Background and Goal:

There have been many projects where I wanted to know the resistance of a wire that is very low (low mΩ range). Actually, most projects somehow involve measuring wire resistance. Obviously a regular multimeter cannot do that directly. I have been able to get rough estimates by measuring millivolts when applying a constant current or by using a resistor. But I am not sure how accurate my multimeter is. While the quality seems great, it was pretty cheap. And this process is kind of tedious.

This means I need to get a 4 wire kelvin multimeter. I have looked around, and they seem to be insanely expensive (100s of dollars). I was able to find one or two that were about $75, but that is still a lot. I would like to make my own for under $25.

Here is the goal. Let me know if parts of it are impractical. So I want to be able to measure with an accuracy of ±10-20 μΩ. I know that means an amp for every 10 μV, and thus large currents and precise, low noise op amps are required. I also want it to have a nice display, and auto-ranging for current and op-amp gain.

My idea:

So I have a general idea of how to go about making this. I would first supply a constant current of up to 5-10 amps. I would then measure the drop across the wire with a precision op-amp. I know it needs a low input offset voltage, and it should be low noise, but other than that I am not too sure what exactly to look for. This would get fed into either a higher resolution ADC or directly into an arduino analog pin (I will need to figure out if 10 bits is enough).

For adjusting the range of the op-amp I am thinking resistors with bypass transistors controlled by GPIOs. I could also do the same thing to make a crude digital potentiometer for adjusting the CC of a CC board.

I think I have some spare arduino nanos I could use, or I could just get one for a few dollars off of ebay. I will probably get a cheap OLED display off of amazon to display the current and measured resistance. I will have to find the right library to be able to use I2C, SPI, or whatever the display uses with the arduino.

Where I need help:

So here is where I could really use some help. I first need a precise constant current board that can supply up to 10 amps. I could use suggestions for that. I have one that can do that except it is not too precise (about ± 50-100mA based on my measurments). If I don't know the exact current, then I may have to measure it using a shunt, adding another complex op-amp circuit and more possibilities for errors.

I also imagine at such low voltages, EMI, RFI, and general noise could really get in the way. So how do I reduce noise to acceptable levels, and what will be major sources of it? What op-amp is best for this application, and what is the best setup for it? Specific part numbers would be great.

And is there a better approach here? Does anyone have comments or general suggestions?

 

One thought is to use an AC current through the UUT for the measurement.
That will allow AC amplification of the signal with high gain and no worry about DC offset, either from the amplifier or due to thermoelectric effects.
You could use a bandpass filter and/or synchronous detection to suppress noise and interference in the signal.

The most difficult part may be generating the high accuracy AC current.
You would probably use a frequency somewhere around 1kHz.

 

Also, I kind of remember that there was some way to see who is viewing a thread. Do you know where to look?

 

What makes you think you need 10 Amps?

Do have a look at this: https://www.eevblog.com/projects/ucurrent/

 

Do have a look at this: https://www.eevblog.com/projects/ucurrent/

That device is for measuring current.
The TS wants to measure resistance.

 

You might get by using a well-heatsinked LM317 as a 1 amp current source. That would greatly simplify your project. The IR drop can be sensed using an inexpensive OP-07 with offset adjustment or a chopper stabilized amplifier like the ICL7650.
Example: http://cappels.org/dproj/X100_Microvolt_amplifier/X100_Microvoltls_Offset_DC_Amplifier.html

Building on crutschow's idea of AC coupling, another way to obtain greater sensitivity is to use a synchronous detector"
Example 1: http://www.analog.com/en/analog-dialogue/articles/synchronous-detectors-facilitate-precision.html

Example 2, milliohm meter using a synchronous detector: www.holographyforum.org/data/pdf/aa-Collection_a_k/aa-Laser/aa-lockin/AVR_experimental_synch_detector/dlmom.html

 

A single chip lockin amp (board cost $ 10) -

https://community.cypress.com/thread/15988

See attached.



Regards, Dana.

 

It is possible to buy a 5 digit 3 amp panel meter. Presumably you could, with patience, make an accurate shunt to get it to 30 amps.
Then you could monitor your current source on the fly as you are reading the voltage across the wire, so the source would be adjustable at the time of measurement to exactly 10 amps.

Another problem I foresee is pumping 10 amps through your wire--will it take that much current?

 

You might get by using a well-heatsinked LM317 as a 1 amp current source. That would greatly simplify your project. The IR drop can be sensed using an inexpensive OP-07 with offset adjustment or a chopper stabilized amplifier like the ICL7650.
Example: http://cappels.org/dproj/X100_Microvolt_amplifier/X100_Microvoltls_Offset_DC_Amplifier.html

Building on crutschow's idea of AC coupling, another way to obtain greater sensitivity is to use a synchronous detector"
Example 1: http://www.analog.com/en/analog-dialogue/articles/synchronous-detectors-facilitate-precision.html

Example 2, milliohm meter using a synchronous detector: www.holographyforum.org/data/pdf/aa-Collection_a_k/aa-Laser/aa-lockin/AVR_experimental_synch_detector/dlmom.html

Thanks for the suggestions. The one issue with AC coupling is that I intend to measure coils, where the inductance would lead to a much higher reading. And how would any of the AC or pulse stuff not give terrible readings for inductors? Aside from that issue, I could see how it would reduce the noise at other frequencies and 1/f reduce noise.

I figured that if you have more current it means noise and other issues will have less of an effect, as higher voltages are easier to measure. I could add an auto-ranging feature by having it apply 10 amps for a few mΩ or less, 2.5 amps for under 30 milliohms, .5 amps for under 100 mOhms, and otherwise 50-100 mA.

So do you recommend using the same setup and ICs as a preamp here? Would I get similarly results? And is their a particular op-amp or op-amp setup that you recommend as the main op-amp?

 

You are right about the inductance introducing errors. Microhoms are pretty small! You can lower the frequency of the stimulus to any frequency needed to make inductance irrelevant but it will slow down the measurement, which might or might not be significant for your application.

If I may ask, what do you intend to measure and why? This is not the sort of thing that you will be measuring with a pair of hand-held probes!

 

You are right about the inductance introducing errors. Microhoms are pretty small! You can lower the frequency of the stimulus to any frequency needed to make inductance irrelevant but it will slow down the measurement, which might or might not be significant for your application.

If I may ask, what do you intend to measure and why? This is not the sort of thing that you will be measuring with a pair of hand-held probes!

I want to measure the resistance of coils and wires to calculate losses and other things. Also to figure out if something is CCA or copper, or some other material. There are many uses.

 

If heating of the device under test is not an issue you will do well with your idea of using a lot of current and a microvoltmenter.

 

I looked on amazon and their was one for 12 grand that used 10-110 Amps to get up to a resolution of .1 uOhms. More current is definitely the way to go. Anyways, I will be going on vacation for three weeks so I may not be too active here on AAC.

 

Also I may possibly add an overkill option with the three 2.7V 500F caps I bought. I could make it a CD pulse spot welder that additionally measures really low resistance. I could already use one for welding 18650 packs, so I may as well get a high current shunt. That would also allow for dead short protection which could damage the fets. But that is a project for later.

 

In my work I use a low resistance digital meter which has a display resolution down to 1mΩ - although I don’t require that level of accuracy.

The meter uses dc current and a 4 wire measuring system to eliminate the volt-drop of the test leads. Even at mΩ resolution, care has to be taken to avoid contact resistance significantly affecting the result.

In operation the meter displays a value once a ‘measure’ button is depressed – rather than continually updating the display. Multiple operation of the ‘measure’ button gives successive readings; with no change to the setup, it is not uncommon for the reading to vary by up to 10mΩ (for no obvious reason). Therefore I would expect to see a meter with µΩ resolution to suffer from a similar variability.

One other observation, passing 10A through a wire is likely to result in significant heating, causing a very significant rise in resistance (when measuring in the µΩ range).

 

One other observation, passing 10A through a wire is likely to result in significant heating, causing a very significant rise in resistance (when measuring in the µΩ range).

I only need +- 5% accuracy. That means I would only need to pass 10 amps through very low resistances. I do not need 10s of μΩ resolution when measuring something in the mΩ range. If it is a high current lug or wire, and just briefly, it should not heat up too much.

 

Here is a circuit proposal that just might work, powered from a lithium-ion battery (3.7V) capable of delivering 10A.

The resistance of interest is connected between terminals A & B.

Transistors Q1/Q2 are configured as a darlington, Q2 must be capable of handling the 10A.

Switch SW1 is a centre off switch such that the emitter current of Q2 is passed through R4, or R4 + R3 or R4 + R3 + R2 depending on the switch position. The resistance values (R2 – R4) are selected to give the required currents at each switch position.

Now all you need is a circuit that can measured μV levels between terminals A & B.

 

One other observation, passing 10A through a wire is likely to result in significant heating, causing a very significant rise in resistance (when measuring in the µΩ range).

And on top of that, that heating is almost certain to cause VERY large thermoelectric potentials-- hundreds of microvolts or even more-- that will completely obscure the small I*R voltage drops the TS is trying to measure.

Thermocouple effects are a notorious cause of measurement errors in low-level DC amplifiers, even without the self-heating that will be encountered in this application. Googling the phrase, "minimizing thermal errors in low-level circuitry" brings up tons of useful information on this, including links to the Keithley Low Level Measurements Handbook. Minimizing parasitic thermocouple effects is no trivial task even for an experienced designer, as it involves not just circuit design but PC board layout and careful selection of interconnect methods and materials.

Frankly, I am skeptical of the design goal of 10-20 μΩ measurement resolution as it seems neither appropriate nor necessary.

 

Thermocouple effects are a notorious cause of measurement errors in low-level DC amplifiers, even without the self-heating that will be encountered in this application.

Thus the reason for using an AC signal in such measurements (where possible).

 

I can try pulses of a few hundred mS. That way it doesn't really get to heat up too much, but avoids issues with inductive reactance. And I could also even try using a fan or something to cool it. Additionally, I would only need this kind of resolution on VERY low resistances, where 10 amps wouldn't heat it up too much.