• State
  • Locked Locked
  • Replies 8 replies
  • Subscribers 49 subscribers
  • Views 269 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.

Noise Calculations in a circuit.

A Banerji
Prodigy 40 points
Other Parts Discussed in Thread: TINA-TI

How can I calculate the noise of a circuit with several opamp stages cascaded...I have followed TI's precision lab videos for noise calculations but it's not clear from those what to do when there are several opamp stages cascaded together. Can some one please just list out the steps I should follow for calculation of noise of this circuit, it would be really really helpful. I have looked all over but I couldn't find any source that gives information on how noise adds up in a circuit such as this. I have attached a pic for reference. 

I am basically confused about whether I should go stage by stage, as in consider noise of one stage as input to other, multiplied by the gain, if any or consider the entire circuits overall gain and bandwidth for noise calculations. In that case, how do I add the thermal noise and opamp voltage noises? Any help is greatly appreciated. 

Please list out the steps I should follow for noise calculations.

Thanks in advance

 

  • 0
    Guru 140200 points

    Hi A,

    have you thought about running a TINA-TI noise analysis? This would heavily simplify the job :-)

    Kai

  • 0 in reply to kai klaas69
    Prodigy 40 points

    Hello,

    Thanks for replying!

    Yes I have done that...however I would like to know how to manually calculate  the noise using the formulae and specified noise values in the data sheet of opamps used. I am required to do this for a project I am a part of, and it will be really helpful if I am able to figure out a standard procedure for noise calculation in a complex circuit which can be used in the future for other circuits we will be using as well.

    Regards,

    A

  • 0
    Guru 48395 points

    Total output noise is one thing, but really after the 1st stage, it is SNR you care about. The signal and noise at the output of the first stage will get the same gain going forward with each stage adding a little bit of noise to what comes in. 

    You can certainly take the output noise of the first stage then RMS that with the input referred noise of the subsequent stages and pass on with those stage responses. 

    Tried to work through this kind of thing in this article, 

    https://www.edn.com/controlling-spot-and-integrated-noise-in-a-two-stage-transimpedance-design/

  • 0 in reply to A Banerji
    Guru 140200 points

    Hi A Banerji,

    noise calculations can be quite sophisticated. In the datasheet of OPA101 is shown a nice example:

    Kai

  • 0 in reply to Michael Steffes
    Prodigy 40 points

    Hello, thanks a lot for your suggestions!

    I have accurately calculated the noise of the individual stages, they are pretty close to the noise obtained from simulations. However taking rms of output noise with input of subsequent stages and so on does not seem to give accurate results. I have done it as follows: (Noise of stage 1)*(gain of stage 1) rms with noise of stage 2(calculated individually)  and then multiplying that with gain of stage 2 and taking rms with stage 3 and so on. I am not sure if my approach of calculating noise for each stage individually first is right or if something changes when several stages are cascaded together.

    Regards,

    A

  • 0 in reply to kai klaas69
    Prodigy 40 points

    Hello Kai,

    Thanks for the suggestion!

    However, the example shows calculations only for one amplifier, which has not been a problem so far. I am unable to obtain a suitable procedure for noise calculation when several amplifier stages are cascaded. There seem to be no examples of this available anywhere :(  

    Any suggestions on how to proceed are greatly appreciated. Also, my only main source of reference has been TI's precision lab videos on noise calculations, but those don't give any hint on how noise will behave if additional stages are present, unless I am missing something. 

    Regards,

    A 

     

  • 0 in reply to A Banerji
    Guru 140200 points

    Hi A,

    you must think in terms like noise voltage density which is frequency dependant and frequency bands as demonstrated in the link I gave you. When you have calculated the noise voltage density versus frequency "Un1(f)" (noise voltage density spectrum) of the first stage, you must take this "signal" and run it through all the succeeding amplifier stages. In these stages the noise voltage density spectrum gets additionally filtered, amplified and attentuated. You end up with the noise voltage density spectrum "Un1'(f)".

    Now you take the second amplifier stage, calculate the total noise voltage density spectrum at its output "Un2(f)" and run it through all the succeeding amplifier stages as well. You will end up with the noise voltage density spectrum "Un2'(f)".

    You do this with all your four amplifier stages and additionally get "Un3'(f)" and "Un4'(f)". At the end you take the sum over the squares of these noise voltage densities:

    Un_tot(f) = SQRT ( Un1'(f)^2 + Un2'(f)^2 + Un3'(f)^2 + Un4'(f)^2 )

    But this is only the noise voltage density versus frequency. The total noise voltage (RMS) you will get after integrating over this noise voltage density "Un_tot(f)" as shown in the link I gave you.

    When you calculate the total noise of each amplifier stage treat the driving output as virtual ground. That's not quite true because the output impedance of an OPAmp is also frequency dependant. But this has consequences only at very high frequencies. And don't forget the source impedance in the first amplifier stage.

    Kai

  • 0 in reply to A Banerji
    TI__Genius 10850 points

    Hi A,

    We haven't heard from you in a while. We hope the issue was resolved. Closing this thread now. Reply to re-open if you have any additional questions.