7

I have designed a common-emitter RF amplifier, with input and output impedance matched for my load and source impedance. I have used Smith Chart for the matching.

The matching is good for the designed frequency, but when using other frequencies the matching is poor (I get large reflections in my spice simulation).

I know that wideband matching (one that works for all frequencies) is not possible.

But if I want to make a fair matching in a frequency band, say 200 MHz to 600 MHz, is there any good technique for this? (Except just trial and error with the Smith Chart).

stevenvh
  • 145,145
  • 21
  • 455
  • 667
JakobJ
  • 2,298
  • 1
  • 21
  • 42
  • 1
    aim for the center of the band, and don't expect to get good performance outside of half an octave either side. – JustJeff Apr 19 '11 at 21:49
  • I think it's possible to do better than just "aiming for the center", but I don't remember the details. You use filters to match at different frequencies. http://www.edaboard.com/thread145663.html http://www.freescale.com/files/rf_if/doc/app_note/AN721.pdf – endolith Apr 26 '11 at 15:04
  • @endolith - 200 to 600 just seems kind of wide. you could probably get a fair match over 200 to 400, and a similarly fair match over 300 to 600. – JustJeff Apr 26 '11 at 19:01
  • 1
    use a transformer? – markrages Apr 27 '11 at 00:04

1 Answers1

3

That's a relatively wide bandwidth to match a circuit over. The Bode-Fano equation governs bandwidth/return loss for reactive circuits, and google has some articles that will help with this. As long as the impedances you are working with don't vary much, you can do a decent job of matching. Resistive matches are pretty broadband, but lossy.

rfdave
  • 1,821
  • 10
  • 7
  • Thanks, I found some stuff on google, which explained it pretty good. When using a RF-transistor, they design for constant gain, but for a couple of frequencies inside the band. So matching is basically done for a couple of different frequencies. – JakobJ Apr 27 '11 at 11:34
  • Yeah, doing all that on a paper smith chart is pretty tough. Take a look at Microwaves101.com, they have some s-parameter calculations in Excel that might be helpful. Link here – rfdave Apr 28 '11 at 01:22
  • If you're feeling ambitious "Closed Form Solutions for the Design of optimum Matching networks" Dale Dawson IEEE TMTTS V57N1. The only downside is this seems to assume you are matching impedances, not a trace on a smith chart. Extracting impedances from smith chart traces is another one of those important skills that's been absorbed into design software. – rfdave May 03 '11 at 00:31