tapered line transformer with bends or curves

Has anyone done an tapered line impedance transformer that had curves or optimal mitered bends? I'm designing a Wilkinson divider and need to turn the straight/tapered 1/4 wave transformers lines into loops to reduce space. I split my tapered line impedance transformer into 3 pieces and put mitered bends between the pieces. I'm confused to to see that my design falls apart. Suggestions? I'm simulating using AWR/Microwave Office. I haven't seen a way to make tapered line curves.

tnx,
GE



I was expecting an optimal mitered bend to be short enough and close enough that I could ignore the bend. When running a tapered trace, am I wrong or is my simulation wrong?

well, the ONLY answer is to use a electromagnetic field solver for the full structure. When you put multiple discontinuities spaced at less than quarterwave, there are no valid simple models to predict the overall performance. This is because one discontinuity generates evanescent modes, which have not died down by the time they reach the 2nd discontinuity.

That was the answer I feared biff44. I've been running AWR's AXIEM EM. I'm a bit surprised that an optimally mitered stripline bend would cause a significant discontinuity. Possibly as you say, the discontinuities do not have a chance to die down.