S-parameters bisection

Hello everyone,

I am trying to use the guide described here. But after extraction of the bi-sected S-parameter, I get spikes:

Instead of the plain line:

If I cacade the bisected results, everything is ok. The spikes appear at the frequencies of the phase shift of the S-parameter:

I have tried to fix the problem in the following way:

s21_bisec = (s_obj.Parameters(2,1,:));
% Adjust s21
phase_desired3 = unwrap(angle(s21_bisec));
mag_S21 = abs(s21_bisec);
[x,y] = pol2cart(phase_desired3,mag_S21);
s21_bisec_desired = x+1i*y;
s_bisec.Parameters(2,1,:) = s21_bisec_desired;

But it also did not help. I am familiar with this topic discussing bisection and the github code at the end of the tread, but this code uses square root from the s-parameters, that seems to be not correct in a case of a cascaded network.

What can I do?

[Sherman Chen]

Any progress on this? Thanks.

Answer 1

[Stadder]

Did you find any solution on this topic? I have the same issue

[Sherman Chen]

It seems to me this bisec() function issue might be related to the matrix processing. You can see the similar issue when applying s2t, t2s for cascading. By applying the method from Vladimir Dimitriev Zdorov described in "Numerically robust, fast and accurate method of combining linear models of arbitrary topology into a single S-parameter model", I successfully fixed the cascading issue. The similar methodology might also apply to bisectioning but I have to leave it for some day when I get time to look at it.

[Stadder]

Thanks for your feedback, was your solution based on an optimized code or is this problem based on the orignal results/measurements?