Why an L-Network Has Only One Q for a Given Ratio
A lossless L-network has two reactances and must satisfy two equations, so its Q = sqrt(R_high/R_low − 1) is set by the resistance ratio alone. A 4:1 match always has Q = 1.73. Any other Q needs a third element or cascaded L-sections.
Engineer’s note: Once the impedance ratio is fixed, an L-network's Q is fixed too; with only two parts you have no freedom to set the bandwidth. For a narrowband match, the low part count makes it the best value around. For the wideband needs of modern systems, its limits are clear.
For practitioners
An L-network is the simplest lossless match there is: one series reactance and one shunt reactance. That simplicity has a cost. With only two elements, the network has no spare degree of freedom. Once the source and load resistances are fixed, the Q is fixed too. Bandwidth follows from Q, so it is fixed as well. This page shows why, and what you have to add when the fixed Q is not the one you want.
Two unknowns, two equations
Take a resistive load R_high that you need to present as a smaller resistance R_low. The shunt element always sits across the higher resistance. The series element sits on the low side.
The unknowns are the two reactances, X_s (series) and X_p (shunt). The requirement is that the impedance seen from the low side equals R_low exactly. That impedance is a complex number, so "equals R_low" is really two real equations:
- the real part must equal R_low
- the imaginary part must equal zero
Two real unknowns and two independent real equations give a unique solution, up to sign. Nothing is left over to tune. The Q isn't a design input. It's an output of the match.
Solving it: where Q comes from
Put X_p in parallel with R_high and convert that pair to its series equivalent. Define the shunt-arm Q as Q_p = R_high/X_p. The real part of the series equivalent is:
That is the real-part equation. It involves only X_p, so it pins the shunt reactance down on its own. Rearranging gives the result this whole page is about:
The imaginary part of the same series equivalent is:
The imaginary-part equation says the series element must cancel this leftover reactance. So X_s = Q·R_low, with the opposite sign to X_eq. The first equation set the shunt arm and the second set the series arm. There are no unknowns left.
The sign is the one real choice. A series inductor with a shunt capacitor gives the low-pass form. A series capacitor with a shunt inductor gives the high-pass form. Both satisfy the same two equations, so both have the same Q.
The series and shunt arms share one Q
The solution has a neat property. At the design frequency, the Q of the series arm equals the Q of the shunt arm:
This isn't a coincidence. It's the imaginary-part condition written another way. So an L-network has one Q, not two, and that Q depends on nothing but the ratio R_high/R_low.
Worked example: 50 Ω to 200 Ω at 1 GHz
R_high = 200 Ω and R_low = 50 Ω, so the ratio is 4.
- Q = √(4 − 1) = √3 ≈ 1.732.
- Series reactance: X_s = Q·R_low = 1.732 × 50 ≈ 86.6 Ω.
- Shunt reactance: X_p = R_high/Q = 200/1.732 ≈ 115.5 Ω.
- Check: X_s/R_low = 86.6/50 = 1.732, and R_high/X_p = 200/115.5 = 1.732. The two arms agree.
At 1 GHz, ω = 2π × 10⁹ ≈ 6.283 × 10⁹ rad/s.
- Low-pass: series L = 86.6/ω ≈ 13.8 nH, shunt C = 1/(ω × 115.5) ≈ 1.38 pF across the 200 Ω side.
- High-pass: series C = 1/(ω × 86.6) ≈ 1.84 pF, shunt L = 115.5/ω ≈ 18.4 nH across the 200 Ω side.
The component values differ and so do the out-of-band responses. The Q is 1.732 in both, because both circuits solve the same two equations.
Q, bandwidth and the ratio curve
The loaded Q of the match sets how fast it detunes away from the design frequency. Higher Q means narrower bandwidth, and fractional bandwidth falls roughly as 1/Q. Since an L-network's Q is a function of the ratio alone, so is its bandwidth.
| R_high/R_low | Q = √(ratio − 1) |
|---|---|
| 2 | 1 |
| 4 | 1.732 |
| 5 | 2 |
| 10 | 3 |
| 50 | 7 |
Q grows roughly as the square root of the ratio. Large transformation ratios therefore force narrow-band L-matches.
Every L-network sits on the curve. Points off the curve aren't reachable with two elements.
Getting a different Q means adding elements
To move off the curve, you need more unknowns than equations.
Higher Q (above the curve, narrower bandwidth). Add a third element to make a pi or T network. The extra reactance is the free parameter. In effect, the network transforms through a virtual resistance that is more extreme than either termination. That forces a larger ratio in one half, and the larger ratio sets the higher Q.
Lower Q (below the curve, wider bandwidth). Cascade L-sections through an intermediate resistance so that each section sees a smaller ratio. For 50 Ω to 200 Ω, an intermediate value of √(50 × 200) = 100 Ω splits the job into two ratio-2 steps with Q = 1 each, down from 1.732 for a single section.
In plain words · for students and beginners
L-network Q in plain words
A lever with its pivot already decided
Picture a seesaw bar with a heavy weight at one end and a light weight at the other. The heavy weight stands for the larger resistance and the light one for the smaller resistance. For the bar to balance, the pivot (the point the bar rests on) has to sit closer to the heavy end. Once you've picked the two weights, there's exactly one place the pivot can go. Put it anywhere else and the lever tips over.
With weights of 4 and 1, the pivot sits 1/5 of the way along from the heavy end. An L-network works the same way. Choose the two resistances and its Q (a number that says how sharply tuned the circuit is) is decided for you. For a ratio of 4, the Q is 1.73.
Where the picture breaks down: the pivot position isn't the Q. The pivot follows the ratio directly, while Q follows the square root of the ratio minus one. A lever also balances with one condition, but a matching network has to meet two, one for the resistance and one for the reactance (the part of impedance that stores energy instead of using it up). The analogy only shows that there's no free choice.
The same example, step by step
We want 200 Ω to look like 50 Ω.
- Divide: 200 / 50 = 4.
- Subtract one: 4 − 1 = 3.
- Take the square root: √3 ≈ 1.73. That's the Q.
- Series part: 1.73 × 50 ≈ 86.6 Ω of reactance.
- Shunt part (the element connected across the line to ground): 200 / 1.73 ≈ 115.5 Ω of reactance, placed across the 200 Ω side.
You can build this with an inductor in series and a capacitor across the line, or the other way round. The Q comes out at 1.73 either way.
Mistakes beginners make
- Trying to "tune" the bandwidth by changing one part. If you change one element, the match breaks. The values are locked together.
- Putting the shunt element on the wrong side. It always goes across the higher resistance.
- Thinking low-pass and high-pass versions differ in bandwidth around the design frequency. They have the same Q. What differs is how each one behaves far away from that frequency.
- Expecting a wide match across a big ratio. At a ratio of 50, Q is 7, which is narrow.
Remember
- Two parts and two conditions leave no free choice, so the ratio sets the Q.
- Q = √(R_high/R_low − 1), and the series arm and shunt arm share it.
- For a higher Q, add a third part (pi or T). For a lower Q, chain L-sections through a middle resistance.
Related pages on RF View
Frequently asked questions
Can I choose the Q of an L-network matching circuit?
No. Between two resistances, the two reactances are fully set by the match conditions, so Q = sqrt(R_high/R_low − 1) is fixed. The only remaining choice is low-pass or high-pass, and both give the same Q.
What is the Q of an L-network matching 50 ohms to 200 ohms?
The ratio is 4, so Q = sqrt(4 − 1) = 1.732. This holds for both the low-pass and the high-pass version.
How do I get a wider bandwidth than a single L-section allows?
Cascade two or more L-sections through an intermediate resistance. Each section then sees a smaller ratio and a lower Q. For 50 to 200 ohms, a 100 ohm midpoint gives two sections of ratio 2 and Q = 1 each.
How do I get a higher Q or narrower bandwidth than an L-network gives?
Add a third element and use a pi or T network. The extra reactance is the free parameter that lets you set Q, as long as it is above the L-network value for that ratio.