High power L matching

[e.sotillet]

Hi guys, I usually work with small signals and today I need to use L matching network to connect a 50 ohm, 40dBm source to a 75ohm load.

But my question is: how I can select the proper capactors for this task? I will use air coil with relative thick wire but I dont know how to select the right capacitor. I think one major key parameter is the Q, rigth? I was thinking in High-Q RF capactors like those from ATC.

So, my questions are:

1.- What determine the selection of the capacitor?

2.- When can I use a 0603 capacitor and when I need to go with a bigger one?

Best regards

[TimFox]

Capacitors have a maximum current as well as a maximum voltage.
To simplify, high Q gives a low ESR, with lower power dissipation for a given current.
You need to know either the manufacturer's current rating, or an estimate for power rating of the capacitor package, and calculate the RF current through the capacitor in your circuit.
In a resonant circuit used for power coupling, the current through the reactances may be larger than that delivered to the load ("circulating tank current").
+40 dBm = 10 W, so this is an important quantitative problem.

[M0HZH]

You probably also need to design for output mismatches.

I use double the normal current as capacitor rating. For +40dBm you would have about 365mA of current into 75ohm, so a capacitor rated for 800mA RF current or more should be all right.

At these power levels, I see 0805 being the most common capacitor size.

[TimFox]

Actually, the current through the capacitor will be higher than the current delivered to the load for a resonant Pi network, where the network Q value can be chosen to be 10 or so.
The L network does not have that degree of freedom.
For 50 to 75 ohms, with a shunt capacitor on the 75 ohm side, the reactance will be roughly -j 106 ohms, so the current through the capacitor is slightly less than the current delivered to the 75 ohm load (same voltage).

[Wallace Gasiewicz]

I would just use an approximate Toroid Transformer.
4 turns primary and 5 turns secondary on the transformer

[e.sotillet]

Dear Tim, so you propose to use a pi network instead of a L network?


On the other hand, the toroid transformer seems a good solution but I have to make several of this boards, so I think it will better something surface mounted.

Best regards

[TimFox]

Dear Tim, so you propose to use a pi network instead of a L network?


On the other hand, the toroid transformer seems a good solution but I have to make several of this boards, so I think it will better something surface mounted.

Best regards

A pi network is more flexible, and allows you to choose the Q of the network.  It has the advantage of reducing harmonics in a narrow-band application, but with higher network Q it will involve higher circulating current in the reactances.
If this is a broad bandwidth application, the toroid is a better choice and can be done as a transmission line transformer for very wide bandwidth.

[TimFox]

Since you need surface-mount, I took a quick look at the MCL catalog for transformers, since they make both surface-mount and connectorized units.
https://www.minicircuits.com/WebStore/Transformers.html
They have many rated for 50:75 ohms, over different frequency ranges.
Their index listing includes frequency and ratio, but does not have power ratings;  however, each datasheet has an RF power rating.
The small SMT packages seem to go up to only 250 mW (+24 dBm), and even ones with N connectors go up to maybe 1.5 W (+32 dBm).
Feel free to check more thoroughly, but if you use a toroid you will need to choose one large enough to handle your 10 W requirement.

[profdc9]

Generally a L network results in the lowest loss and the highest bandwidth because it is the lowest Q matching network, however, it requires the largest component values.  The component values can have more loss because of the materials needed to achieve these high values in a small volume.

Depending on your wavelength, you might be able to use a quarter wave transformer.  Because of the small difference between impedances the usable bandwidth will probably be quite high.

A 1/12 wave transformer is even more compact

https://www.cv.nrao.edu/~demerson/cs/twelfth.htm
https://www.microwaves101.com/encyclopedias/twelfth-wave-transformer

There are actually multiple solutions for an L match.  I wrote a MATLAB program (you can use with Octave) to calculate all of the solutions to figure out which might be the easiest to attain.  You could use equivalent stubs for impedances (open stub for capacitance, shorted stub for inductance, keep in mind that vias have inductance too).