RF impedance matching pcb design

Thread Starter

yoamocuy

Joined Oct 7, 2009
84
Hey, I designed an impedance matching circuit as shown in the attachment. I was able to simulate the VSWR of the impedance matching circuit. I then added real traces in the simulator I am using and for simplicity sake just set all the trace widths and lengths to 1 mm each for now. I re-simulated the VSWR with real traces but my response is not very close to the simulation of the ideal circuit as can be shown in the attachment. I'm only operating between the frequencies of 150 MHz and 300 MHz, so unless I'm mistaken, I shouldn't need to worry about the trace widths as long as I keep my trace lengths shorter than about 55 mm since that is approximately 1/20th of a wavelength at 300 Mhz. I was able to build a filter using this same approach. Any ideas as to why this approach wouldn't work for an impedance matching stage at this frequency?

The specs of my board are FR4 board, 35um copper thickeness, .062 inch board thickness, and 2 layer board.
 

Attachments

Thread Starter

yoamocuy

Joined Oct 7, 2009
84
All the component values are the same as that of the ideal schematic. The only difference is that there are traces of 1mm length and width between each component. I've attached a close up of a section of the real schematic for clarity.
 

Attachments

Kermit2

Joined Feb 5, 2010
4,162
RF is funny about how things are laid out on the board. The angles between traces and the number of traces. Also the pattern of the traces affects RF much more than one would think.

RF layout is an art, and geometric(pretty) layouts are almost always not the best way to do it. Odd angles and the complete absence of parallel traces will most time perform better.

Perhaps some shielding can be added around portions of the circuit the way it is laid out now and improve the results
 

Thread Starter

yoamocuy

Joined Oct 7, 2009
84
It depends on the trace width, operating frequency, and board characteristics. To get 50 ohm traces I'd need 3.05 mm thick traces. Since my traces are so short though the trace impedance shouldn't make a difference. I've tried changing my trace thickness to 3.05 mm just to check, and the output isn't any different from that of the 1 mm thick traces. What is interesting though is that varying the trace length appears to change the output whereas when I was building the filter the trace length didn't matter as long as it was less then 1/20 of a wavelength. Is there any reason that the trace length would matter for an impedance matching circuit and not a filter?
 

billbehen

Joined May 10, 2006
39
I'm not much on RF design, but does the source or load impedance have more effect with a matching circuit than with a plain filter? As the frequency increases past 200MHz, maybe one or the other of these is no longer negligible....
 

SgtWookie

Joined Jul 17, 2007
22,230
A straight piece of 1mm wire that is 55mm long will have an inductance of about 51.2nH.
So roughly, for every 1mm of 1mm wide trace, you add 0.931nH inductance to the path. That's not counting coplanar parasitics.

I see you have a 60pF cap in there. What kind it is will make a big difference in the real world. A 47pF multilayer ceramic 0805 cap tested out at around 160pF @ 500MHz using an HP material analyzer.

And impedances on the input and output have to match what it is designed for.
 
Top