Resistance of high power and small resistivity

#10? No way 2 watts through that will exhibit much heat... at all.

--
Yeah, you\'re right.

Thanks. :-) I don\'t know what I was thinking...

JF

I don't know what "measure accurately" means to you, but to me it means at least

1% accuracy. The OP hasn't said it directly, but from the previous post, I get the feeling that the best his voltmeter can resolve is 1 mV. So the shunt you are proposing here would have .5 mV across it with the 1 amp he proposes to use to calibrate it. How does he measure .5 mV with 1 % accuracy if his meter can only resolve 1 mV?

Same problem as before. During the calibration with 1 amp, as he proposed, he has 5 mV across the shunt. How does he measure this with 1% accuracy with his voltmeter which apparently can only resolve 1 mV?

calibrate

to get

steel, its

such

And,

have no

of my

So it could, but what is the relevance of this to the OP's situation? He certainly doesn't have this kind of instrumentation.

--
He stated that resolving 1A into 10mV would be OK, so that\'s at least a
+/- 5% error, which would make the shunt the most accurate thing in his
setup, obviating its need to be calibrated.

least

No need to calibrate it? How shall he make it, then? Just measure the length of a piece of copper wire?

He hasn't told us the accuracy of his meter. He hasn't even said if it's digital or if it's a Radio Shack analog job. So how do you know that a 5% shunt would be the most accurate thing in his setup?

In this day and age, he probably has a low cost DVM, which is going to have better than 5% accuracy (of full scale, of course).

He said he can measure 2 amps with some meter, presumably his DVM, and that range will be better than 5% if it's on a DVM. Although, in order to calibrate his homemade shunt, he will need a separate ammeter and voltmeter. Or, perhaps he's going to use the meter on his power supply to indicate the 1 amp calibrating current. I tried to ask about this, but he didn't respond.

At any rate, you don't know what the accuracy of his existing instruments is, because he hasn't told us.

Furthermore, I wouldn't say that a 5% shunt allows one to "measure accurately".

--- Yes, and its diameter, and then consult a magnet wire table.

If he can measure the length of a 12" long piece of wire to +/- 1/16" that's an accuracy of about 1 part in 192 which is just short of +/-

0.5%.

If he has no way to measure the diameter and get the resistance from cross-sectional area, resistivity, and length, then from:

we find that #10AWG has a maximum resistance of 1.019 ohms/1000', a nominal resistance of 0.9988 ohms/1000', and a minimum resistance of

0.9795ohms/1000', so that comes out to:

Rmin = 0.9795 \\ dR = - .0193 ohms / Rnom = 0.9988 \\ dR = 0.0202 ohms / Rmax = 1.019

The resistance uncertainty about nominal then varies from 2.02% to

-1.93%, say +/- 2%, which when added to the +/- 0.5% uncertainty due to length comes out to a total error of about +/- 2.5%.

Not too shabby for an off-the-shelf chunk of wire, huh?

---

--- Because, as I stated earlier:

"He stated that resolving 1A into 10mV would be OK, so that's at least a

+/- 5% error, which would make the shunt the most accurate thing in his setup, obviating its need to be calibrated."

In other words, if he chooses his calibration range as 10mV and his meter has a granularity of 1mV, then if the meter's actually at 5.5mV it has to read either '5' or '6', so the error in the reading will be either plus or minus 500µV, which is +/- 5% of the 10mV scale, making the shunt twice as accurate as the meter at that point.

---

--- Yes, but in actuality the meter will have its own errors which will add to the inherent +/- 5% realized by using a 10mV range with 1mV granularity.

---

--- The accuracy of the meter has little to do with it; what his limitation will be will be the 1mV granularity over a 10mV span.

---

--- No.

All he needs to know is the length of the copper wire, its cross sectional area, and the resistivity of soft-drawn copper.

Temperature, too, if he really wants to get into it.

---

--- Nor does it matter since if he's going to use a 10mV span to do his "calibration", the uncertainty in the resistance of the cut-to-length shunt will be less than the uncertainty in the meter reading.

---

---

You wouldn't, but the OP may well be since he's willing to accept a 5% "calibration" and the shunt is twice as accurate as that, right off the bat.

Since his meter is more than likely a garden-variety 3-1/2 digit DMM using the ubiquitous ICL7136:

there's an easy way to get better accuracy, and that's to make the shunt longer.

For example, to make full use of the ADC's capability, the input signal span should be 199.9mV, where +/- 1 count is equal to 1 part in 2000, or

+/-0.05%.

To do that would require the shunt to drop 199.9mV with 20A through it, so its resistance would need to be:

E 0.1999V R = --- = --------- = .009995 ohm I 20A

Since 10AWG has a resistance of 0.9988 ohms per 1000', the length required to get 0.009995 will be:

0.9988R 0.009995 --------- = ---------- = 120.08" 12000" x"

A cutting error of 1/2" results in a resistance uncertainty of about a quarter of a percent but, unfortunately, the uncertainty in the diameter will still add about 2% to the uncertainty of the resistance, which might be OK for the OP.

Dunno; he vanished... JF

The OP, and apparently you also, feel that the phrase "measure accurately" will be satisfied by 5%+ accuracy. I don't share the feeling. As I said earlier, to me "measure accurately" means at least 1% accuracy.

The way for him to overcome the granularity problem is to use a shunt of higher resistance. Such a shunt will dissipate more power and get hotter; hence the need for a resistance of low temperature coefficient.

shunt

You apparently meant by the word "setup", just the current measuring capability he's trying to build. I took it to mean all of his instruments, and their accuracy specs.

I understand all of this, and it's the reason why he won't be able to "measure accurately" a 20 amp current doing things that way. Of course, if "measure accurately" means 5%+ is ok, then no further discussion is needed. I'm not persuaded that he will be "measuring accurately" if he uses a 1 foot long piece of copper wire for a shunt. But, if he is, and if you are, then he should have at it.

Again, if he wants to "measure accurately", he most definitely needs to take account of the temperature. Copper changes resistivity by about .4%/degree. For about a 12 degree change in temperature, he has incurred another 5% error.

Only if he takes the temperature into account. A relatively small change in temperature destroys your assumed 2.5% initial accuracy. And, it's not enough to know the ambient temperature; he has to know the temperature of the wire. I passed 20 amps through a 1 foot long piece of 10 gauge wire in free air, and it gets noticeably warmer; how much I couldn't say with any accuracy, but it felt like 5 or 10 degrees. The temperature uncertainty will be a substantial obstacle to getting even 5% accuracy. He would have to put it in a temperature controlled oil bath.

We should take a poll of readers, to see how many think the phrase "measure accurately" is satisfied by 5%. Even old analog meters can do better than that.

If he spends a little money on a low tempco shunt of 10 milliohm resistance, then he can get 1% accuracy at 20 amps, and not have to worry about the temperature. He could get an analog meter on eBay that can measure 20 amps DC without a shunt and do better than the 1 foot piece of copper wire.

If he uses a 1 foot long piece of 10 gauge copper wire, taking into account all the sources of error, including the important one of wire temperature, then he is measuring roughly, not accurately.

If this is good enough for the OP, then more power to him, but he should be aware that he isn't "measuring accurately".

This discussion has come down to an apparent disagreement between you and me about what "measure accurately" means. I think you will agree that he can't get

1% accuracy with his meters and a 1 foot piece of copper wire.

Using a .01 ohm shunt rather than a .001 ohm one will take care of the granularity in his measurement, but the other sources of error, including the all-important temperature rise uncertainty and the high tempco of copper puts him back in the 5%+ ballpark. A commercial low tempco shunt of .01 ohms would give him about 1% accuracy.