Problem with simple branch line coupler design (multilayer circuit)

[mfarhan1]

dear all
i am working on multilayer circuits, i made a simple branch line coupler in stripline configration burried in dielectric, its working fine, but i need to specify connecting point so i put vias at ends and take it to upper ground which is visible, now the problem is the results are totally different and wrong, please any one having experience in multilayer circuit advice me,

 

[vfone]

Re: Multilayer circuit

In a branch line coupler design very important is the “Line of Symmetry”, where I=0 and V=max.
If your added vias changed the line position, the coupler performances would be affected.

 

[mfarhan1]

Re: Multilayer circuit

thanks but how could i know that the line of symmetry is not disturbed, i want to take the signal line out of burried strip line to external layer to make connection!

 

[Azulykit]

Re: Multilayer circuit

If I understand your problem correctly you have a stripline branch line coupler covered with conductive sheets top and bottom and you now have taken the feedlines to the upper conductive surface surface so you can make connections.
Is that surface line CPW? Consider carefully where the reference groundplane is and keep the line and spacing at the same impedance as the coupler input and outputs.

Another area to look for trouble is with stitching vias. As a first cut I would try to keep them spaced less than a tenth of a wavelength or so. You might also consider several rows. This comment may start a spirited discussion.

I would look at those connections very carefully. If you used vias are there unintended stubs on the lines? You may consider a blind via from the top to the buried conductor (and no further). The transition to the upper layer and a through via may have introduced unexpected reflections in your coupler.

If you have an EM simulator model the coupler and the traces. You may be able to see where things have gone astray.

Good luck.

 

[mfarhan1]

Re: Multilayer circuit

@Azulykit thats right you completly uderstood my problem, i think the top ground is pure copper sheet with a slot in it where i pass via and make a small CPW line for connecting port, the vias are installed on each end of coupler so they are at least lambda/4 appart i cant take them closer, why are you asking to close them by lambda/10? secondly why do you think i need more than one via? i have ADS momentum how can i see where things went wrong, i can use the post processing visualization i plot current now i am new to troubleshooting and dont know how to see the problem, If you used vias are there unintended stubs on the lines?no the design is stright forward with no extra stub, ( i dont understand what you want to ask about stub) any help in this regard would be appreciable.

 

[Azulykit]

Re: Multilayer circuit

I normally use HFSS and am not familiar with teh features of ADS.

I was trying to make two separate points about the via's. The via that you use to connect the upper driven trace (cpw) th the inner driven trace (stripline) should be cut at the inner trace (stripline) without leaving the end of the via hanging (through via). People usually don't like blind vias because they are more expensive to fabricate. Leaving a open stub on your transmission line will cause a suck-out in the bandpass. The longer the stub the lower the frequency of the suck-out.

My second comment was about stitching vias tying the upper and lower groundplanes together. think of them as converting the stripline and CPW lines into a more coaxial like transmission line structure. Those are the vias that I would place close.

In HFSS one can build this structure and watch the fields propagate through a device to see where standing waves occur and get an idea of where things need to be adjusted.

I missed your last post until just now. Hopefully this will help.



Regards

 

[mfarhan1]

Re: Multilayer circuit

thanks Azulykit, i would most probabbily using blind via, yes ADS can not tell about standing waves visually so i am thinking to simulate it on HFSS, so that i could figure out the problem

 

[Azulykit]

Re: Multilayer circuit

I guess there is some restriction about using the term suck-out. maybe I should clean up my vocabulary and say "a sharp decrease in s21" instead.........:?:

Added after 51 minutes:


Just another thought: before you build the complete model make sure both the CPW and striplines are at 50 Ohms. Once they are at the same impedance then attach the two and go from there. The structures are easier to de-bug is I am reasonably confident that the parts are working first.

 

[t2imoor]

Re: Multilayer circuit

I don't know whether this is the right thing to do... BUT, could you guys help me out here:-