• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 47 subscribers
  • Views 436 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.

Is the noise of op-amps Gaussian/mean=0?

Joel772
Prodigy 30 points

I understand that white noise is not always Gaussian. In general, is the Vpp/Vrms noise stated in operational amplifier data sheets Gaussian/has a mean of 0? Any exemptions to be aware of while selecting an operational amplifier? 

Guess for interpreting the Vpp values:  x V of noise is a Gaussian curve with values ranging from +x/2 V to -x/2 V. A 10 V of Vpp noise means that the noise is a Gaussian curve with a +5 V max value on one side of the Gaussian curve and the -5V on the other side. Can it be more than x/2 on either the positive or the negative side? The Vpp values that are given in sheets are within how many sigmas of the distribution?

  • +1
    TI__Mastermind 27360 points

    Hi Joel

    We actually have a really good video covering the different types of noise. It can be seen here: https://www.ti.com/video/series/ti-precision-labs-op-amps.html under the spectral density video. 

    Joel772 said:
    In general, is the Vpp/Vrms noise stated in operational amplifier data sheets Gaussian/has a mean of 0? Any exemptions to be aware of while selecting an operational amplifier? 

    As far as I'm aware of, there's no exceptions. Noise will have a Gaussian distribution centered around zero. Except popcorn noise. Popcorn noise usually is a sign of defects in a device and are a whole other topic. 

    Joel772 said:
    Can it be more than x/2 on either the positive or the negative side?

    Pulling a slide from the presentation I linked, you can see that there's a difference in the max and min. However, because 1/f is very low frequency noise, if you had a longer time sample, the gaussian curve would be closer to zero. 

    Joel772 said:
    The Vpp values that are given in sheets are within how many sigmas of the distribution?

    The typical values are +/-1 standard deviation, or about 68.3% of the distribution. 

    Is there a reason you are concerned if the noise is more positive or negative? 

    Best Regards,

    Robert Clifton 

  • 0 in reply to Robert Clifton56
    Prodigy 30 points

    Hello Robert,

    Just considering how much of the noise can be averaged out and calculating the numbers of samples needed to be averaged to get to as close to zero as possible. The goal is to produce 1 sample by averaging many samples such that in the samples produced by averaging, the variance due to noise is reduced to 0.01 microvolts(in the original values before amplification). The input values are in microvolts. The random sample-to-sample variation is the concern, we can ignore any constant positive/negative offsets. 

    Any suggestions? 

  • 0 in reply to Joel772
    TI__Mastermind 27360 points

    Hi Joel,

    Noise can be tricky to predict. Starting with a low noise part would help. Here's some questions for you: 

    - Do you know what frequencies are important to you?

    - What part numbers were you thinking?

    Best Regards,

    Robert Clifton 

  • 0 in reply to Robert Clifton56
    Prodigy 30 points

    We are just considering which specific op-amps. Will you be able to provide more information or run tests if we provide the specific part numbers?

  • 0 in reply to Joel772
    TI__Mastermind 27360 points

    Hi Joel,

    What kind of tests are you wanting? I would recommend running simulations to compare the devices noise floors. 

    Best Regards, 

    Robert Clifton