# Convert complex number to phase shift in simulink

I am currently modelling some dynamic phenomena, combining pure simulink blocks and simscape blocks.

I have measurements (real-valued) on which I perform computations with complex-valued matrices, which give me a complex-valued output.

However, I use this ouput in a feedback loop so I have to convert it back to some real signal for the simulation to run.

My idea was then to convert this complex value that has a relative phase to my input signal into some real signal with same amplitude, but showing a time shift matching the argument of my complex number, but I can't figure it out.

Any suggestions ?

I know of the transport delay / time delay (though I am not sure how to use them in such a situation), but it feels overkill for something that looks so basic. Moreover, I need to perform some linear analysis for stability on this model, which seems to be incompatible with such delay blocks as proposed in shifting signal in simulink.

What is the right way to do this ?

• This is unclear. Is the output a sample sequence where samples are complex numbers? What does the index number of the sequence present (=are the output samples in time or in frequency or in what domain)? May 14, 2020 at 12:06
• @user287001, the signals are continuous, there is no numeric processing per se, so no time sample except for the computation step. All signals (input and output) are in time domain. May 14, 2020 at 12:15
• Your comment, “...convert this complex value that has a relative phase to my input signal into some real signal...” makes me think of correlation. Here is link. See section 3.2.3 May 14, 2020 at 12:18
• @relayman357, thanks for the link. Though it presents an interesting approach, I do not think this is applicable here. In fact, for the correlation to work it seems one has to know the actual frequency of the signal, which is in my case a consequence of the dynamic phenomena I am studying. May 14, 2020 at 12:28
• @BambOo Correlation will let you determine the frequency too. But, this still may not be what you need here. See section 3.2.4 May 14, 2020 at 13:00