I'm attempting to calibrate a precision current source, @ 1A, and 3A. I have a Fluke 87v, and a Brymen 857.
I also have a precision 100 mOhm resister, 0.1% tol., 10ppm, 3W:

https://www.digikey.com/en/products/detail/vpg-foil-resistors/Y14750R10000B9R/8538226

When I adjust my current output for say, ~3A, and measure across the resistor, I'm reading 300mV. Because it's 0.1% shunt, my slop might be 300.3mV, or 299.7mV. Wattage under test is 0.9W, well within this shunt resistor's specification.

Now, the % difference between the ideal 300mV, and a possible outlier of say 300.3 is = 0.1%, as expected. However, my Fluke in current measurement reads 2.978A, which is .73333333% difference, when it should read no more than 0.2% deviation, according to it's specifications. My Brymen 857 current reading is 2.9761A, .79666666%, again way outside of the meters specifications.

This is the current spec. for the Fluke 87v:
Scale: 6A, Resolution: 0.001A, Burden: 0.03 V/A, Accuracy: ±(0.2 % + 4)

I don't know who to trust! I paid nearly $30 for that resistor, is it lying to me?

 

Hard to say since you can't confirm the actual current.

 

how are you using it? shunts require Kelvin connection. that product has 4 terminals just for that. any chance they got shorted by soldering?

 

Yes, that's a possibility. Let me check and get back to you. Thank you!

 

How are you making the measurement with your meter? are you replacing the shunt with the meter? If so, you are changing the circuit by replacing a 100 mΩ resistor with a 30 mΩ resistor. Different circuit, different current (though I would expect the current to go up in that case). If you are putting it in series with the shunt, then I would expect it to go down, but then the voltage across the shunt should go down, too.

Do either of these meters have a current, valid, calibration?

You give no hint as to what this circuit is that you are measuring or how you are adjusting the current in the first place.

Are both current meters in series with the shunt at the same time? If so, the fact that the two meters agree very closely (within about 0.03%) would bias me toward accepting their values. But, they could both be out of tolerance in the same direction by about the same amount.

There are also temperature effects to be considered, both for the meters and for the shunt.

You say that 0.9 W is well within it's spec'ed wattage range. While this is technically true, given that the rating is 3 W, you are actually running it at a good fraction of what most people consider good practice for power ratings, which is to not run components above half their rating. For metrology, this is even more important as running them at 50% of rated power will normally mean that they will be running much hotter than ambient, which will magnify their temperature sensitivities.

How has this shunt been treated? Was it soldered according to the specifications? Even if it was, that could let it change by 0.25%. Has it been exposed to humidity? How old it is it -- new off the shelf? It can change by up to 1% over a 2000 hr operating life.

But perhaps most importantly, what level of accuracy do you really, truly, NEED for you application?

 

what level of accuracy do you really, truly, NEED for you application?

Was wondering the same thing.
Also, how are you negating the test lead ohmage? Are you completely zeroing out their added line resistance?

 

how are you negating the test lead ohmage?

For which particular measurement technique?

However, my Fluke in current measurement reads 2.978A

Are the shunt and the Fluke both connected when you do the shunt measurement?
They should be for an accurate measurement comparison.

 

At 1 A and 3 A into 0.1 Ω resistor is sure to be difficult to measure.
There are sources of error that will need to be considered.

1) Resistance of your cables
2) Internal resistance of the ammeter
3) Temperature dependency

You need to take voltage measurements or a use Wheatstone bridge with a reference resistor, i.e. one with known calibration.

 

how are you negating the test lead ohmage?

For which particular measurement technique?

With high number readings such as high voltage, high current or high resistance, the test lead ohms are negligible. But with low numbers such as low value resistance or micro - currents down where you're looking to calibrate to - the test lead resistance is important. MrChips seems to agree.

1) Resistance of your cables

 

Depending on where you put the resistor or the ammeter, it could be interfering with the feedback mechanism of the constant current source. It must be outside the feedback loop for an accurate reading.

 

Are your meters calibrated and certified? Did you zero out the lead resistance for the meters? What is the exact resistance of the resistor measured on a certified calibrated instrument? Why does it even matter? So many variables...

 

With high number readings such as high voltage, high current or high resistance, the test lead ohms are negligible. But with low numbers such as low value resistance or micro - currents down where you're looking to calibrate to - the test lead resistance is important. MrChips seems to agree.

The resistor he is using isn't just a resistor, it is a four-wire Kelvin-configured current shunt. Even a volt meter with a measly 100 kΩ resistance will only shunt 1 ppm of the current from a 100 mΩ shunt.

 

Let's look up the resistance of copper wire.

Diameter of #18 AWG is 1 mm.
Diameter of #12 AWG is 2 mm.

Given the resistivity of copper is 1.71E-8 Ω-m at 20 °C,
I will be generous and allow for 1 m of cable.

Resistance of 1 m of #18 copper cable is 0.022 Ω
Resistance of 1 m of #12 copper cable is 0.005 Ω

You can try making your own 100 mΩ standard resistor.
Resistance of 1.2 m of #24 copper wire is 100.06 mΩ

Temperature coefficient of resistance for copper is 0.393% per °C
Thus 1 °C change in temperature will affect the resistance by 0.4%.

 

Let's look up the resistance of copper wire.

Diameter of #18 AWG is 1 mm.
Diameter of #12 AWG is 2 mm.

Given the resistivity of copper is 1.71E-8 Ω-m at 20 °C,
I will be generous and allow for 1 m of cable.

Resistance of 1 m of #18 copper cable is 0.022 Ω
Resistance of 1 m of #12 copper cable is 0.005 Ω

You can try making your own 100 mΩ standard resistor.
Resistance of 1.2 m of #24 copper wire is 0.10006 Ω

Temperature coefficient of resistance for copper is 0.393% per °C
Thus 1 °C change in temperature will affect the resistance by 0.4%.

Not sure what the point here is. Maybe I missed it, but I don't see where anyone has suggested that he jury rig his own shunt or that the shunt he is using isn't adequate for his needs.

How did you use numbers with one to three sig figs to arrive an a resistance good to five?

What are the tolerances on the resistivity of the copper used for the wire?

What are the tolerances on the diameters of the wire?

The temperature coefficient alone would seem to make a copper-wire current shunt completely impractical in any measurement situation in which accuracy and precision are important (which still leaves a lot of applications where these aren't critical considerations). What is the temperature rise going to be of 1.2 m of #24 copper wire carrying 3 A of current? It's dissipating 300 mW of heat over a very small surface area. Plus, the rated ampacity of #24 wire in free air is only 3.5 A.

If this is related to needing to take the resistance of the lead wires into account, it seems more detail needs to be given to make that point.

 

with low numbers such as low value resistance or micro - currents down where you're looking to calibrate to - the test lead resistance is important.

Only if they are in series with the current and the circuit providing the current is not a stiff (high-impedance) current source.
The TS stated it was a "precision current source" so I would think it would be fairly stiff, but we don't know for sure.

 

the discrepancy could be result of used method. for example using one DMM (such as Fluke) to measure voltage across shunt... and then using same DMM to measure current. in both cases test leads length is the same but when measuring voltage this has no effect on test. when measuring current, test leads mean that entire circuit has higher resistance (because current also has to go through leads) and if PSU is not compensating for that, measured current will of course be lower (added resistance of DMM shunt and two leads).

 

The problem described by the TS is precisely the reason that calibration traceability is required to have confidence on the instruments you’re using and the values you’re reading.
And that is again the reason that when one purchases a reference, resistance, voltage, whatever, it should come with a reading obtained with a traceable instrument.

 

The problem described by the TS is precisely the reason that calibration traceability is required to have confidence on the instruments you’re using and the values you’re reading.
And that is again the reason that when one purchases a reference, resistance, voltage, whatever, it should come with a reading obtained with a traceable instrument.

And, that traceability is only useful if the instrument that provided the calibration had a current calibration certificate, and the instrument that was calibrated with it is also current. I've seen people claim that they are using a meter with a NIST-traceable calibration and they were technically telling the truth, just leaving out that the last time it was calibrated was more than thirty years ago.

The TS says that he is calibrating a precision current source. Taken at face value, that would mean that he should be using equipment and techniques that allow him to guarantee the performance of that source over a set of operating conditions for a set period of time. But "calibrating" is such an amorphous term that we don't really know how to interpret it, which gets back to that critical question that many ask and so few answer -- what is NEEDED in order for this source to be considered "calibrated" for the intended application.

 

Though I've never used a Wheatstone bridge - I believe they are used to measure precise values of an unknown resistor. I would think that this description could be expanded to say "Verify the exact resistance of a suspect resistor".

The resistor he is using isn't just a resistor, it is a four-wire Kelvin-configured current shunt.

I've never known such a resistor like this. Have no understanding of how it works, nor do I know if a Wheatstone bridge would work. I only offer food for thought.
View attachment 360304

 

https://www.allaboutcircuits.com/textbook/direct-current/chpt-8/kelvin-resistance-measurement/