Resistance (ohm) Electrical Measurements with a DMM - Verify Shunt?

[AlbertPaglinawan]

electrical measurements

I have questions and I would be very glad if anyone will help me in these.

ok, I have a current shunt (50A, 50mV) with a resistance of 1 milli ohm ± 0.01 milli ohm (rated resistance). The actual, based on the calibration certificate is 0.996 milli ohm. I tried to measure the shunt using our DMM with these specs:

accuracy = ±0.024515 milli ohm
readability/resolution = 0.0001 milli ohm
range is 10 ohm

After measuring the shunt, i read 0.0082402 ohms on my DMM.

My set-up is this;

I used 4 banana connectors (23-inch long), wrapped it with copper shielding foil, and then wrapped it still with knitted wire mesh tape (made from nickel-copper alloy). I connected one end to my DMM, then the other end, I shorted to set my zero offset, then connected it to my current shunt. Then I measure. <b>8.24 milli ohms</b>. What went wrong?

<b>note:</b> The certificate said they used 2 multimeters (HP 34401A and HP 3458A) and 1 current shunt (Leeds & Northup 7638). My DMM is Wavetek 1281.

I tried to make this as detailed as possible but if I missed something, please tell me.

another thing, is there another more accurate way of calibrating my shunt?

cheers!

 

[Jerry Eldred]

It's been a while since I've used a Datron 1281. I just looked up the specs. Although it's a very good multimeter, you may have difficulty in working with 1 milliohm on it.

Remember for measuring such a low resistance, you make what I have rhetorically called "Ohm's Law Measurements". That's just one of those terms I made up.

What I mean by that is in order to make a good measurement of a 1 milliohm shunt (I am presuming it is an Empro or equivalent shunt; it's brass on a black plastic-looking base), you need to use ohms law. In this case R=E/I. Your shunt is the R. You then need to measure E (V) and I (Current).

You need one of two methods, a substitute shunt (which was apparently the method used, and is desirable) is the common one. I've used some of the Guildlines. Those are pretty much the standard of choice (no plug for a vendor here. Only my lab experiences).

The second method is a good milliohm meter. There are some on the market.

Using the Ohms law method, run a pretty significant current of known accuracy through the shunt. Such as from a good meter calibrator is a good example. Measure the current with a good DMM with current function (and adequate range). Ten amps would be a good level (more if you've got it available).

Measure the current with a good meter. Make sure you maintain your uncertainty ratio o f 4:1, etc.

Measure the millivolts across the shunt being calibrated with a good meter (4:1).

Do the math, and calculate the value of the shunt.

If you have to use four wire ohms function on a high accuracy multimeter, shielding of the leads is not quite as important as lead resistance compensation. You need a GOOD zero on the leads. I've done it with about 12 Gauge solid copper wire, and got passing results. But it takes some finesse. You also need to take care for things such as current on the outer lugs, and voltage on the inner lugs.

I would strongly recommend using the Ohms Law method if you must do the shunt inhouse.

I would strongly recommend that the BEST method is using a standard shunt with 4:1 uncertainty ratio and pumping at least 10 amps or more through your shunt.

Measurement errors are really easy to get doing those low resistance readings.

 

[Ryan Wilde]

The Datron 1281 is a great meter, I love ours and wouldn't trade it for the world. Were you using it in the "True Ohm" function (recommended for low ohms applications)? Had you run an autocal recently? Does your 1281 read 0.0000000 when the leads are all shorted together? All have a huge effect on ohms measurements using the 1281. It also helps to use low-thermal leads, since stray current from all of the little thermocouples you just made makes a big difference.

But, that may not be the problem here....

Current shunts MUST be measured with current applied, and not ohmmeter current. They are also NOT COMPLETELY FLAT (although 8 times the nominal is way out of line). In other words, my resistance at 1 A will most definitely NOT be my resistance at 50 A (every one that I've seen increases with current, your results may vary). Hence, a good calibration of a current shunt will have multiple current points checked, and a power curve computed. The power curve of a good Weston or Leeds & Northrup shunt will be quite linear on some slight slope. Cheaper shunts tend to have a square curve (parabolic) with the increase in current.

Okay, that said, I just checked my shunt with both our Datron 1281 and an HP 3458A, and it read right where I expected it to read according to my power curve (0.000993 Ù) said it should.

That said, I would still highly suggest using Jerry's method of running a known current through the shunt and measuring the voltage drop across the shunt in this application.

Or maybe your shunt is bad, or annealed, or something like that...

Ryan

 

[Jerry Eldred]

I may have missed one small detail in my original reply. You asked what went wrong. Although I talked a lot about good methods etc., I did not mention that my best "guess" is that assuming your shunt is good (within tolerance), and your meter is accurate, exactly what the error is...

My guess is that it is lead error. The banana leads are good leads for many applications, but at 1 milliohm, there is significant likelihood that "small" differences in lead resistance will produce your error.

I read the post again. I initially had a mental picture of a reading of .842 milliohms versus 1 milliohm. I re-read and realized you read 8.42 milliohms compared with ~0.996 milliohm nominal. That's about 845%. Way off the map. As Irecall from times I've done those Empro shunts that, trying to use banana leads made me pretty well crazy.

We even keep Pomona gold plated spade lug wires here in the lab, and I wouldn't trust those at 1 milliohm (I get pretty skiddish using them at 1 ohm).

Ryan,

Thanks for the good input on the 1281. I think I've used them a long time ago. It may actually have been the 1271 (the old brown meter with what I seem to recall as being nixies, or some older looking LED's). I'm a 3458A person for quite a few years. Have you seen the "Fluke-HP 3458A meter with the added Fluke diode reference. They tightened up a few of the specs. The other meter I'm in the middle of doing an interlab comparison with (shipping it to China next week) is the HP 34420A. That is an awesome meter for voltages between 1 mVDC and 100 mVDC, and resistance ranges of 1 Ohms to 1 Megohm. I would actually trust it's readings of a 1 milliohm shunt better than that of a 3458A. Sorry, I digress.

If you have any possible interest in participating in our ILC, let me know offline via email. I've sent it to 4 subcontractor labs, and 6 inhouse labs in the US, Europe and Asia. That's what I'm working on this moment.

 

[AlbertPaglinawan]

Thanks!

Yes, the 1281 is in True Ohm 4-wire mode in 10 ohm range, 8 resolutions. when I short it, it reads 0.0000052 ohms.

I have a multifunction calibrator which can give 10A ± 6.44mA (DC), and a DMM that has an accuracy of 0.00012mV. So doing math, It would seem that I have a standard that has an error of; UL=1.001 milliohm and LL=0.9993684 milliohm. This would make a TUR from 10:1 up to 15.83:1.

Bottom line, can I use this to calibrate my shunt without acquiring a more accurate current shunt? Can I just use my multifunction calibrator (Wavetek 9100) to pump 10A into my shunt and measure the corresponding voltage?

Cheers!

 

[Ryan Wilde]

Originally posted by Lord Ituralde
Thanks!

Yes, the 1281 is in True Ohm 4-wire mode in 10 ohm range, 8 resolutions. when I short it, it reads 0.0000052 ohms.

I have a multifunction calibrator which can give 10A ± 6.44mA (DC), and a DMM that has an accuracy of 0.00012mV. So doing math, It would seem that I have a standard that has an error of; UL=1.001 milliohm and LL=0.9993684 milliohm. This would make a TUR from 10:1 up to 15.83:1.

Bottom line, can I use this to calibrate my shunt without acquiring a more accurate current shunt? Can I just use my multifunction calibrator (Wavetek 9100) to pump 10A into my shunt and measure the corresponding voltage?

Cheers!

Your specs are a bit off actually, as the 1281 has a small asterisk on all of its specs, which is that you must add the uncertainty of the standards used to calibrate the 1281 to the equation, as the 1281 printed specs are for its own stability. It does not take into account the uncertainty of the standard against which it was calibrated. Therefore, the better the lab that calibrates it, the better it is to the true measurement, and the better its uncertainty. In other words, you are probably closer to the 15 PPM region than the 6 + 0.5 PPM region that they quote.

All that notwithstanding, when you use the 9100, you ARE calibrating the shunt against a more accurate shunt, it is just internal to the 9100. So that is covered. The problem is that you can only calibrate your shunt up to 10 Amps, and above 10 amps it is basically uncalibrated (if you can't prove it, it isn't correct). If you can live with the 10 Amp limitation, then go for it, your equipment is easily up to the task. Depending on the quality system that you work under (QS 9000, ISO 9000, or the dreaded ISO 17025), you may even be able to get away with no limitation whatsoever. If you are under 17025, you can't do it without the limitation. If you are under ISO or QS 9000, you may, with creative procedure writing and a bit of research, so read on...

Option 1: Call the manufacturer of the shunts you use. Ask them what the 'typical' power curve for your particular model is. If they don't know, I wouldn't buy their shunts. Most likely, they will know. Calculate out the probable error at 50 Amps, ensure it still falls within the 1% tolerance window, and consider it cal'd. Make sure your procedure reflects the method, and be prepared to answer questions, and keep the manufacturer response on file, even if you have to write it down yourself.

Option 2 (the better option): Have your shunt calibrated by an outside source for its initial calibration. Make sure they give you at least 4 or 5 points so that you can calculate some form of power curve. Make sure one of those points is 10 Amps. Calibrate the shunt yourself after this, comparing the 10 Amp point versus the intial calibration. You can keep this up until the first time the shunt is overcurrented, or left at full current for some length of time, as the power curve will change as the resistive element anneals. At this point, send it out for cal again, and start the process over.

In the long run, you will probably save your company some money with these methods, and your president won't have to spend holidays domestically, even with the bad economy, and we all know how important that is.

Ryan

 

[Ryan Wilde]

Jerry,

I don't trust the gold plated spade wires either. If it matters enough that I'm worried, I always use shielded low-thermal leads (copper bananas into copper connections which are standard on all high-precision DC equipment). I've seen 10 PPM shifts using the gold-plated, and sometimes that is enough to kill a measurement. The only place I've found the low-thermals is from Fluke, and they are not cheap, but having a set of 4 is worth its weight in gold when you are in doubt.

The 1281, as long as it is calibrated by a great lab, is still the standard against which all other meters are measured. We have the 3458A with the internal zener reference option, and it is very close to the accuracy of the 1281 (BTW, it sounds like you did have the 1271, which is also a great meter). I haven't run across a 34420A yet, but it sounds like I would really like to do so. Low voltage ranges are always very dicey, and if Agilent came up with a way to measure it more reliably, I will beg for one for the next several years until they give in to shut me up.

I should also note that I am also a 3458A person, and I wouldn't use the 1281 in most applications. The 3458A is universal and more software is written for it, and its speed is also a factor in daily operation. I hate writing my own code, so rarely do I use the 1281 automatically. But, when I need to make a measurement that I trust manually, you'll usually find me in front of the 1281. For example, on a whim I decided to measure all of my standard DC references with both the 3458 and the 1281, just to see how they agreed with my miles of data on the References. The 1281 was within 1.5 PPM in all cases, while the 3458 seemed to use more of its uncertainty (right around 3-6 PPM as I remember). I know, it is down in the 'don't care' region, but still, I really love nominal.

Well, back to real topics and such, I think I've strayed enough.

Ryan

 

[AlbertPaglinawan]

In the long run, you will probably save your company some money with these methods, and your president won't have to spend holidays domestically, even with the bad economy, and we all know how important that is.

pretty important alright

Thanks! You have given me a new perspective in calibration certificate (plus many other things as well ), and just in time coz our 1281 will be due on april. I will remember to request the accuracy of the standards used.

This is the first time the our 1281 will be calibrated (in my opinion) as we just purchased this last year. We only have the "Certificate of Calibration" sent by fluke with nothing but a phrase that "...meets or exceeds all relevant published specifications." No test data, no standards used, no metrologist signature (but it does have a calibration date though plus equiment details). I don't think we can use this as valid certificate of calibration (More like a certificate of... I dunno, conformity?).

Anyway, those are just random thoughts. Thanks again.

Cheers!

 

[Ryan Wilde]

Originally posted by Lord Ituralde


Thanks! You have given me a new perspective in calibration certificate (plus many other things as well ), and just in time coz our 1281 will be due on april. I will remember to request the accuracy of the standards used.

What really helps is to ask them for their scope of Accreditation, which outlines the uncertainty that they are capable of on the elements that you use. I can tell you that Fluke is VERY good. In DC Voltage (the basis of all 1281 measurements) Fluke has characterized their working reference cells to less than 0.5 PPM, which is amazing for a working standard.

This is the first time the our 1281 will be calibrated (in my opinion) as we just purchased this last year. We only have the "Certificate of Calibration" sent by fluke with nothing but a phrase that "...meets or exceeds all relevant published specifications." No test data, no standards used, no metrologist signature (but it does have a calibration date though plus equiment details). I don't think we can use this as valid certificate of calibration (More like a certificate of... I dunno, conformity?).

Unless you specify "accredited with data" you will not get all of the elements required by ISO 17025 from any manufacturer. Our purchase orders have a statement to the effect that all calibrations must be provided to us "accredited with data". But, I can tell you that in the interim, Fluke is very forthcoming with the information that you seek. If you call and ask them, they will tell you the uncertainty at which your unit's calibration was performed. I guess that is why we all keep buying their equipment.

Cheers back at you!
Ryan

 

[Graeme]

Calibration of your own standards

Originally posted by Lord Ituralde


Thanks! You have given me a new perspective in calibration certificate ... and just in time coz our 1281 will be due on april. I will remember to request the accuracy of the standards used.

This is the first time the our 1281 will be calibrated (in my opinion) as we just purchased this last year. We only have the "Certificate of Calibration" sent by fluke with nothing but a phrase that "...meets or exceeds all relevant published specifications." No test data, no standards used, no metrologist signature (but it does have a calibration date though plus equiment details). I don't think we can use this as valid certificate of calibration (More like a certificate of... I dunno, conformity?).

The practice that I reccomend, and have seen in many calibration labs, is that ALL of the measurment standards (transfer and working) get "before" and "after" test data. If you send a standard out for calibration it costs more, but it can be worth it. When purchased new, you should ask for a Report of Calibration with data. Again, it costs more, but if you don't ask you will likely get only a piece of paper that says (in effect) "we built it and we say it meets our specs, so there."

The "before" data is the results of a calibration run without doing anything else. The value of this data is that it tells you what the as-found condition of the standard was and if there are any out of tolerance values. You can also compare it to the "after" data from the previous calibration to determine peformance trends over time.

If everything is in tolerance, or if is is a fixed device such as a standard resistor, this will probably be the only set of values.

The "after" data is the results of a calibration run after the instrument has been adjusted, repaired, optimized or otherwise has value added to it. This is the condition when it is sent back to you. You can use this (see above) to monitor the equipment over time and to establish some of your Type B uncertainty.

Depending on the instrument, the data can range from a single value with an uncertainty (as in a standard resistor) to a fairly thick report. If my memory is correct, when Fluke (formerly Wavetek) calibrates my 1281 the before and after reports are about seven pages each. Other instruments are thicker, many can be only one page. But the data (nominal, limits, measured value, measurment uncertainty, and more) is very valuable and is necessary to build the Type B uncertainty database of that measurement system.

If you don't ask for the data you probably won't get it.
If you don't have the data then you really have very little knowledge about how the standard is performing.