# What is the advantage of a current shunt ic versus a differential ADC across the shunt?

I'm trying to measure DC 25A across a .005 ohm shunt resistor. From my research I've found the TI INA169 which is a current shunt monitor ic. However the datasheet says it has a total output error of ~1%. At 25A, this would be .25A. I'm trying to measure to 0.01A accuracy.

With 25A through the resistor, that is 125mV. If I had a differential ADC like the ADS1115 (16-bit, 1 for sign and 15 for value), then if I had it measuring the voltage differential between the shunt would that not be a more accurate solution?

With the INA169, what is the real accuracy I could expect if I measured the output with the ADS1115 in a single ended configuration? Would it realistically be more than 1%?

• One chip is an ADC, the other produces a current proportional to the input voltage differential. Two totally different applications, not sure why you try to compare them. Jul 4, 2016 at 10:12
• With the current sense amplifier chip, I would measure the output with an ADC. However with the differential ADC, I can just measure the voltage differential of the shunt resistor without the chip. I'm wondering if that is actually possible. Jul 4, 2016 at 10:14
• Instead of "throwing" a more accurate (is it really more accurate ??) component in your design, you need to think about the tolerances in the complete system. The INA169 is a completely different component than the ADs1115 so how can those be compared ? You should learn about tolerances and accuracies first before you can have a meaningful discussion about these. You can build a better than 1% setup with the INA169 if you calibrate it. Jul 4, 2016 at 10:17
• 0.01A accuracy over 25A is 0.04%. So, to ensure this, the first thing that comes to my mind is the tolerance of the shunt resistor. Did you actually find a 5mOhm resistor with 0.04% tolerance? I'm not sure what you want to achieve is realistic.
– dim
Jul 4, 2016 at 10:18
• You do not need overall 0.04% intrinsic tolerance/accuracy as long as you can calibrate. Also component drift can be calibrated out but when using components with a high drift (they change properties over time) you will have to calibrate more often. Jul 4, 2016 at 10:20