milliohm meter, ghetto style

[dannyf]

I wanted to build a simple milliohm meter.

Decided that the simplest I could do is to use a constant current source to drive the dut and then measure the voltage cross the dut.

Here is the schematic.

[dannyf]

Briefly, R1 sets the current - slightly below 100ma.

R2+R3 is a pot, adjusting it will fine tune the current to 100ma - our target current levels.

The regulator used is an ams1117-2.5v.

[dannyf]

The first dut I tested is a piece of solder - maybe 4 - 5 inches long.

The voltage across it is 2.6mv, at 100ma (99.4ma to be exact), yield a reading of 26 milliohm.

[dannyf]

The same solder, shorted about 1/3 from the test points.

Reading of 0.9mv -> 9 milliohms.

[dannyf]

3rd test on the same solder, shorted about 2/3 away from the test points.

Reading suggests 19 milliohms.

[dannyf]

At those levels, test leads' resistance cannot be ignored. So for accuracy purposes, you will need a kelvin clip to do that.

You can build one fairly inexpensively if you want.

[robrenz]

What about some means for thermal emf cancellation?

[(In)Sanity]

I often measure current shunts and such by just passing exactly 1.00000 amps through them and reading the voltage drop.   Nothing Ghetto about this technique. 

Jeff

[robrenz]

I often measure current shunts and such by just passing exactly 1.00000 amps through them and reading the voltage drop.   Nothing Ghetto about this technique. 

Jeff

You can eliminate the thermal emf errors by reversing the current for a second reading and averaging the two readings.

[(In)Sanity]

I often measure current shunts and such by just passing exactly 1.00000 amps through them and reading the voltage drop.   Nothing Ghetto about this technique. 

Jeff

You can eliminate the thermal emf errors by reversing the current for a second reading and averaging the two readings.

Hmm, good to know.  Normally the items I would be measuring in that manor don't have any heating that appears to impact the reading,  although I'm sure that is in fact happening. 

Jeff

[owiecc]

I often measure current shunts and such by just passing exactly 1.00000 amps through them and reading the voltage drop.   Nothing Ghetto about this technique.

It will be ok if you can estimate the error of your measurement. Then your measurement can be said to be ok. Otherwise it is an indication of the resistance level, not a good measurement.

E.g. why don't you use a 1.00000000000A current instead of 1.00000A? Is it different form a 1A current source? Do you measure 1.00000 or do you set 1.00000A? What is the accuracy of your current source? What is the accuracy of your voltage measurement? When where they calibrated? Is 1.00000A still a 1.00000A? How many times do you do the measurement? Were the instruments warm when you made the measurement? What about other error sources like self-heating and emf? Where did you do the voltage measurement? Is moving the voltage sense lines change the measurement?

[dannyf]

Design considerations:

1) ams1117-2.5: ams1117 was used since I have tons of them; Other devices, like 7805 can also be used; or even tl431 + power regulator; I picked the 2.5v version to 1) minimize dissipation on the 33ohm resistor and 2) allow the adapter to work under a 5v source;

2) 33ohm resistor: it is picked so to generate a current of just slightly less than the desired current. In this case, the 33ohm resistor is actually a 33ohm resistor (visible on the top of the perf board) in parallel with a 270ohm resistor (not visible on the bottom side of the perf board). This allows me the flexibility to run the whole thing at 100ma (x10 factor for readings) or 10ma (x100 factor for readings). The 10ma current level is desirable if you wish to power the adapter at high voltage to measure higher resistance.

the power dissipation on the 33ohm resistor is about 0.3w. I used a 1/4w resistor instead. Didn't feel warm and the readings were steady over a 2hr run.

3) the pot: I used a 1k single turn pot, as that's all I had. Ideally, you want to use a lower value multi-turn pot, 470ohm or 330ohm would have been great - with a single turn pot, setting the precise current levels is difficult.

The adjustment range with the 1k pot is about 90ma - 130ma.

4) power supply: the adapter is designed to take 5v sources -> it minimizes power dissipation over the ams1117 (a smt part used here). If you wish to power it at a higher voltage, like 12v or even higher, you need to make sure that 1) it doesn't exceed the rated voltage on the regulator (15-17v typical); and 2) the power dissipation on the regulator isn't too much. I think for 12v or higher @ 100ma, you need a to220 device, potentially with a heatsink.

5) current stability vs. power supply: I set the current to 99.4ma at 5v and ran it ***briefly**** at 12v too, giving a current reading of 99.2ma. Reasonably stable I may say. The regulator did get fairly warm (but not hot) quickly.

Hope it helps.

[dannyf]

Readjusted the current to 99.9ma - good enough.

Here is a measurement of a 1ohm resistor:

A reading of 99.6mv equates to a value of 99.6 * 10 = 996 milliohm, vs. its norminal value of 1000milliohm.

Or more precisely, 99.6mv / 99.9ma = 996.997milliohm.

[(In)Sanity]

I often measure current shunts and such by just passing exactly 1.00000 amps through them and reading the voltage drop.   Nothing Ghetto about this technique.

It will be ok if you can estimate the error of your measurement. Then your measurement can be said to be ok. Otherwise it is an indication of the resistance level, not a good measurement.

E.g. why don't you use a 1.00000000000A current instead of 1.00000A? Is it different form a 1A current source? Do you measure 1.00000 or do you set 1.00000A? What is the accuracy of your current source? What is the accuracy of your voltage measurement? When where they calibrated? Is 1.00000A still a 1.00000A? How many times do you do the measurement? Were the instruments warm when you made the measurement? What about other error sources like self-heating and emf? Where did you do the voltage measurement? Is moving the voltage sense lines change the measurement?

Constant current source adjusted in to my 3457A and I'll either use my Fluke 8520A or Phillips PM2535 to measure the voltage drop.   I never said it was a precise measurement,  but if your going to do it you might as well get as many zero's in there as you can.  Self heating is a problem,  however if I'm using this technique to measure a current shunt I almost want the self heating error included.   I can also just use the 4 wire measurement mode on one of my 3 meters that have the option and get pretty good results.   

Jeff