• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 79 subscribers
  • Views 1442 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Comparing Two ADC performances (SD16_A and MCP3901)

BasePointer
Expert 1570 points

Hi,

I want to compare two ADC performances provided by TI and Microchip for energy measurement purpose. But I dont know how to compare. I need more accurate measurement with more bandwidth. Is comparing SINAD parameters enough?

 

Thanks,

BP.

  • 0
    Guru 227245 points

    SINAD is surely not the only parameter (don't forget, a gain of 32 adds another 30dB to the SNR which more than compensates for the decreased SINAD). Impedance, distortion, input range etc. are other things. And the handling.

    One thing that's great with the MSP is the simultaneous capture of all channels (depending on the MSP) and even the adjustable conversion delay to offset the angle error introduced by the current transformer.

    And personally, I don't like microchip. Their PICs are a pain in the a.. to program, and I don't want to try their other stuff if I don't have to.

  • 0 in reply to Jens-Michael Gross
    Expert 1570 points

    I completely agree with you about their pics :) But unfortunately they provide very aggressive offer that cause a pressure on us comes from top, so we need some technical facts to attenuate such as salvo. For the same SINAD value, it seems they are faster than TI. But I'm not sure I do really need high sample rates. Linearity is important for me, I mean the ADC that I need should properly measure a current from 1mA to 100A. But I'm far enough to ADC terms so I couldn't decide if I need to increase OSR which means decreasing sample rate or increasing sample rate which means decreasing OSR and SINAD. Actually I dont need to measure to many harmonics, up to 10th harmonics is highly ok for the standard which means 50/60Hz*10 = 600Hz * 2 Nyquist = 1.2 ksample/sec. I wonder If I apply pure sine signal to the meter, is there any relation between accuracy and sampling frequency? For example I applied 50Hz pure sine signal and I used 1khz sample rate and 4 Khz sample rate with same OSR. Will the result be exactly same, or faster one will be more accurate?

     

    Thanks.

  • 0 in reply to BasePointer
    Guru 227245 points

    BasePointer said:
    I mean the ADC that I need should properly measure a current from 1mA to 100A.

    That's 17 bit resolution. And 10ppm precision.
    Well, the SD16_A module, which allows simultaneous sampling of up to 7 channels, offers more than 16 bit bresult. Since teh SINAD is related to the signal, with 1mA you'll get 16 0-bits and then the significant bits follow. I don't kow about the Microchip, but on the MSP, you'll get (at least with an OSR of 1024) more than 16 bits result.
    With only 16 bit output, the maximum resolution is 1.5mA (absolute value, including noise and jitter) if 100A is your full-scale signal. If you don't scale the signal so that 100A is exactly the reference voltage, then your resolution is even worse.

    E

    BasePointer said:
    I wonder If I apply pure sine signal to the meter, is there any relation between accuracy and sampling frequency? For example I applied 50Hz pure sine signal and I used 1khz sample rate and 4 Khz sample rate with same OSR. Will the result be exactly same, or faster one will be more accurate?
    Sure. The samples are just points on the waveform. The more points, the more exact your reproduction of the sine wave will be.
    But if you KNOW that you're sampling a pure sine wave, then you can reconstruct the signal by both sampling frequencies equally well. Just more calculations needed, as you cannot simply interpolate between the sample points. As long as you sample faster than with 100Hz, you can (with the appropriate math) 100% reconstruct the frequency and amplitude of the (50Hz) sine wave.
    If it isn't a pure sine wave, faster is better.

     

  • 0 in reply to Jens-Michael Gross
    Expert 1570 points

    There is an graphics shows SINAD vs OSR in the datasheet. When OSR=1024, it seems SINAD is 90db. According to my calculation ENOB should be (90db-1.76)/6.02 = 14.6 bit.  That means my scale is 1:25834. So if I adjust my full scale to 120A, each LSB will correspond to 4.65mA. If my calculations are correct, I need to think..

     

    Thanks,

    BP.

     

  • 0 in reply to BasePointer
    Guru 227245 points

    BasePointer said:
    So if I adjust my full scale to 120A, each LSB will correspond to 4.65mA.


    Not exactly. At 120A you have a useful resolution of 4.65mA. But if the measured signal is lower than full-scale, then some of the bits which are noise at full scale, become trustworthy information.

    The MSP SD16 offers more than 16 bit at OSR=1024, so at a smaller signal, you can have a better resolution.

    It makes no sense telling 'current current is 119.756A', yet it might be possible to tell that the current current is 0.756A. The SD16 offers this without knowing the actual size of the signal and preparing a prescaler/gain before the conversion.

    See the other thread about the SD16 module where we tried to figure out how this all works. It's only a few days old.

     

**Attention** This is a public forum