Improve return loss results in wilkinson power splitter

[chailingleow]

Hi all,

Anyone how to improve return loss results in wilkinson power splitter?Thanks

 

[RF-OM]

Your question is not clear. Do you design your own splitter or already have one? What are design details? If you design it you need to use right transmission lines and connectors. For 50 Ohm 2-way splitter you will need two 70.7 Ohm lines and one 50 Ohm line, all about ¼ of wavelength long. Also it is very important to have right footprints for connectors. Next point is bandwidth. Wilkinson splitter is narrowband device. Depends on central frequency you can expect good return loss for bandwidth about 20%, but even for 40% S11 & S22 may still be acceptable. If there is not enough bandwidth you may build cascaded Wilkinson. Good starting book may be this one: in case you need it.
Do not forget isolation resistor. When you use printed transmission lines FR4 substrate usually good enough up to 4, sometimes even 5 GHz at least for 2-way splitter. I made 2-way Wilkinson on double layer FR-4 for 3 - 5 GHz band and got acceptable parameters. Building 4-way Wilkinson on double layer board is not easy but possible task.

Best regards,
RF-OM

 

[chailingleow]

Hi,

I want the insertion loss about 13dB and isolation 5.5dB.
I already have one circuit.But i need to improve on its parameter.If u have any good document on how to design wilkinson power splitter within its spec,pls email me.Thanks for helping..

 

[RF-OM]

S23 (isolation) of 5.5 dB is not a big problem. But again, you did not tell what is frequency and what is design type. S11 of 13 dB is more difficult to get . Check the book that I told you and you will find some good information. There are a lot of book about splitters, try Google search and you will get tons of information.

Best reards,
RF-OM

 

[chailingleow]

Hi,

Frequency from 10MHz to & GHz.what do u mean by design type?
May i know which book u recommend?because there is few there and it take time to download..its possible then it can send me email than its easier..

Thanks a lot

 

[RF-OM]

Dear chailingleow,
Well, my congratulations with so challenging task. Definitely no one off-the-shelf splitter will work in such a big range. Your goal is close to my. Currently I design splitter for 5 to 5000 MHz. It is considered to be impossible, but almost 50 years ago one of my coworkers and teachers taught me an excellent engineering rule: “your eyes may be scared but hands are doing”. Since that I am using this rule and it really works. In your particular situation if you restricted to Wilkinson combiner solution there is only one way to build it: to use multi stage Wilkinson. I can roughly estimate you need about 10 stages and such a combiner implemented on high quality double sided Rogers substrate may be about 25 inches long. The book what I recommended to you is Peter Vizmuller’s "Systems, Circuits, and Equations". There are many other books, but this one is a good start.
If you are not restricted to Wilkinson circuit check RF transformer based splitters. They will not provide results as good as multistage Wilkinson, but will be much smaller and less expensive.
Best regards,
RF-OM