Visualize matching with a quarter-wave transmission line. Vary source line Z_0, load Z_L, substrate epsilon_r and design frequency to learn the intermediate impedance, physical length and bandwidth.
Parameters
Source line Z_0
Ω
Load Z_L
Ω
Substrate ε_r
—
Design frequency f_0
GHz
Bandwidth is computed for |Γ| < 0.1 (return loss > 20 dB). For simplicity, √ε_eff ≈ √ε_r is assumed.
Results
—
Intermediate Z_T = √(Z_0·Z_L)
—
Quarter-wave length l = λ_g/4
—
Effective wavelength λ_g
—
Bandwidth (relative, |Γ|<0.1)
Line Layout and Reflection Coefficient
Top: line Z_0 → intermediate Z_T (λ_g/4) → load Z_L / Bottom: |Γ| vs f/f_0 (red dashed line = |Γ|=0.1)
Theory & Key Formulas
Characteristic impedance of the quarter-wave line inserted between $Z_0$ and a real load $Z_L$:
$$Z_T = \sqrt{Z_0\,Z_L}$$
Effective wavelength on a substrate (relative permittivity $\varepsilon_r$) and physical length:
The relative bandwidth $\mathrm{BW}/f_0$ satisfying $|\Gamma|<\Gamma_\text{max}$ is obtained by inverting the cosine; it shrinks as $Z_L/Z_0$ grows.
What is the Quarter-Wave Impedance Matching Simulator
🙋
In RF circuits I keep hearing "match it with a quarter-wave line." What is actually happening? Why can't I just connect a 100 Ω antenna directly to a 50 Ω cable?
🎓
You can't, really. If you connect Z_L = 100 Ω directly to a Z_0 = 50 Ω line, the boundary produces a reflection coefficient Γ = (100−50)/(100+50) = 1/3, so about 11% of the power bounces back. The trick is to insert an extra section of line in between, a quarter wavelength long, with characteristic impedance $Z_T=\sqrt{Z_0 Z_L}=\sqrt{50\times100}\approx 70.7$ Ω. At the design frequency, the reflection vanishes completely. That's exactly the default in the simulator.
🙋
Why the geometric mean (square root of the product) and not just the average?
🎓
Good question. When the line is exactly a quarter wavelength long, its input impedance becomes $Z_\text{in}=Z_T^2/Z_L$ — that's the magic of a quarter-wave line. Setting this equal to Z_0 gives $Z_T^2=Z_0 Z_L$, so the geometric mean is the right answer. The arithmetic mean would not match.
🙋
If I change the substrate ε_r, the length l changes too. Higher permittivity means shorter line?
🎓
Right — the guided wavelength is $\lambda_g=c/(f_0\sqrt{\varepsilon_\text{eff}})$, so it shrinks inversely with the square root of the permittivity. On FR-4 (ε_r ≈ 4.3) it is roughly half the free-space wavelength. So a quarter wavelength at 2.4 GHz is about 31 mm in vacuum, but only about 15 mm on FR-4. Drop ε_r to 1 (air) in the simulator and you'll see l roughly double.
🙋
Looking at the lower graph, |Γ| is zero exactly at f_0, but it rises quickly as the frequency moves away.
🎓
That's the main weakness of quarter-wave matching. The "quarter wavelength" condition only holds at f_0, so the bandwidth is limited. The larger the ratio Z_L/Z_0, the narrower it gets. At the default 50→100 Ω the relative bandwidth is about 38%, but pushing to 50→500 Ω drops below 20%. For broader bandwidth you go to multi-section Chebyshev or maximally-flat transformers.
Frequently Asked Questions
In microstrip, the characteristic impedance depends on the conductor width W, substrate thickness h and relative permittivity ε_r. Once Z_T is decided, W is computed from Hammerstad's synthesis formula or an electromagnetic solver. For example, on FR-4 (1.6 mm thick, ε_r = 4.3), 50 Ω corresponds to W ≈ 3.0 mm and 70.7 Ω to W ≈ 1.6 mm. The layout narrows the line over the Z_T section and then returns to the original width.
The condition l = λ_g/4 is only exactly satisfied at the design frequency f_0; off-frequency, the input impedance drifts away from Z_0. The larger the impedance ratio Z_L/Z_0, the steeper the |Γ| curve around f_0, and the narrower the bandwidth. For broader bandwidth, switch to multi-section (Chebyshev / maximally flat) transformers, tapered lines, or L-section matching networks.
The voltage standing wave ratio VSWR is one-to-one with |Γ|: VSWR = (1+|Γ|)/(1−|Γ|). The condition |Γ| < 0.1 used here corresponds to VSWR < 1.22 (return loss > 20 dB). A typical relaxed criterion for antennas and amplifiers is |Γ| < 0.33 (VSWR < 2, RL > 9.5 dB); |Γ| < 0.1 is a tighter target.
Single-stub matching (adding a short- or open-circuited parallel stub) can match complex loads directly, but requires tuning two variables — stub position and length. A quarter-wave transformer only needs one extra line section for real loads and produces a simpler layout. For complex loads, the standard recipe is to cancel the reactive part first, then apply the quarter-wave transformer to the resulting real impedance.
Real-World Applications
Antenna feeding: A patch antenna often has an input impedance of 100-300 Ω depending on the design, and a quarter-wave transformer (sometimes combined with an inset feed) is widely used to match it to a 50 Ω coax. A single narrow line section on the PCB does the job, with no extra components required.
Amplifier output matching: Power amplifier output stages have low impedances of a few to a few tens of ohms and need to be matched to 50 Ω systems. Multi-section transformers (Chebyshev matching) cascading several quarter-wave lines, or tapered lines for very broadband matching, are common solutions.
Wilkinson power divider: The Wilkinson divider uses two quarter-wave lines of impedance √2·Z_0 ≈ 70.7 Ω and a termination resistor of 2·Z_0, simultaneously achieving port matching and isolation between output ports. It is one of the most elegant applications of quarter-wave matching.
Filters and couplers: Branch-line couplers, rat-race hybrids, combline and interdigital bandpass filters and many other passive microwave circuits are designed using combinations of quarter-wave sections as the basic building block. The quarter-wave concept is the lingua franca of microwave circuit design.
Common Misconceptions and Cautions
The most common mistake is to take Z_T as the arithmetic mean of Z_0 and Z_L. The correct answer is the geometric mean $Z_T=\sqrt{Z_0 Z_L}$. For example, matching 50 Ω to 200 Ω, the arithmetic mean is 125 Ω while the geometric mean is 100 Ω. Set Z_L to 200 in the simulator and check the Z_T card. A line designed with the arithmetic mean would leave a residual reflection even at the design frequency.
The next most common error is to compute the line length using the free-space wavelength. On a substrate you must use the guided wavelength $\lambda_g=c/(f_0\sqrt{\varepsilon_\text{eff}})$, which is much shorter than c/f_0. At 2.4 GHz on FR-4 (ε_r ≈ 4.3), a quarter of the free-space wavelength is 31.2 mm, but the line you actually build is 15.07 mm — roughly half. Toggling ε_r between 1 (air) and 4.3 (FR-4) in the simulator shows l changing by a factor of about two.
Finally, beware of the misconception that "if it matches at the design frequency, bandwidth doesn't matter". As this tool demonstrates, a quarter-wave transformer only achieves low reflection over a finite band centered on f_0. The Wi-Fi 2.4 GHz band (2.4-2.5 GHz, about 4% relative bandwidth) easily fits, but UWB or broadband antennas requiring 50% or more of relative bandwidth call for different approaches — multi-section transformers, tapered lines or matching filters. Keep an eye on the BW card and verify that your application sits comfortably inside the achievable bandwidth.