On precision resistors

[IconicPCB]

I would like to get an idea of the performance / cost relationship of high end metal film resistors in current use  both in reference source building and other applications.
What kind of construction is available, pros and cons.

[cellularmitosis]

Here are my notes: https://docs.google.com/spreadsheets/d/1ehaeZ3XRQLUmOXZ1fhGlg-V5EXQ1VFFxHjfW145CqyA/edit?usp=sharing

edit: attached a screenshot of the summary

[Edwin G. Pettis]

What you are asking depends a lot on what you want to accomplish in your design, there is no cut and dried answer.  If you want lowest noise, precision wire wound resistors are the lowest, they are also comprobable in stability, TCR and matching TCR.  For price performance, it is hard to beat PWW resistors for best specs, only hermetic sealed resistors can be better at a very high price point.  In order to give a more specific answer, you need to provide some specifics of what you want to accomplish.

[dl1640]

My specific requirement is to measure up to 30A dc.
Consider to use a shunt resistor but seems difficult because I need 0.01% uncertainty, at least.

[cellularmitosis]

Vishay and Caddock make precision low-value resistors which can handle lots of current.

[quarks]

for resistor values below 1 Ohm I like BURSTER Serie 1240 very much

[dl1640]

Quarks

Thank you, I found BURSTER and its affiliate in my area.
I will reach to them for detailed and also need to find a way to calibrate these precision ones.
Then apply them to my measurement.

Thanks 

[Andreas]

edit: attached a screenshot of the summary

Hello,

good survey.
But for the T.C. I would not list the "typical values" as they might lead to false expections.

The typical values are usually the average of a large production batch.
So individual resistors migth have much larger values.

E.g. for the Z201 resistor the typical value is 0.2 ppm/K.
The "spread" is 0.6 ppm/K.
So according to data sheet it is the sum of both = 0.8 ppm/K max.
My measurements over 10-40 deg C indicate most of them being
between 0.2 - 0.8 ppm/K with some outliers slightly above the 0.8 ppm/K.
Only one resistor has near the typical 0.05 ppm/K which is claimed for a 0-60 deg C temperature range.

So a "max" spec of 3 or 5 ppm/K of a PWW resistor which practically turns out being typical between 1-2 ppm/K
is not much worse than the typical spec of 0.2 ppm/K.

From the construction side (especially strain relief of the wires) there may be of course differences.
Here the S102/Z201 resistors have a good construction.  (wire connected on the other side of the resistor).
And also the metal cap of metal film resistors has some advantages.

In any case you should avoid stress + excess heat on the leads of precision resistors.

With best regards

Andreas

[dl1640]

One problem is if I buy a new resistor the factory normally calibrate it with a relatively small current (low power dissipation).

However the resistor would be used with a relatively large current then the factory calibrated value would not be practical.
Do I ask factory to calibrate with my specific current value?

[alm]

A current sense resistor should be calibrated at its full scale current, otherwise you are just extrapolating based on assumptions about the temperature coefficient, rather than calibrating. Page 7 of this document makes that same point supported by data.

[Kalvin]

Why bother calibrating the sense resistor instead of calibrating the measurement using known test currents? If the sense resistor is stable (low tempco etc.), the calibration of the measurements will produce consistent results even if the resistor may not be completely accurate in the first place.

[dl1640]

Kalvin

I need to calibrate current source of 10A and 30A dc.
So I will need a shunt resistor calibrated at those particular points.

The accuracy for the current source is about 0.03%/0.05% for 10A/30A.
For this, it is better to keep the measurement system (shunt + dmm) uncertainty below 0.01%.
The shunt may not need to be 0.01% accurate to its nominal value however it must be calibrated to less than 0.01% uncertainty.
Not only to find a proper shunt but the calibration of it is important, either by factory or external cal service.

A current sense resistor should be calibrated at its full scale current, otherwise you are just extrapolating based on assumptions about the temperature coefficient, rather than calibrating. Page 7 of this document makes that same point supported by data.

alm

Thank you, I will read this paper later.

So far, I do not get a reply from Burster if they could calibrate the shunt by 10A and 30A dc.

[dl1640]

short update:

local burster rep quote me about 400-euro for each 1240-0.0005 or 1240-0.001

?••?!

and they seem not to calibrate at max/rated current, in this case 1240-0.001 is specified for 30A in air but factory calibrate it with 9A, and i would use it under 10A..