Callibration Resistor Standard

[skipjackrc4]

Does anyone here have an experience with resistance standards?  I've decided to try to get one to calibrate my micro-ohm meter, but new ones are a bit out of my price range...

I found some old Soviet ones on eBay for a reasonable price here: http://www.ebay.com/itm/0-1Ohm-100mOhm-0-01-Resistor-Standard-Resistance-an-g-ESI-GR-GenRad-/230775231142?pt=LH_DefaultDomain_0&hash=item35bb4692a6

These are old and are shipping from Ukraine to the US.  Would the accuracy of these things be damaged in shipping or by aging?  They will still be much better than a 1% current shunt resistor, but do you think it would be worth the price?

[nukie]

How do you know the used Soviet unit has not been overloaded?

Vishay Z foil if price is right.

[notsob]

How about 1 of these - 3 tested 0.1% resistors & you can check you DMM as well

http://www.voltagestandard.com/DMMCheck.html

[robrenz]

Some questions that will affect what you may need:

Is your micro ohm meter 4 wire Kelvin connection?
Does it have current reversal or offset compensated ohms to minimize thermal emf errors?
What are the ranges and resolutions and what is the current drive level for each range?
Are the leads shielded and does it have a stated CMRR for the sense leads
What is the meters stated accuracy?

[skipjackrc4]

It is not a commercial meter, but one I am designing.  The current source is 100mA, controlled by a buried zener voltage reference.  I'm not too worried about the source, as the components are very tight tolerance.  I am putting a precision instrumentation amp to multiply the sense voltage by x1, x10, or x100 to increase the sensitivity, but this is the part that worries me.  I have no accurate way to check that the multiplication is accurate.  For a $30 IC, it had better be, but I'd like to know. 

There are 0.01% resistors in 100ohm and up (I'm using one in the current source) but this is too high for my design to measure.  Once you start getting into the fractional ohm range, low tolerance resistors get very expensive.

The eBay seller claims that the standards are tested working, but does not mention being checked for cal.  His eBay reviews are positive, and one reviewer posted that the resistor was within 0.005% tolerance.  So they seem legitimate.  I just don't know much about these things.

The leads are 4 wire and shielded.  There is a current reversal switch.  The meter will be a bench Fluke model, I'm not sure of the model number as it is not in front of me right now.  I'm not expecting 0.1% accuracy on a meter that costs $100 to build.  But I would like to be able to check everything.

[Harvs]

Sorry OT:

I am putting a precision instrumentation amp to multiply the sense voltage by x1, x10, or x100 to increase the sensitivity, but this is the part that worries me.  I have no accurate way to check that the multiplication is accurate.  For a $30 IC, it had better be, but I'd like to know.

Wow, what part are you using that costs $30?

I'm currently doing a similarish thing conditioning the current shunt voltage in an eLoad prior to ADC, and an auto-zero op-amp + analog mux auto-ranging and resistors will be < $6 (though I will be correcting for resistor mismatch in software.)

If you're just trying to check the amp's acuracy, surely that's going to be easier by just input a stable reference voltage, and measuring the input and output?

[skipjackrc4]

I don't remember the part number, and I don't have it with me, but it's a TI (or AD, don't remember) InAmp with pin-selected fixed gain of 10, 100, and 1000.  0.05% tolerance for the gain at x1000, lower for smaller gains.  I'm keeping an open mind about using something cheaper, though    I'll continue that search later tonight.

I'm using a TI Ref102 10V, 0.05% voltage reference with a 100-ohm 0.01% foil resistor in the current source.  Aging specs for the reference aren't too bad and the tempco is 2.5ppm.  It will be operated only for short intervals without heating up much.

[grenert]

Texas Components will make Vishay bulk foil precision low tempco resistors to whatever specifications you need (within the limitations of their technology, I think they go down to about 0.5R):
http://www.texascomponents.com/

This set of DMM calibration resistors includes 1R and 10R ones:
http://www.precisionresistor.com/product.php?productid=144&cat=507&page=1

[skipjackrc4]

Grenert, those links look good.  I'll give them a call tomorrow.  That DMM calibration set would be good to have on hand anyway.

[robrenz]

If you are using a 8846A or 8845A I have a suggestion that will probably give better accuracy and cost less.
Make a low precision 1 amp source and spend all your money on a 0.1 ohm precision high wattage 4 wire shunt with low TCR. Pass the one amp thru the shunt and and DUT with the source leads and use the shunt voltage drop as the reference voltage into the fluke in RATIO mode. Then input the sense voltage from the DUT sense leads into the input voltage of RATIO mode.  Ratio mode will do the ratio of input V to reference V to 1ppm.  That means you could theoretically resolve 0.1µ ohm with this setup.  The beauty of this setup is the current  does not have to be precise or even stable. You are making a direct ratiometric comparison to the precision shunt. Your error will be the ratio mode error plus the shunt error.  This way you get -140 db of CMMR on the sense leads of the meter and Increased accuracy of being able to take readings at 100 NPLC for increased stability and effective resolution. You still need a current reversal switch to remove thermal emf errors.
You could use reduced current levels for higher resistance but at 10 ohms and up the 8846A is very precise so not much point in that.

[skipjackrc4]

Unfortunately, I do not have an 8845 or 8846 (though I wish I did).  Mine is quite a bit older.  Still, this is a very clever and useful procedure.  I'm not sure that it would work with my equipment, though.  I don't have anything to precisely divide signals.

PS- I saw your video on making the Kelvin clips.  Very nice work!  Clearly this is an area you know a lot about.  I'm certainly grateful for your input.

[nadona]

This is old thread but instead of PM to robrenz, I decided to continued it. Maybe someone can use of this same information in the future.

1) "spend all your money on a 0.1 ohm precision high wattage 4 wire shunt with low TCR."
   Like these? http://bit.ly/1gnovOk  bit.ly/1dCQupP

2) Any other DMMs have the "Ratio" functions? The "Ratio" function of HP 3456A & 3458A, are they same with Fluke's?

[Conrad Hoffman]

Possibly the technique described at http://conradhoffman.com/HamonResistor.html will be of use. Once you have a master divider to compare against, choosing resistors for another is easy. Just set it up as a bridge and compare the two dividers with the most sensitive meter you have. Never let an old thread die!

[nadona]

Possibly the technique described at http://conradhoffman.com/HamonResistor.html will be of use. Once you have a master divider to compare against, choosing resistors for another is easy. Just set it up as a bridge and compare the two dividers with the most sensitive meter you have. Never let an old thread die!

Thank you.
Above thing can be used to measure 1mm ohm resistor to 10ppm or 100kohm to 10ppm at 1KHz?

[Conrad Hoffman]

Stray capacitance and probably even skin effect will throw things way off at 1 kHz. It doesn't seem like that high a frequency, but when you talk ppm, everything matters. You might find that ratio transformers are a better tool for AC, or at least could be used to cross check results. Good ratio transformers will be accurate to a couple PPM, look just like resistive dividers and will have 6 or more decades of adjustment.

[nadona]

According to the GR 1689 manual, to calibrate the meter 4 resistors calibrated to 20ppm at 1KHz are required.

I bought the 7 decade Gertsch RT to get the 95.3K ohm measured at 1KHz. The problem is to begin with at least I need 1 resistor measured at 1KHz accurate to at least 20ppm.

Thank you.

[Vgkid]

The only place for that is to send it off to the npl, that is what iet quotes for their calibration measurements.

[nadona]

What is that npl stands for? Wiki has no entry for it.

All I need one. From there I can measure the others. And I have another GR RCL machine which need same resistors.

IET sells calibration kit which is DC measured and accuracy isn't good as manual says.

[Vgkid]

NPL is the UK version of NIST,
here is the advertisement datasheet for those calibration resistors.
http://www.ietlabs.com/pdf/Datasheets/1689-9604.pdf

[nadona]

I have more accurate resistors than their calibration resistors.
With their cal set I can't cal 0.01pF to 0.02% accuracy. May be 1pF to 0.05%

Their cal set. "Accuracy: 0.1% of nominal value" puzzle me.
1) Manual say it requires 20ppm accuracy measured at 1KHz. IET Cal set is measured with DC
2) GR 1689 meter's spec is 0.02%.
3) More accurate resistors are required to compensate on board four 0.1% resistors.

[Vgkid]

Here is the paper I was looking for, on the calibration of digibridges
http://www.ietlabs.com/pdf/application_notes/OntheCalibrationofDigibridgesandtheVerificationHPH.pdf

[nadona]

Thank you very much for the PDF.

It looks like it doesn't have to be measured at 1KHz.

Do you have URLs pointing to the Appendices?
A. “Frequency Response of Standard Resistors”
B. “Resistance Measurements: AC vs DC”
C. “ Using Digibridges for “DC” Resistance Measurements”
D. Values of HF Standards

[nadona]

Thank you for the 3458A ratio explanation.

1) Is that the shunt in the URL is right type? How precise it should be?
2) 3456A's ratio is different from 3458A's?

[nadona]

DiligentMinds.com: I am very happy with your answer! Thank you.

[Conrad Hoffman]

A silly thing, but I sent a L&N resistor out for cal and the lab wanted to reject it because it was too far from its nominal value. What they forgot is that between the time the resistor was built, and the calibration, NIST changed the definition of the ohm slightly. I can't remember what year, but a lot of the used L&N stuff is easily that old. Personally I prefer older standards- any drifting has probably already happened! Do change the oil in standards, as the oil can become corrosive over time and change the resistance.