0
\$\begingroup\$

I am working on an RF project and I notice that many sources I have used, recommend using a pi network (CLC), such as the image below, to impedance match the RF trace to the antenna.

enter image description here

I have a really short distance between my RF output and antenna and the impedance is controlled the entire way.

If the antenna impedance is well matched to my characteristic impedance, can I then simply ignore my pi network, open up the caps, and short the inductor ?

If the antenna impedance is does not match, and requires tuning, I've been shown this website (https://home.sandiego.edu/~ekim/e194rfs01/jwmatcher/matcher2.html)

When I enter I know, as seen below, there is one parameter I do not know how to determine - Desired Q. The default is set to 3. In this example, I set the source and load to be 50 ohms.

enter image description here

How do I select an appropriate Q ? What is the general workflow when working with RF that requires impedance matching or tuning ?

\$\endgroup\$
3
  • \$\begingroup\$ Read this paper from ARRL: Quality Factor, Bandwidth, and Harmonic Attenuation of Pi Networks \$\endgroup\$ Commented Nov 15, 2019 at 20:59
  • \$\begingroup\$ @Captainj2001 will do! Will check it out later tonight. \$\endgroup\$
    – efox29
    Commented Nov 15, 2019 at 21:00
  • \$\begingroup\$ how much variation will your RF carrier require? 0.54 to 1.64 MHz ( the entire AM band)? or 313Mhz +- 1MHz ? \$\endgroup\$ Commented Nov 16, 2019 at 3:28

2 Answers 2

1
\$\begingroup\$

If the antenna impedance is well matched to my characteristic impedance, can I then simply ignore my pi network, open up the caps, and short the inductor ?

Quite often not.

Reason: the \$\pi\$ network is also a decent "remover" of high order harmonics and can make the difference between a product that is able to pass certification tests and a product that will fail. The chip's output will very likely have unwanted harmonics and it will likely need to be filtered to stop cross-channel interference.

Regards estimating the \$\pi\$ filter values, you can also use this calculator. It gives you all the theory behind \$\pi\$ networks and shows how you can manipulate values to achieve a tighter bandwidth (aka cutting out those unwanted harmonics).

This is an example of a 10 MHz \$\pi\$ network that has 50 Ω input and 300 Ω output impedance and varies the transfer impedance from 10 Ω to 35 Ω: -

enter image description here

The top graph is probably most important in your situation: it shows that the high frequency attenuation can be as much as 7 dB at only 12 MHz. By the time you get to the 2nd harmonic (20 MHz) it can be around 20 dB to 25 dB. It's a 3rd order filter so 18 dB per octave (60 dB per decade). It will be approximately the same for a 50 Ω to 50 Ω converter.

The middle and lower graphs show the phase and magnitude of the input/output impedances.

\$\endgroup\$
2
  • \$\begingroup\$ Indeed, understanding that this is there for harmonic suppression is key. In practical terms, the asker's approach of trying to determine the Q first is probably mistaken. Rather, what they should do is model the harmonic attenuation. Then verify that the desired frequency range will be passed, though that's fairly likely - it would be hard with ordinary components to build this filter in such a way that would be too narrow for ordinary usage, the question is more will it have sufficient harmonic attenuation. \$\endgroup\$ Commented Oct 10, 2020 at 16:54
  • \$\begingroup\$ The antenna however might only be a good match over too narrow a range, and the network might help with that, but only when considered in the context of the actual antenna. \$\endgroup\$ Commented Oct 10, 2020 at 16:55
0
\$\begingroup\$

\$Bandwidth = \frac{ω0}{Q}\$ Where \$ω0\$ is your center frequency. So choose your \$Q\$ to set your bandwidth.

Probably more than you've wanted to know about RF matching.

can I then simply ignore my pi network, open up the caps, and short the inductor ?

Assuming that you are using small components, relative to the structures on your PCB you should be able to do that. Although a 0 ohm resistor probably isn't very good above 1Ghz. I'd look at some way of replicating the structure you have on your board, or do a board spin once you've verified that you don't need the matching network.

If it's an 0402 L, then you should be able to just solder bridge it.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.