I'm not sure what you mean by a regular T. An H-plane T doesn't have any phase difference between the arms. An E-plane T has 180 degrees. If the matches aren't perfect then reflections will change the phases if the network isn't symmetrical.

How about using hybrid Ts with loads on the fourth arm. These behave like a Wilkinson divider and damp down reflections.

Hi Malcolm, thank you for your reply. These are E-plane waveguide Tee-s, and I simply used waveguide ports, with Zwave characteristic impedance.

Other parameters behave like normal, I mean insertion loss, return loss... But the phase shift is not. Not a big thing, because I'm sure it will work when we build it in real life, but I would like to know what is the problem with HFSS.

It might not work in real life because simple T junctions don't absorb reflections and unless the whole thing is well matched then differences between reflections in different arms result in different split ratios.

I don't use HFSS so I can't help there. Have you matched the T junctions? I think just joining the waveguides together won't work, and there needs to be a post or wedge at the junction to get a match. I can't see why it shouldn't be 180 degrees, though. If I was modelling just one I would put an electric (PEC) wall down the middle anyway because of the symmetry.

I forgot to mention but the T junctions are matched.

Hi there,

Have you checked vector fields (E or H)? Are they correct?

Hi Andrei!

Yes, I checked it recently, here is the thing:

Basically, the fields are correct.

When I use it as a splitter the fields are correct, but when I use it as a combiner... well, they are correct, but I have to use "edit source" and correct the phases at the appropriate outputs. For e.g. I designed a 1to4 combiner for X-band, and all the 4 output ports were in-phase. Nonsense.

Are you sure the ports are the same way up? It is possible that the way HFSS defines ports, the excitation of opposite ports of a e-plane T could be automatically out of phase because the reference vector is pointing the other way.

Hey Malcolm,

yes, that's what I thought too, but even in this method, I can't get back the theoretical values at a 1to8 combiner.

I attached a picture. S21 is the first output port, S31 the second, and so on. So

the port 1,3,5,7 are in-phase, and port 2,4,6,8 are 180deg. (didn't want to picture in eight different pics.)

Correct me if I am wrong, but I think the phases have to be like this:

port 1,4, 6,7 - in phase

port 2,3 ,5,8 - 180deg.

(see the pic below)

Anyway, I really appreciate your replies.

I'm assuming you are using wave ports. You must confirm the integration lines on all ports are defined and pointing in the same direction. HFSS seems to arbitrarily assign initial phases to the ports if the integration lines are not defined.

I agree about the phases to expect. You need to check what Mohamed K. Emara warns about - I don't know HFSS.

Hello,

Some papers are attached for your help.

Thanks,

Dear friends,

thank you for all the replies. All the integration lines are on the same direction, and as I wrote above: "the port 1,3,5,7 are in-phase, and port 2,4,6,8 are 180deg.".

See my graphs and my comment above.

To my surprise, I have drawn out your circuit, putting in the polarization direction on each arm, and if all the bends are in the e-plane as you have shown, there is no difference in phase due to the e-plane "T"s, when the arms are bent downwards as shown.

In the arms straight after the "T" the fields are in opposite directions, but the 90 degree bends after this bring them back into phase because the two arms are rotated 90 degrees in opposite directions which is 180 degrees rotation difference.

Well, another strange thing. :)

I guess, if you finer the mesh, you might get other phase results.