Taking into account electric fringes effect on transmission line length

[dirac16]

I have been working on a recent paper. Basically I want to re-do the simulation results provided in the article. The circuit is a type of Wilkinson power divider with wide passband and stopband, and a also wide isolated frequency band. The circuit with its parameters is shown below.

Basically I want to re-draw the layout as shown above and do a post layout simulation. The substrate is Rogers RO4003C with the dielectric constant 3.55, loss tangent 0.0027, and layer thickness 0.813mm. First for the microstrip coupler lines I have to find the associated width (W), length (L), and separation (S) using Linecalc tool. I did this for the middle coupler and found the following physical lengths (@f0=2.4GHz):

W4=4.19mm (as compared with W4=3.96mm)
L4=4.52mm (L4=4.07mm)
S4=0.71mm (g2=0.44mm)

As you see my calculated lengths are slightly bigger than those illustrated for the middle coupler in the bottom right table. This is clear because Linecalc did not take into account the electric fringes which causes the transmission line to have a slightly shorter effective length. Now how can I find the physical lengths by considering the electric fringes effect? Is there a clear way to find the effective lengths?

Thanks

 

[vfone]

Tuning the physical length of the line during an EM simulation would give the best results..

 

[BigBoss]

Line Calculator does not take the fringing effects into account.It calculates the length and others as a standalone and very long transmission line.
Optimization is not easy especially for such complex EM structures and it does frequently not give a reliable result.Manual Iteration is better approximation.Overall Scaling( minor steps) may be a solution.

 

[dirac16]

Tuning the physical length of the line during an EM simulation would give the best results..

So how can I do the tuning during EM simulation? Is there any relevant tutorial on how to do this?

 

[volker@muehlhaus]

EM-based tuning depends on your simulator.

One nice method often suggested by Sonnet Software: if you break a line and insert two ports left and right, you can take your EM results and insert some transmission line length in schematic simulation. And that length can be tuned/optimized at circuit simulation speed.
This method requires good port calibration, then it is very efficient.

The same "tuning port" idea also works for shunt C or series L, using single series or shunt port. For inserting a "tuning transmission line" we need two ports.

Good luck!
Volker

https://www.sonnetsoftware.com/support/downloads/techdocs/port-tuning-a-microstrip-folded-hairpin-filter-revc.pdf