Impedance matching

Impedance matching is the practice of inserting a passive network between a Transmission line and a mismatched load to eliminate reflections — making the input of the matching network look like $Z_0$ to the feedline.

The motivation: a mismatch ($Z_L\ne Z_0$) produces a Reflection coefficient $\Gamma \ne 0$, with fraction $|\Gamma {|}^2$ of the incident power bounced back to the source. This is energy wasted, and at high power levels, also potentially damaging to amplifiers and creating standing-wave hot spots on the line.

The trick: don’t change the load; instead insert a reactive network between the line and load that transforms the apparent impedance to $Z_0$ at the line’s input. Reactive components (inductors, capacitors, stubs) absorb no real power — they only redistribute the wave phases. All the source power ends up at the load.

What the matching network is

A matching network sits at some plane $M{M}^{\prime }$ between the line and the load, with the load’s actual impedance $Z_L$ on one side and the matching network on the other side facing the line.

After matching:

• Looking into $M{M}^{\prime }$ from the line side: impedance is $Z_0$ (matched, $\Gamma =0$).
• Multiple reflections can still occur between $M{M}^{\prime }$ and the load, but they’re confined to that segment — no reflection on the line feeding $M{M}^{\prime }$.

Three common topologies:

1. Lumped element (L-section, π, T networks): single $L$ and $C$ at the matching plane.
2. Stub matching: a short or open-circuited transmission line stub connected in parallel (or series) with the main line.
3. Quarter-wave transformer: a section of line with characteristic impedance $\sqrt{Z_0{Z}_L}$, length $\lambda /4$.

Quarter-wave transformer

For a real load $R_L$, insert a $\lambda /4$ section of transmission line with characteristic impedance

$Z_T=\sqrt{Z_0{R}_L}$

between the feedline ($Z_0$) and the load ($R_L$).

Why this works: from Wave impedance and input impedance, a $\lambda /4$ line terminated in $R_L$ presents input impedance $Z_T^2/R_L=Z_0{R}_L/R_L=Z_0$. The feedline sees $Z_0$, perfectly matched.

For a complex load $Z_L$, place the quarter-wave transformer at a position where the line impedance is purely real — a voltage maximum ($Z=S{Z}_0$) or voltage minimum ($Z=Z_0/S$), where $S$ is the SWR. The transformer matches that real impedance to $Z_0$.

Drawbacks: quarter-wave transformers work at one frequency (the design frequency); they detune at others. Limited bandwidth.

Single-stub matching

A short- or open-circuited transmission-line stub of length $l_s$ is connected in parallel with the main line at distance $d$ from the load.

The procedure (using a Smith chart):

1. Plot the normalized load admittance $y_L=1/z_L$ on the chart.
2. Move along the SWR circle (toward the generator) until you intersect the unit conductance circle $g=1$. Read off the distance $d$ in fractions of $\lambda$ — this is where the stub goes.
3. At that point, the normalized admittance is $1+jb$. The stub needs to provide $-jb$ susceptance to cancel.
4. From the susceptance scale, find the stub length $l_s$ for a short- (or open-) circuited stub of $Z_0$ that gives $-jb$.

After this surgery, the parallel combination at the stub plane is $g+jb+(-jb)=1$ — normalized matched.

Two solutions exist (the SWR circle intersects $g=1$ at two points); engineers pick the one with shorter $d$ or more bandwidth.

Lumped-element matching

For low frequencies or small structures, use lumped capacitors and inductors instead of stubs. The L-section is the simplest: one series element + one shunt element.

For example: load $Z_L=50+j100$ Ω, want to match to $Z_0=50$ Ω at 100 MHz.

Add a shunt inductor across the load to cancel the imaginary part of the load admittance, then a series capacitor. The values are computed from the admittance/impedance arithmetic; the Smith chart visualizes the path (the normalized susceptance $b$ below is read off the chart after moving along the constant-$|\Gamma |$ circle to the $g=1$ admittance circle).

A worked example from the Electromagnetics notes (at $\omega =2\pi \cdot 100\times 10^6$):

After determining the normalized susceptance $b=-1.58$ to cancel the load susceptance: $L=\frac{1}{\omega \cdot |b|/Z_0}=\frac{1}{2\pi \cdot 10^8\cdot 0.0316}=50\text{ nH}.$

Or a capacitor $C=50$ pF if the susceptance needs to be positive.

Why matching matters in practice

RF transmitters: a 100 W transmitter into a load with $|\Gamma |=0.5$ reflects 25 W back. The reflected wave can overheat the final amplifier, distort the output, or trip a protection circuit. Matching ensures the transmitter sees a clean 50 Ω load.

Receivers: maximum power transfer to the receiver requires the receiver’s input impedance to match the line. Mismatch wastes signal power and degrades signal-to-noise ratio.

Antennas: an antenna’s input impedance is frequency-dependent. A matching network at the antenna feedpoint flattens the SWR over the operating band.

High-speed digital: not “matching” in the RF sense, but controlled-impedance traces with termination resistors at source or load suppress ringing.

Tradeoffs

• Bandwidth: matching networks are tuned to a frequency. Wider bandwidth requires more complex networks (multi-stub, multi-section transformers). The Bode-Fano limit sets a fundamental tradeoff between bandwidth and match quality.
• Loss: ideal reactive elements are lossless, but real inductors and capacitors have parasitic resistance, which dissipates a small fraction of the matched power.
• Tolerance: component values drift with temperature; tuning may need adjustment.

In RF circuit design, the matching network is often the dominant subcircuit. Modern simulators (ADS, AWR) automate the Smith-chart procedure, but the principles are unchanged from the 1940s.