width of ground plane for grounded coplanar

[ge]

Two questions that has always bothered me...

1) volker_muehlhaus and others has shown em plots where the current/field rapidly decreases as we move away from the narrow slot, (https://www.edaboard.com/threads/254994/). I've seen this also in sonnet. From an isolation perspective, is it better to have two narrow grounds (>1.5*w) or one much wider ground between two RF traces?

2) If using grounded coplanar, is the straps over the strip necessary?

I'm working on a 500 - 3GHz switch. The current dream is 4350 w=30mil, s=10mil, h=20mil.

tnx,
Gene

 

[ge]

Suggestions would be appreciated.

tnx,
Gene

 

[volker_muehlhaus]

Gene, have you tried to simulate both cases and compare the results?

 

[ge]

That is my Friday project. I have access to Momentum but no 3d simulators.

 

[jossion]

I don't think you need a 3d simulator to do this. Momentum will help you. The ground layer will be infinitely spread. Place the coplanar waveguide structure on the substrate layer and put sufficient vias in the ground strip on cond layer and then do a simulation. Now by narrowing the width of the ground strip you can see what you wanted. But the moment you told you need a grounded coplanar waveguide you have to maintain the width of RF track and spacing between track and ground strip to get a 50 ohm RF line. In case of GCPW there can be multiple combinations of s and w which will give 50ohms.