optimization of microstrip chamfered bends
I m using IE3D to optimize chamfer bends at a particular frequency. In KC gupta's book on microstrips some relations have been given for optimum chamfering of microstrip bends. These relations are dependant only on width and height, but not on length of arms. But i noticed that by changing the length in IE3D, the graph of s11 vs frequency changes and also the value of optimum chamfering changes. Shouldn't it be independent of length according to the formulae? , or i m doing something wrong?
ideally Gupta's equations are valid but assuming transmission line length having no effect i.e. they are 50 ohm and no interaction between after the bend transmission line.
Inn IE3D it accounts the coupling between the Tx lines too so that it will give less IL. SO IE3D is good on integrated circuits.
Thanks for the quick reply, I have to find some empirical relation for optimum chamfering of microstrip 90 degree bends at 2.45GHz for GaAs substrate and gold metal strip. Can this be done accurately using IE3D? what variables should i consider as independent variables? width, height ? As I already mentioned as I change length of arms of bend in IE3D, value of optimum chamfering changes. But in most cases length is not known in advance. How to get around this problem? Should I just keep the length as constant while varying width and height? In that case what value should i take for length?
Precisely asking, can we categorically say that given the value of width and height, a particular value of chamfering is best at a given frequency?
Hi,
Its possible to calculate it good...
K.
Yes, it should be independent. Did you shift the port reference planes to the bend?
By shifting port reference planes to the bend does it mean as shown in the first figure?
Why should the above two models yield different S parameters?
1. If the feedline is not exactly 50 Ohm, additional length will change the S-parameters.
2. In the model where the ports are located directly at the bend, your EM results might be bad because of fringing fields between ports and the bend (discontinuity).
1. If the feedline is not exactly 50 Ohm, additional length will change the S-parameters.
2. In the model where the ports are located directly at the bend, your EM results might be bad because of fringing fields between ports and the bend (discontinuity).