Dear Gyu,

welcome,

I see only in the figure S21 which shows ripples that may be due to mismatching at the input and output ports. So, before making any design changes you have to be sure of input and output matching at all required frequencies. What is the expected maximum value of S21.

Best wishes

HI,

First of all, you need to double check of that the exploited waveguide ports used to feed the GCPW is providing a 50 Ohm characteristic impedance; otherwise, the simulation results are totally inaccurate. This can be simply performed by manipulating with the waveguide's dimensions since there is no build-in CST solver which you could utilise to calculate the proper width and height of the waveguide. The attached file is useful for such a circumstance.

All the best

The ripple is almost certainly due to a reflection at the start of the taper at each end of the line. The ripple has period roughly 4 GHz, so is likely to be caused by reflections 125 ps apart (half a period). In alumina (roughly 100 mm/ns) this will be about 12.5 mm apart. This coincides with the distance between the two tapers, so the tapers (or the start of them) are almost certainly to blame. There modulation is caused by a -22 dB down signal, so the reflection at each end may be at about -16 dB.

Abdelhalim Zekry

Dear Prof. Dr. Zekry,

Thank you for your answer.

I want to have as same a S21 result as the attached file.

Honestly, I'm not quite familiar with electrical port system.

Would you please explain me the detail of "matching the ports"?

Sincerely,

Gyu-young Park

The CST waveguide ports that you have used need to be the same impedance and have the same mode content as the transmission line they go into. Then they are matched. If you have made the ports big enough this will probably be ok. If the ports are not matched then there will be a reflection at each port which will cause ripple. It could be this, but I think the reflections are probably closer together than the distance between the ports. You can find out by making the parts next to the ports longer and seeing if the period changes.

You did not show S11 and S22 at the connection between the feed lines and your wave guide tapers. It is important to see how far your device is matched at its input and output.

According to your targeted S21 curve you don not allow any reflections which requires perfect matching.

Best wishes