How Do You Design an RF PIN Diode Switch for High Isolation?

[jmckennon]

Hi,

This is my first encouter with anything to do with RF design. I'm trying to design an RF PIN diode switch to meet certain specifications. I'm having considerable difficulty putting together a circuit schematic that will accomplish this.

My specs are:
Center frequency: 15GHz
Bandwidth 3GHz
Insertion loss<0.5db
Isolation>25db
Return loss<15db

I have a seedling schematic that has values I can't modify, but I'm allowed to do really whatever I want to design the bias circuits and switching networks. I'm using ADS 2009 to do the schematic design. Attached is the seedling schematic I'm working with.

Since I've never really done any design work with RF circuits, any help would be greatly appreciated. My first task to tackle, is how to achieve isolation >25db. Does anyone know how I would do this? Thanks in advance!

 

[marcusl]

Whoa! The first thing you should ask is why they assigned such a demanding task to a newbie! These specs are achievable but are not simple--and nothing at Ku band is simple anyway. Your attachment is illegible, so we can't help you there. In general you'll need to compensate for the parasitic reactance of the device package, and may need to use a shunt-iterated topology. Do you understand S-parameters, return loss, etc? ADS? If you build something do you understand enough about microwave instrumentation to measure and characterize your design? If you are truly new to RF, then you have a lot to learn.

It's unreasonable to ask for an education in RF/microwave engineering from the folks here. Suggest you start with a book or two on RF/MW design. Add classes and tutorials offered by Besser, the Agilent / ADS folks, etc. Then read up on pin diodes (you can find app notes online, here's the first one that Google listed: http://www.qsl.net/n9zia/wireless/pdf/an922.pdf").

As specific questions come up, or if you get stuck, then put up a post here and we can help you.

 

[jmckennon]

I understand S parameters and return loss etc, but I've never built anything like this before. It doesn't have to be physically realizable it just needs to work in a simulation. I don't have too many things to worry about like a professional design would need, I just need it to meet certain design specs (this is an educational project). I'll try attaching the schematic I've put together so far. I guess the place where I am stuck right now is, how can I modify this so that I have an isolation >25 GHz?

 

[marcusl]

I understand S parameters and return loss etc, but I've never built anything like this before. It doesn't have to be physically realizable it just needs to work in a simulation.

Oh, that's a relief! Put your switching elements in between the splitter and the output ports. If you use a shunt as you've done, you'll need to include a lambda/4 line so the other side sees an open. Use an iterated shunt to get more isolation, as explained in pin diode app notes. Compensating for package parasitics at your operating frequency will help as well.