high frequency power splitter

I am now designing a tri-way power splitter covering the frequency of 5.9-6.4GHz.
using the traditional wilkinson's methods.But the VSWR is very poor and the isolation between the 3 ouputs is also poor.
As the frequency goes high,the circuit will be an distributed one.so how to increase
the perfomances?

thanks
Liu Yun

Is it in simulation (ADS)?

firstly optimized in Serenade,and then fabricated.
the test result is poor.

Did you put the three balancing resistors to a common floating node? Do the three output lines couple together (they should not be)? Is the connection from all three outputs to the balancing resistors with zero length traces (or close to zero)?

The main problem for Wilkinson divider at high frequency is a capacitance of the balance resistor. If you want to have a good isolation, you must use a resistor with very little capacitance. But for high power device it's very difficult . May be better to use a hybrid(90 degree) model divider.