dearest mohammad

hope this picture could help

ali

What is the comics for, Mohd...?

Do you know what "8-way" means...?

Hi Mohammad, you may use unequal power division Wilkilson power divider as shown in page 40 of this lecture:

http://www.aps.anl.gov/asd/people/nassiri/USPAS2003/Lecture10.pdf

Before that you will need to define what should be the power division for each power divider in order to obtain the power distribution that you want: you start with the top ones by comparing the ratio of power going in each output. For the next bottom ones you add the output powers of the top power divider to get the new output powers and you compare the ratio of output powers to design the power dividers etc.

I hope this is clear and maybe there is other techniques to design your power divider such as using serial feed. Best regards

Simple Wilkinson divider is not suitable for 2-way (or n- way) power division with isolation between output ports because the output ports are not matched and the isolation resistor dissipates power, lowering efficiency. Two way isolated but unequal division can be realised with rat-race or branch-like couplers. 

Wilkinson's power divider is the device of choice, when one needs isolation between the output ports, and one wants the latter to be matched, as is reflected both by undergraduate-level courses in microwave engineering and by experience.

Perhaps for equal power division yes. Not for unequal power division - you need something more comp!ex to ensure matching without incurring losses.

I was addressing your general statement; unequal power division is another ball game indeed... Regardless, I would do not know how to ensure isolation between the output ports, in a the alternative structures that you suggested (?)

Here is an example of a branch line coupler with a 3/4 and 1/4 power split that is matched at all ports into 50 ohms. Upper an lower line impedance is 1/sqrt(1/50*(1/50+1/150)) ohms , and left and right line impedance are sqrt(50*150) ohms. The numbers 150 and 50 follow from the 3:1 power split.

You can do a similar design with an unsymmetrical rat-race if you need an in-phase split.

The traces show isolation S21, transmission S41 and S31, and reflection (traces labeled V(port_33)-1 and V(port_11)-1.

thanks denis and pieter for your help

i have to resoulve my power divider 

Here is a second example, this time with a ratrace in-phase splitter. (With the branch line coupler there is a 90° phase shift between the output signals.)

The figures show the design formulas and also a few responses of an ideal implementation for a splitter to 3/4 and 1/4 output power to ports 1 and 3.

If the input signal is applied to port 4, the output signals will be 180° out of phase.

(Port 2 is also known as the sum port and port 4 the difference port.)

There is one other configuration for an isolated combiner with unequal split that should also be mentioned: the Gysel (also known as Gysel-Wilkinson) combiner/divider.

This divider uses two 100 ohm isolation resistors compared to the rat-race's one 50 ohms resistor.The design equations are the same as for the rat-race. I swept the impedance of the balancing half-wavelength line. Like the rat-race and branch line couplers, it has the advantage that the isolation resistor(s) are terminations, as opposed to the Wilkinson's floating resistor.

I attach an example of this divider with an unequal split. Like the rat-race, the total length of transmission lines is 1.5 wavelength, compared to the branch-line coupler's 1 wavelength total length.

All the analyses were performed with LTSpice, an excellent implementation of Spice available free from Linear Technologies (www.linear.com).