If not having GND vias in the structure the return current path for both nets is not closed. Adding them closes the return current path between the two reference planes close to the traces.
Thanks for the reply.
But similar via if i extract in line-sim return and insertion loss graph are coming accurate with no ground vias around them.
so i mean to say line sim and 3D results should be same.
The differential 3d via modelling in linesim until release 9.2 was a bit simplified for a quicker response of the simulator. When I remember correctly you can see this when you look at the mode conversion curves in the mixed mode model for the via (S13 and S14). I think they stuck at -200dB. This shows that some effects left out of the simulation which was seen as OK for very symmetric structures.
I already gave ports with lower reference.
Can you please elaborate on addition of return vias effects.
No, you need to convert to mixed mode to see the mode conversion. I think when you open the geo file from linesim in MGRID you will see that it uses differential ports which internally ends up in a 2 port differential model. once the simulation in the 3d solver is finished it will be internally combined with the result from a second solver (forgot which one) which finally ends in a single ended 4 port model. As I said before I would prefer using the solution in HL9.3 where we changed the 3d solver.
I'm not sure if I understood correctly. Are you looking for an introduction why stitching vias make sense?
Yes, With and without stitching difference in return and insertion loss profile.
Without stitching vias return loss graph was obtained as below:
While converting to differential return loss was obtained as:-
with stitching vias return loss graph was obtained as below:
I know that with ground vias proper return path is obtained when signal switch layers in order to maintain trace impedance.
Can you please explain the graph of return loss with and without stitching vias.
Thanks for the reply
I simulated two cases in differential via with back drill and without back drill also taking ground vias into effect, below is the geometry that i took into consideration ground vias have the connectivity in LAYER (2,4,6,9,11,15).
Differential vias traces entry layer is TOP and exit layer is 10, so extra stub is created from layer 11 to layer 15.
While simulating return and insertion loss profile in touchstone viewer, return loss and insertion loss is more in case of backdrill as compared to non backdrill, it should not happen.
With backdrill those ground vias are still from TOP to BOTTOM those are not backdrilled, my point of view is whether it is having larger return current loop from layer 11 to layer 15, I know that it will go to shortest return path.
whereas without backdrill differential vias and ground vias are adjacent to each other means return current loop will be less.
Do those ground vias are creating that change in insertion and return loss profile