1
\$\begingroup\$

As someone new to practical RF design (the theory is easy by comparison!), I have a relatively simple question when it comes to matching networks, and it has to do with best practices.

Obviously, a theoretical match is easy to make with software. What I've generally done is measure the antenna, come up with a suitable matching network design. Then starting at the antenna, I add the components one by one measuring resulting impedance from each new component.

It seems to me, the biggest question is what to do when the results are so far off that the remaining component values are meaningless since they assume an intermediate impedance that just doesn't exist in the real circuit.

I guess there are two situations, one where the impedance from the new component is off but at least is along the same reactance/resistance curve. The other of course is where the new impedance is also rotated.

I try to adjust the theoretical matching network to values that produce the same impedance as measured. And sometimes I feel good about the adjustment when, for example, I can be fairly certain its just stray capacitance. Other times I'm not so certain.

What is the best way to proceed in these situations?

Based on Dakota's comment below, let me clarify.

When putting together a matching network, my approach is to put together a design first using something like JJSmith. Then, starting with the first component at the antenna:

  1. Add the matching component, then measure with a VNA.
  2. If the measured impedance doesn't match the design, adjust the component values until I get the planned impedance at that point in the matching network.
  3. Go back to the Smith chart and adjust the component value (and Q if necessary) in the charting software until the software agrees with the actual reading. Doing so allows me to adjust the component values yet to be added so that once again I'm headed toward the target impedance. It also gives me an idea of the parasitic capacitance and impedance.
  4. Repeat 1-3 until the complete matching network is in, and at the target impedance.
  5. Connect to a sweep generator and see if the antenna and matching network is generally working as I expected. If not, because I kept the Smith chart "in sync" as I built the actual network, I at least have a few clues of which component to tweak if, for example, the circuit isn't resonating at exactly the right frequency.

My question: does anyone have a method that is more efficient. I really appreciate any comments.

\$\endgroup\$
  • \$\begingroup\$ I think you'll get a better answer if you rewrite your question to address the specific design problem you're working on. \$\endgroup\$ – DakotaD Jul 20 '16 at 14:36
  • 1
    \$\begingroup\$ This is very dependent on what frequency range you're working in. I assume this is "low frequency" RF, i.e., f < 3 GHz, where you can still use some lumped components for matching. Most suppliers show in their datasheets what the self-resonant frequency of components is, and give models for the parasitic capacitances/inductances. Perform your matching based on these models, not by assuming ideal components and then measuring to see what parasitics there are. Above 3 GHz forget about lumped components altogether (for the RF matching) and use transmission line matching techniques. \$\endgroup\$ – Captainj2001 Jul 22 '16 at 20:57
  • \$\begingroup\$ @Captainj2001 Are you saying you don't usually check the new impedance after adding each component to see if you're on track? \$\endgroup\$ – Lance Beasley Jul 22 '16 at 21:35
  • \$\begingroup\$ @LanceBeasley That isn't a practice I've seen where I worked in the past. RF components are generally characterized quite carefully and include parasitic modelling values and the design process (with these good models) shouldn't need more than one iteration. \$\endgroup\$ – Captainj2001 Jul 22 '16 at 22:15
  • \$\begingroup\$ @Captainj2001 I see what you're saying. But I was referring more to the parasitics on the PC board, and I also that passives behave differently depending on the frequency. Doesn't an inductor have a slightly different inductance at different frequencies (and capacitors as well)? At 345MHz, I haven't been able to get anywhere without a lot of incremental testing, especially if the matching network includes additional stages for filtering. My experience so far has been that I have to test everything... \$\endgroup\$ – Lance Beasley Jul 22 '16 at 22:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.