Using Simulink's Averaging Spectrum Analyzer

Hi,

I built a model of a dynamic system in Simulink. Now I want to get its Bode diagram, using Averaging Spectrum Analyzer. The question is: how should I set values of block parameters to get proper result of a simulation?

I've tried many options, but none of them let me get a Bode diagram I achieved solving this problem analytically.

[Kaustubha Govind]

Which Blockset Library does this block belong to? I can't find anything from a documentation search.

[Michal Tomaszek]

Hi there, thank you for response.

You can find it in Simulink Extras->Additional Sinks or simply enter Averaging Spectrum Analyzer in search window and you'll get it.

I found it hard to get any info while searching the internet, so any help would be appreciated.

Answer 1

[Andreas Goser]

Maybe this is only a partial answer, but hopefully helps on the way to a full answer.

Some years ago, I learned that the blocks in "Simulink Extras" are purposely in this library and that it means they are undocumented and unsupported. I personally compare this with a situation where you download a free blockset from FileeExchange.

But what you always can do is to look under the mask of the blocks to see how they work. In this case, the block is actually an S-Function written in MATLAB Code:

edit msfuntf

And this code is documented and may answer your question:

%MSFUNTF an MATLAB S-function which performs transfer function analysis using ffts.
%   This MATLAB file is designed to be used in a Simulink S-function block.
%   It stores up a buffer of input and output points of the system
%   then plots the frequency response of the system based on this information.
%
%   The input arguments are:
%     npts:        number of points to use in the fft (e.g. 128)
%     HowOften:    how often to plot the ffts (e.g. 64)
%     offset:      sample time offset (usually zeros)
%     ts:          how often to sample points (secs)
%     averaging:   whether to average the transfer function or not
%
%   The spectrum analyzer displays three plots: the time history,
%   the phase and magnitude of the transfer function.

[Michal Tomaszek]

Thank you, I'll try to make that information useful.

Currently I'm trying to learn how to bring into play this block by using it in simple sysem. I know so far, basing on different approaches, that (which is pretty obviously, but I give you all I got) length of buffer and sample time are connected with simulation time by equation:

buffer length=(simulation time)/(sample time)

This let me to get all time response of a signal. Sample time has to be chosen properly. It depends on input signals frequency - the higher frequency is the smaller sample time has to be.

I'm looking for further answers. Hope we'll be able to solve this together.