• State Resolved
  • Replies 1 reply
  • Subscribers 46 subscribers
  • Views 32 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

ADC128S102: Datasheet does not spec noise directly?

May-Li
Prodigy 30 points
Part Number: ADC128S102

Tool/software:

Hello-

I noticed that the ADC128S102 datasheet does not spec noise directly for use by the Analog Engineer's Calculator TUE Calculator. Or if I missed it, could someone point it out??

What I ended up doing was computing the ADC noise for a given input voltage. Please let me know if this makes sense.

_______________________________________________

For an input voltage of 1.65V, rms value 1.167V. SNR 73dB @ 5V VA

Anoise,rms = Ainput,rms / sqrt(SNR)

                  = 1.167V / sqrt(73dB)

                  =0.136V

Does this make sense / seem right?? Seems pretty high.

Thanks!

-M

  • +1
    TI__Expert 5190 points

    Hi Madeline,

    SNR is defined as 20*log_10(Full-Scale Codes RMS / Noise Codes RMS)

    Algebraically, this can be represented as 10^(SNR/20) = Full-Scale Codes RMS / Noise Codes RMS

    The Full-Scale Codes RMS can be taken by the sqrt(2)/2 * Full Scale Codes. Since the ADC128S102 is a 12-bit device, Full Scale Codes = 2^12 = 4096. Therefore the RMS codes are sqrt(2)/2 * 4096 = 2896.309. 

    Since the ADC128S102 SNR is 73dB, with some algebra, we can say Noise Codes RMS =  2896.309 / 10^(73/20). This is equal to about 0.6484028 ADC codes RMS. So our noise codes RMS over our full-scale codes RMS is 0.6484028 / 2896.309, which equals 223.87e-6. Alternatively, this is just SNR = 20*log(1/x). Insert the SNR of the ADC and solve for x. 223.87e-6 represents a scaling factor to determine the ADC noise at close to a full-scale input. So with a 5V input, we take the RMS. This is equal to about 3.536V RMS. Multiplied by 223.87e-6, and you get 791.5uV RMS worth of noise.

    Let me know if I can help explain this further. At the end of the day, this is the fundamental nature of logarithms, and all we are doing is comparing the magnitudes of a full-scale signal to the measured noise, thus signal-to-noise ratio (SNR).

    Regards,
    Joel