Current measuring - which approach?

Discussion in 'The Projects Forum' started by illusive, Jan 7, 2016.

  1. illusive

    Thread Starter Member

    Jul 9, 2015
    46
    0
    Hi all,
    I'm building a variable power supply and i'm currently working on the current measuring system. I need to measure 0-3A fairly accurate, down to mA. In other words i need 3 positions after the decimal point (for example 2.455A). The value will be displayed on a LCD screen so i will be using a AVR microcontroller. My idea is to pass the current through a shunt resistor and i want the maximum voltage drop on it to be no more than 20-30mV.Than use a low offset voltage op amp (offset in the microvolts range) to amplify that voltage drop to some usable value for the microcontroller ADC. The problem is that im don't know if the microcontroller ADC will have good enough resolution and accuracy for the job. It's a 10bit ADC, so 1024 steps. Another way will be to use an external ADC chip with a 16bit resolution. What do you think of that ? Is there a better way? I was looking at hall effect current sensors like ACS712, but they don't have the accuracy for the mA range (1.5 percent error). Are there any other prebuild chip solutions? I don't really need spot on accuracy, let's say -+1-2mA is ok.
     
  2. #12

    Expert

    Nov 30, 2010
    16,257
    6,762
    So, you want to build a digital meter with accuracy in the range of 1/2 of one tenth of a percent from the viewpoint of an amateur. I would have to do some Googling to find out if it is even possible to buy that quality from the professional meter makers. So, my opinion is, "no".

    The math is entirely doable, but the practical application stretches my imagination.
     
  3. illusive

    Thread Starter Member

    Jul 9, 2015
    46
    0
    If it is too complicated task to do i'm willing to let go on the 3rd digit after the decimal point. Anyway, can you recommend me a low input offset voltage op amps (lets say no more than 100uV) that can go to 0V output with single supply (i.e the op amp must have a common range that includes negative voltage or ground, i'm not sure if that means that it is rail-to-rail op amp). I'm on a budget so i can't spend more than 4-5 bucks for a single op amp.
     
  4. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,518
    1,247
    Unless your 0.01 ohm current shunt resistor has a tolerance of 0.1% or better, even the 2nd decimal place will have little meaning. Measuring multi-amp current accurately is a difficult design task. Work up what is called an error budget, calculating the output effect of all sources of error, and see what your real output accuracy is. Shunt tolerance, opamp offset errors and their temperature drifts, gain errors due to feedback resistor tolerance, common mode rejection limitations, A/D reference tolerance, A/D quantizing error, A/D non-linearity errors... It's quite a list. I'm not trying to be discouraging, but you've selected a very non-beginner-level project.

    ak
     
    #12 likes this.
  5. #12

    Expert

    Nov 30, 2010
    16,257
    6,762
    I can imagine it (or, I saw one), therefore I can design and build one, is the battle cry of the uninitiated. There are physical limits to all measuring systems. They have been explored. That's why the math is entirely doable. Look to the experts to find the State of The Art. Study them, learn from them, and especially, notice the price required to perform 4 or 5 digit accuracy, then back down your expectations by at least a power of ten for an amateur project.;)
     
Loading...