ADS Momentum stacking multi layer and single-layer component simulations

[usx]

I'm simulating a 3-layer board (energy is coupled by a 'waveguide' from layer 1 to 3) followed by a coupled line bandpass filter.
I've simulated coupler and am satisfied with the results. I've also simulated a bandpass filter (albeit, on a single-layer board) and am satisfied with it. I stack them in ADS schematic (coupler and bandpass filter share a ground) and S-parameters add linearly. Next, when I copy and paste those two structures together, I get distorted signal, even though input and output impedances and matched to 50Ω relatively well.



I expected above two circuits give me identical response but it wasn't the case. I don't expect there to be crosstalk between the layers as they are well isolated, so who wouldn't response be identical? Does it have anything to do with how ports are set up in Momentum?

 

[Hamid.Kiumarsi]

Hi,
Show the substrate definition for your single-layer and multi-layer boards.
One possibility is that since the surrounding environment of circuits is changed effective dielectric constant is also have been changed.

 

[usx]

Hi,
Show the substrate definition for your single-layer and multi-layer boards.
One possibility is that since the surrounding environment of circuits is changed effective dielectric constant is also have been changed.

That's what I'm thinking, too. I used the same substrate definition for all structures.

 

[Hamid.Kiumarsi]

If you have used the same substrate definition for all of your circuits, there shouldn't be the problem that I mentioned.

 

[usx]

ADS Momentum recommended substrate layer dimensions (x, y) for multi-layer design?



I have't considered checking it but but now that I have, turns out frequency response of my circuit is depended on substrate length and width.

I attached the a picture showing what happens when I only changed substrate dimensions.
What are the recommended dimensions of a substrate for EM simulation?

- - - Updated - - -

A little clarification on my post. What I meant by "substrate layer" was the ground plane layer. I do believe simulations are bounded by it as dielectric layers labeled with "infinite width" according to stackup definition.