Calibrating a Current Shunt

[misterT]

I have this design for current measurements around uA to mA range.. maybe in the future few amps. Nice IC-chip amplifier for this purpose (INA21x). I have a relay to switch between different shunts to little bit change the range.

The problem is calibration. Could there be a design that I could switch on a relay to a precision current source. And what kind of current source would you recommend.. is there an IC for that, or what would be a good way to calibrate this thing? I have a DAQ with analog and digital IO, so using that I hope the system could be self-calibrating.

 

[danadak]

To what precision/accuracy are your goals ?

Quite easy to create a precision current source/sink. OpAmp based comes to mind.
Using a precision Vref part and a precision R to set current.

How to Create a Precision Current Source | Maxim Integrated

The purpose of this application note is to describe how to create a current source (or sink) that is precise, highly flexible, yet inexpensive and simple to implement. The DS4303/5, Voltage Sample and Infinite Hold IC is a unique solution to create

www.maximintegrated.com

https://www.ti.com/lit/an/snoaa46/snoaa46.pdf?ts=1658483466665

Regards, Dana.

 

[crutschow]

You need an accurate 4-terminal shunt resistor, and an accurate voltmeter to measure the voltage across the shunt, to provide an accurate current.
What accuracy do you want?

 

[Reloadron]

I have this design for current measurements around uA to mA range.. maybe in the future few amps. Nice IC-chip amplifier for this purpose (INA21x). I have a relay to switch between different shunts to little bit change the range.

The problem is calibration. Could there be a design that I could switch on a relay to a precision current source. And what kind of current source would you recommend.. is there an IC for that, or what would be a good way to calibrate this thing? I have a DAQ with analog and digital IO, so using that I hope the system could be self-calibrating.

As mentioned much of what you are asking depends on the accuracy you expect to achieve.

I can add the following. Using a current shunt to measure currents down in the uA regions is not easy and is not practical. Here is a typical 4 wire current shunt.

About the lowest value current shunt I have used and seen is about a 10 Amp 50 mV shunt. Keep in mind the mV output of shunts is a very low voltage. Switching between shunts is not normally done using relays unless you have relays with gold or silver plated contacts with very low contact resistance. Normally high quality switches are used with gold plated or silver plated contacts. Switches which look like this.

The contacts are usually treated with conductive silver paste. The image is of a precision decade resistance box but you get the idea as to very low resistance contact switching. Hard to see but there are multiple fingers spring loaded on each contact.

Consider even with a very good precision shunt with uA current the voltage out will be way down in the dirt. A matter of doing the math.

Next consider the A/D even after the signal is amplified. Figure a 10 bit A/D (1024 quantization levels) with a precision 5.0 volt reference can only resolve 4.88 mV and that does not take into consideration noise. This is where you start looking at 16 bit A/D.

This is why I say measuring uA is not an easy task. Especially if you want to do it accurately. For calibration there are precision current sources or run your current through a known precision resistance and accurately measure the I * R voltage drop.

Again this all goes back to the uncertainty you can live with.

Ron

 

[danadak]

If you are using a processor in design there are processors out there that
have high precision A/D, 20 bits or more, that can do differential measurements,
even ones that can do this outside of the rails to 100 mV. So eliminate the
INA21X. Note has analog mux onbaord as well, can easily to 16 channels,
I think I did one one with 48 channels. Get rid of your relays possibly. Single
chip, see whats on it below.

A lot more stuff on this chip :

Regards, Dana.