
order to compare their long term behaviour.
It will be interesting to see just how good these non-hermetic types 
behave over time.
> Using an oil filled container is not going to help if the resistor is not hermetically sealed.  
But the point is to use an oil filled *hermetically sealed* container to 
hold the non-sealed resistors. Fluke did this for the
main resistors of the first decade of their 720 KVD, Tegam, formerly 
ESI, do it for their SR104, widely considered to be one
of the best, if not the best, standard resistor there is. So it appears 
this idea might not be as daft as you make it seem.
> The
> same goes for using any other kind of fill material.  It is going to impact the
> non-hermetically sealed resistor and its going to impact the leakage.
>   
How, exactly, is it going to impact the resistor?
> The biggest
> point for leakage is the connection post insulation material; most are junk !
>   
Hmm, this got me curious enough to dig out my insulation resistance 
tester. Unfortunately it seems the batteries
are flat :-( To be continued ...
> The second biggest point will be dirt between the connection posts.
>   
I am well aware of the neccesity of keeping a clean shop.
> However, it is not so simple as just pouring in some oil.  Oil has a multitude of
> properties and the selection is not going to be an easy one.  
Eh, yeah, that's one of the main reasons I came here for advice :-)
> You talk of baking the resistor/can combination.  What are the limits of the
> resistors relative to heat ?  Heating it to 85c is going to change its
> characteristics and would require you to re-test the resistor all over again.
>   
I have already 'pre-cooked' the first batch of 9 Z201's for three weeks 
at around 60 °C and found a
downward shift of maybe a few ppm, as compared to a 1kOhm reference; a 
20 year old VHA412, from Vishay.
The TCR of this older resistor is, at around 2 ppm/°C, a lot higher than 
that of the resistors under test and
may have had a considerable influence on the measurement.

I have not been able to find any references to changes in TCR, caused by 
temperature cycling, in these devices.

The baking may also help relieving some of the stresses that might built 
up during soldering.
> What about that low emf connector ?   You need to consider what that material
> (insulation) will do when you “bake” it at 85c.  Most likely will ruin it.
>   
The binding posts are rated at 115 °C max. operating temperature and are 
not going to be baked anyway.


Best regards,
Rob.