Current sense resistor

[jacobchanjal@gmail.com]

I need help in choosing a current sense resistor for current measurement in an embedded system. Can anyone please help me?

[TriodeTiger]

What are the current ranges you wish to measure? 1uA? 100mA? 1A? Is accuracy or minimal voltage drop important to you?

Alexander.

[Rerouter]

a few points,

how much current are we talking? uA, mA or A, and what range do you want to measure
what accuracy, e.g. 1mA worst case error, (precision comes down to your adc,)
and finally how much of a voltage drop / heat waste can you handle,

[poorchava]

What you need to know:
-what current range are you going to measure?
-what voltage drop can you accept?
-what accuracy do you need?
-what voltage range can your measurement circuitry accept (opamp, adc)
-ac or dc?
-do you see a possibility of short-circuit (eg. voltage drop across shunt = output voltage)

Now come the calculations:
the only formulas u need to know are: P=V^2/R=I^2*R and R*I=V. Example requirements: current range 0....2A, max drop across shunt 0.2V, 1mA of minimal resolution.

So, maximum allowable voltage drop of 0.2V occurs at 2A current across 0.2V/2A=0.1R resistor. So far so good, it's a realistic value. Maximum power dissipation at these conditions will be 2A^2*0.1R=0.4W. Realistic value, but standard 1/4W or 1206 smd resistor will not do. If whole device doesn't heat up (or is well cooled) 1/2W resistor will do, otherwise aim for 1W. Resistor tolerance as good as you can get, with as low as possible thermal coefficient. 1% resistors are common, 0.5% and lower are more expensive and less available. Best you can get would be a foil resistor, but these are very expensive. More attainable alternative would be a precision thick film resistor (easily available).

Next: we have maximum drop of 200mV at 2A, so 1mA of current translated into 0.1mV. Most common adc reference voltages are 2.56 and 4.096V (easy binary calculations). 4.096V gives LSB=1mV with 12 bit ADC. 2.56 gives 0.625mV at 12 bits. So we would want to amplify the current drop to utilize the biggest portion of full ADC range, which would be 4V. So we want to amplify the voltage across shunt 20 times. This can be easily accomplished with precision opamp (like OP07). If we sense at high side, we need to build a differential amplifier (google it), and if it's low-side sensing (one end of shunt connected to GND) simple non-inverting amp will be enough. More proffesional approach would be to use an instrumentation amplifier (like AD620 for example) or specialized current shunt amplifier.


Anyway, now your 1mA of current translates into 2mV, which is pretty reasonable and gives you 0.5mA resolution with 12bits ADC@Vref=-4.096V. Of course there are some pitfalls like noise and pcb design, for 1mA resolution to have sense (no point in such high resolution if you have some 5mV of noise coupled from other line into your amplifier input)