Milliohmeter

[dr pepper]

Anyone built one?
I just prototyped a circuit.
I want to measure the resistance of some cables on a machine, down to milliohms.
This needs to be battery powered so 100ma is the most I want to have test current wise.
As some know offset voltage is an issue for op amps at dc, so amplifying the voltage across a 1mOhm resistance with 100ma through it requires gains of 1000 for a sensible o/p voltage, and with the average op amp offset of 2mv that ends up a large voltage at that gain, so I prototyped a meter using a lmp8358 chopper instrumentation amp with a gain of 1000 found in a thermocouple amplifier circuit, and a lf358 as a current source controlled by a voltage reference, and the other half the lf358 drives a 1ma meter.
Seems reasonably accurate (I only have 5% resistors for the prototype), and stable.
The other idea I had was to pulse the test resistance at 1 or even 10a, and sample/hold.

 

[chemelec]

This one is Good and I can also supply the parts, If Needed.

 

[Colin]

If you know the current is 100mA, just measure the millivolt drop across the cable. 10mV = 1milli-ohm.

 

[AnalogKid]

Wouldn't the input offset voltage error amount to a significant error in the results? As in post #1, a 2 mV offset voltage would be a 20% error in the output.

ak

 

[AnalogKid]

I prototyped a meter using a lmp8358 chopper instrumentation amp with a gain of 1000 found in a thermocouple amplifier circuit, and a lf358 as a current source controlled by a voltage reference, and the other half the lf358 drives a 1ma meter.
Seems reasonably accurate (I only have 5% resistors for the prototype), and stable.
The other idea I had was to pulse the test resistance at 1 or even 10a, and sample/hold.

I vote for the chopper. A 10 A pulse, no matter how short, can damage things. To prevent localized heating skewing your results, I also recommend pulsing the 100 mA.

ak

 

[Les Jones]

I agree with AK that using a short pulse of a higher current is a better solution. With only 100 mA the voltage you are trying to measure will only be 100 uV per milliohm. The thermo electric effect between connections of different metals will result in significant errors. (Unless you can keep everything at the same temperature.)

Les.

 

[dr pepper]

I've assembled a rough prototype, 100ma constant test current, and 1000x gain from an auto zero amp, it works well, readings are unstable till you connect the leads then its dead steady due to the low impedance.
I'm thinking of how I'm going to protect the op amp from noise and spikes on the leads.
1a is a little much for 4 aa batterys and a hour to 2 hours test sesion, unless I pulsed the test current and sample/holded it.
I'm also putting a 1000v insulation checker in the same unit.

 

[spec]

Anyone built one?
I just prototyped a circuit.
I want to measure the resistance of some cables on a machine, down to milliohms.
This needs to be battery powered so 100ma is the most I want to have test current wise.
As some know offset voltage is an issue for op amps at dc, so amplifying the voltage across a 1mOhm resistance with 100ma through it requires gains of 1000 for a sensible o/p voltage, and with the average op amp offset of 2mv that ends up a large voltage at that gain, so I prototyped a meter using a lmp8358 chopper instrumentation amp with a gain of 1000 found in a thermocouple amplifier circuit, and a lf358 as a current source controlled by a voltage reference, and the other half the lf358 drives a 1ma meter.
Seems reasonably accurate (I only have 5% resistors for the prototype), and stable.
The other idea I had was to pulse the test resistance at 1 or even 10a, and sample/hold.

UPDATE 2014_04_10. Please ignore this post doc. I was way out with the calculations.

Hy doc,

You could cut the current consumption to around 20 mA by using a couple of OPA192s: input offset voltage +- 5uV, input current, 0nA. RRIO. ouput current, +-60mA.

Then all you would need would be a precision voltage reference, like the ADR5041 or LT1634, a PMOSFET, a moving coil meter, and a PP3 battery. Probably cost $25 US all up. Be a nice little project to do too.

If you want a suggested circuit, let me know.

spec

 

[dr pepper]

Yes possibly, the 8358 has digital gain inputs, parallel or serial, and the gain in precision so it kinda makes things more straightforward.
I'm using a precision volt ref, its a LM338, as part of the current source.
The plan is to have amplifier, current source and 1000v generator all on seperate boards, and seperate bannana sockets.
I think I'll invert the scale, so low resistance is fsd, I dont like the way the needle gets bent when the leads are disconnected, I did use a couple of diodes to limit the voltage to the meter, but it still slams the endstop.
The 1000v inverter upsets the 8358 inamp, even on x10 gain (which is the only gain settign it will be used on while the 1000v is being generated), the test leads go direct to the + and - i/p's of the op amp so ground or power noise shouldnt get in, and supply rejection for the inamp is over 100db, it might be the fact its breadboarded, I have some very old dc to dc converters I'll use one of those and isolate the amp/meter altogther if I have to.

 

[spec]

This seems like a remarkable miliohmeter- only 5mA test current. I wonder how it is done:

https://www.farnell.com/datasheets/1724071.pdf

spec

 

[schmitt trigger]

Ohm's law states that 5 mA across 1 mΩ means a voltage of only 5μV !!!
So it must use a good quality chopper stabilized opamp as its front end.
But even then thermoelectric effects between the different materials could create a larger offset. There must be some ingenious nulling ciruits.

 

[spec]

Ohm's law states that 5 mA across 1 mΩ means a voltage of only 5μV !!!
So it must use a good quality chopper stabilized opamp as its front end.
But even then thermoelectric effects between the different materials could create a larger offset. There must be some ingenious nulling ciruits.

Hy st,

Yeah, that's how I saw it, although chopper stabilized opamps are not necessary these days. You can get less input offset with trimmed opamps, like the opa192 which I keep on about. This eliminates any side-effects of the chopping action. The OPA192 has a raft of other benefits too, including RRIO and practically no input current.

The measurement of 1 miliohm accurately seemed, at first sight, to be simple, especially if you are three orders out with your calculations. Even injecting a high current, of around 1 amp to increase the voltage, would not work with all units under test (UUT), because a high current would be intrusive and may change the characteristics of some UUTs. In certain cases, it may even blow them. A high current also has the disadvantage of a heavy demand on the power supply, which would be significant for a battery powered instrument. On the other hand, a high current may better reflect the current operationally flowing in some UUTs.

Nulling would possibly do the job if you wanted to stick with a DC approach: measure the voltage, galvanic etc, and store it in a sample and hold. Then inject a current and store the resulting voltage. Subtract the first voltage from the second and you should have the resistance, but I imagine this would be difficult to implement and I'm not sure the principle is even correct. It may be necessary to inject a positive current and then a negative current and then do some calculations on the three readings.

A more promising method, I suspect, would be to inject a very low frequency, low current sine wave into the UUT. As galvanic voltages are DC they would not show up on the resistance reading, at least that is the theory. The resulting sine wave voltage could then be amplified by a low noise active notch filter. Opamp DC offset would not affect the reading, although precision low offset opamps would still be wise. The very low frequency would avoid reactive effects and the notch filter would limit the noise power. A precision rectifier, possibly synchronous, could then convert the amplified AC signal into DC to drive a meter, digital voltmeter or, even a microcontroller. The main problem with this approach would be noise, both conducted and induced and it may be necessary to have more than one measurement channel and do some processing on the outputs before the rectifier.

These are just some thoughts though.

spec

 

[dr pepper]

Yes it is a good chip, it was 8 quid, offset is in the order of nV, its a auto zero chopper amplifier like you said, at x1000 I get 1v full scale for 100ma through a 10mR resistance.
Yes the reading varies if you blow on it, but when wired to a busbar it works well, biggest problem so far is bannana sockets, contact reistance is naff, moving anything changes the current and affects the reading.
I dont want to end up with dinze sockets on this thing.

 

[spec]

Yes it is a good chip, it was 8 quid, offset is in the order of nV, its a auto zero chopper amplifier like you said, at x1000 I get 1v full scale for 100ma through a 10mR resistance.
Yes the reading varies if you blow on it, but when wired to a busbar it works well, biggest problem so far is bannana sockets, contact reistance is naff, moving anything changes the current and affects the reading.
I dont want to end up with dinze sockets on this thing.

Offset in the order of nV. Wow. I will have to have a look at that.

Dinze sockets.

I could imagine the circuit would be sensitive to pressure etc. Have you thought about gold plated probes. Presumably you are using a four wire approach for injecting the current and measuring the voltage drop.

A pretty challenging project.

spec