Multisim and Ultiboard

Showing results for 
Search instead for 
Did you mean: 

How can I isolate individual noise sources in a noise analysis simulation?




When I do an noise analysis simulation I can get a total RMS value for the band simulated, or I can get a spectral plot.


Is there any way to identify each noise source in the simulation and it's contribution to the overall output noise?


I tried a sensitivity simulation and selected all variables for output and that may have given some of the info I am looking for.  Is that correct?  Or does it not include any of the noise sources that are present in the noise analysis?



0 Kudos
Message 1 of 9

Hi dbur,


What do you mean with that?


Isn’t the option on the tab Output what you want to configure?






0 Kudos
Message 2 of 9

This would give the noise at those nodes rather than the desired output, would it not?

0 Kudos
Message 3 of 9

Yes, this would return the noise on all of those nodes. Isn't that what you were looking on the first place?


Maybe there is something that we are not understanding, can you give us more details, so we can try to help you better?



0 Kudos
Message 4 of 9

Noise at the specified output node is the RMS sum of the contribution of all the noise sources multiplied by their respective gains between the sources and output.  To zero in on the most significant noise contributing sources you need to be able to see that summation before it just becomes one RMS number.  This could be done by turning on/off each noise source and re-running the simulation.  This would actually be a very easy thing to provide for the developers who have access to the source code that runs this simulation.


I don't see a way to do that as it is, or to get any other form of a report of the various noise source contributions to the output.  -Unless selecting each noise source for output does that, but I would like to confirm if this is true or not.  What I don't want is to get the total noise at each of those source nodes as if they were the output instead of at the actual output node desired.  Is that a correct understanding of how this is implemented?


I'm hoping I am wrong and the output will be as seen at the actual circuit output rather than at each source.  That would be the useful way to have it work.


Now that I think about it I suppose I could create a simulation to test this, if there is no answer from other sources.



0 Kudos
Message 5 of 9

OK, I did a check. 


I selected all the sources for output. 

I copied and pasted the results in a spreadsheet. 

I RMS summed the results for each individual source.


This should match the noise_total value from the simulation.


It's not even close.  It's about 10x lower that the simulation result.   So what does this data actually represent?


0 Kudos
Message 6 of 9

Hi dbur,


Thank you very much for your reply, we really appreciate your question.


Multisim performs the noise analysis calculating the contribution of every resistor and semiconductor that can be considered a noise generator. It does it by taking into consideration the transfer function.


Regarding your question, I am not very sure that you will be able to obtain noise information that is delivered from each power source to each component that you would like to analyze. I would like to show you several links that might give us a little more information about this and other noise analyzing tools in Multisim:

Configuring a Noise Analysis in Multisim

Noise Analysis using Analog Devices Operational Amplifiers

Configuring a Noise Figure Analysis in Multisim



I hope this information might help us.




0 Kudos
Message 7 of 9

The first article gives a good overview and implies the analysis treats each individual noise source as I would hope it does.


The problem is still that if I look at the output from all noise sources and sum them RMS the answer does not match the total noise computed by the MS simulation.


This is a big problem and I would like to know why this is happening.  I will still devise some more direct tests when I get a moment.



0 Kudos
Message 8 of 9

Delayed reply, but an explicitly desigedn and simulated circuit to test this behavior showed that in this case it behaved as expected.  All independent contributors to the output noise were provided and they did sum properly to match the total noise.  I don't know why I had bad results with my first circuit.

0 Kudos
Message 9 of 9