• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 47 subscribers
  • Views 356 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

SAR Noise-Free Resolution with DC input other than GND

Robert W McKellar
Prodigy 90 points

Hi guys

The standard method for measuring the noise free resolution of an ADC is to ground the input and then plot a histogram of the output codes to find the standard deviation.

What I have found is that if I use a different DC input voltage, say VREFH/2, the measured standard deviation increases roughly linearly with the applied DC voltage.

Can anyone help explain why the count variance increases as the DC input is increased?

I have observed this effect on the SAR input of an MSP430, but also on a Sigma Delta converter.

Regards

Bob

  • 0
    TI__Guru* 92655 points
    Bob,



    The effect that you are seeing is likely a problem with a noisy reference.

    For any ADC converter, it is comparing the input signal with a known reference voltage to get a digital output code. If this reference voltage is noisy, then the input signal measurement will reflect the noise seen in the reference.

    As an example, the lowest noise you see is when the input is 0V, that is because regardless of the reference input, the output code will be 0 compared to the reference. Let's just say that if you have a reference of 2V and an input of 1V. As the noise varies the reference up and down, the comparison of the input signal to the reference will be basically 1/2 of the full scale defined by the reference. In this case the noise is also basically divided in half as well. As you increase the input to close to 2V and compare it to the reference, all of the reference noise now is reflected in the digital output code that you would get.

    For a SAR, the input is compared to the reference over a small interval of time. In a Delta Sigma converter, the sampling may be averaged over a larger interval. The latter may be slightly more tolerant depending on the amount of oversampling.

    Regardless, if you are having problems with the measurement noise and it is proportional to the input signal, I'd clean up the reference noise.



    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Hi Joseph

    Thank you for taking the time to reply.

    I suppose I could check if the measured noise is coming from the reference by replacing the reference with a battery?

    Regards

    Bob

  • 0 in reply to Robert W McKellar
    TI__Guru* 92655 points
    Bob,


    That would be a good idea. I don't know exactly how big the voltage noise is on a battery, but it's generally very small. One other thing would be to consider how much noise there is in whatever you were using as the reference before. At least you could get a ball park figure on it.


    Joseph Wu
  • 0 in reply to Robert W McKellar
    Prodigy 110 points
    You could also try making a LPF with a very low cutoff frequency to reduce the noise from the reference.
  • 0 in reply to Joseph Wu
    Prodigy 90 points
    Hi Joseph

    I think you are correct - the observed increase in noise as the ADC input voltage is increased comes from the reference voltage. In my case, I was using the reference to drive a bridge sensor to get a ratiometric ADC reading. As the sensor stimulus is increased, the bridge present a greater proportion of the reference to the ADC. The ADC reading may be designed to be ratiometric, but the noise from the reference on the bridge is uncorrelated to the noise from the reference at the ADC reference pin, so the noise powers add.

    I need to decrease the noise on my bridge driver circuit.

    Regards

    Bob