Part 6 · Chapter 23

Transmission Line Equations and Parameters

A short wire is just a node; a long one is a medium. The moment a cable's length grows comparable to the wavelength of the signal it carries, the voltage is no longer the same at both ends at the same instant — it travels, reflects, and interferes, exactly like the plane waves of Part 5. Transmission-line theory is the bridge between field theory and circuit theory: it keeps the convenience of voltage and current while smuggling in the wave physics through four parameters spread along the line. This chapter builds that model, derives the telegrapher's equations, and reads off the two numbers — the propagation constant and the characteristic impedance — that govern everything a line does.

Electromagnetic Field Theory Prof. Mithun Mondal Reading time ≈ 52 min
i What you'll learn
  • Why a line must be treated as distributed once its length approaches a wavelength.
  • The four primary line constants \(R, L, G, C\) per unit length, and their forms for coax, two-wire, and parallel-plate lines.
  • How a differential section yields the telegrapher's equations for \(V\) and \(I\).
  • The line wave equation, the propagation constant \(\gamma=\alpha+j\beta\), and the travelling-wave solution.
  • The characteristic impedance \(Z_0=\sqrt{Z/Y}\) tying voltage to current.
  • The lossless line (\(Z_0=\sqrt{L/C}\)) and the distortionless Heaviside condition \(R/L=G/C\).
Section 23-1

From Circuits to Lines

Ordinary circuit theory makes a silent assumption: that a change in voltage appears everywhere on a wire at once. That holds when the wire is electrically short — far shorter than a wavelength — because the signal crosses it in a time too brief to matter. But a signal travels at roughly the speed of light, so its wavelength shrinks as frequency rises: \(\lambda=u/f\). A one-metre cable is negligible at \(50\,\text{Hz}\) (\(\lambda=6000\,\text{km}\)) yet is several wavelengths long at \(1\,\text{GHz}\) (\(\lambda\approx0.3\,\text{m}\)).

Once the line is a meaningful fraction of a wavelength, the voltage genuinely differs from one end to the other at a given instant. The signal becomes a wave on the line. We could attack this with the full field equations of Part 5, but that throws away the comfortable language of \(V\) and \(I\). The transmission-line model keeps that language and recovers the wave behaviour by smearing the circuit elements continuously along the line.

🔑
When to switch models
treat a line as distributed once its length exceeds about a tenth of a wavelength

Below \(\ell\lesssim\lambda/10\) the lumped approximation is fine; above it, voltage and current vary along the line and the telegrapher's equations are needed.

Section 23-2

The Distributed-Parameter Model

Any two-conductor line is described by four primary constants, each defined per unit length: the series resistance \(R\) (\(\Omega/\text{m}\)) of the conductors, the series inductance \(L\) (\(\text{H/m}\)) from the magnetic field around them, the shunt conductance \(G\) (\(\text{S/m}\)) of the imperfect dielectric between them, and the shunt capacitance \(C\) (\(\text{F/m}\)) of that dielectric. Their values depend only on the cross-section geometry and the materials, not on the signal:

a b
Coaxial
d a
Two-wire
w d
Parallel-plate

The standard closed forms for the three workhorse geometries are collected below, with \(R_s=\sqrt{\pi f\mu_c/\sigma_c}\) the conductor's surface resistance (the skin-effect resistance of one square of metal):

Per-unit-lengthCoaxialTwo-wireParallel-plate
\(R\) (Ω/m)\(\dfrac{R_s}{2\pi}\!\left(\dfrac1a+\dfrac1b\right)\)\(\dfrac{R_s}{\pi a}\)\(\dfrac{2R_s}{w}\)
\(L\) (H/m)\(\dfrac{\mu}{2\pi}\ln\dfrac{b}{a}\)\(\dfrac{\mu}{\pi}\cosh^{-1}\!\dfrac{d}{2a}\)\(\dfrac{\mu d}{w}\)
\(G\) (S/m)\(\dfrac{2\pi\sigma}{\ln(b/a)}\)\(\dfrac{\pi\sigma}{\cosh^{-1}(d/2a)}\)\(\dfrac{\sigma w}{d}\)
\(C\) (F/m)\(\dfrac{2\pi\varepsilon}{\ln(b/a)}\)\(\dfrac{\pi\varepsilon}{\cosh^{-1}(d/2a)}\)\(\dfrac{\varepsilon w}{d}\)

Two relations cut across every row and are worth keeping in mind: the product \(LC=\mu\varepsilon\) and the ratio \(G/C=\sigma/\varepsilon\) hold for all these lines. They say the wave's speed depends only on the dielectric, and that the loss is fixed by the dielectric's conductivity — geometry cancels out.

Section 23-3

The Telegrapher's Equations

Take a slice of line of infinitesimal length \(\Delta z\). It carries series resistance \(R\,\Delta z\) and inductance \(L\,\Delta z\) along the conductors, and shunt conductance \(G\,\Delta z\) and capacitance \(C\,\Delta z\) across the gap. This is the elementary cell whose repetition is the line:

R Δz L Δz G Δz C Δz I(z) I(z+Δz) V(z) V(z+Δz) Δz
The per-unit-length equivalent circuit of a differential section Δz of line

Kirchhoff's voltage law around the top loop says the voltage drop across the section equals the series impedance times the current; Kirchhoff's current law at the right node says the current lost across the section equals the shunt admittance times the voltage. Dividing by \(\Delta z\) and letting it shrink to zero gives a coupled pair — the telegrapher's equations:

Telegrapher's equations (time-harmonic, phasor form)
\[ \frac{dV_s}{dz}=-(R+j\omega L)\,I_s, \qquad \frac{dI_s}{dz}=-(G+j\omega C)\,V_s \]

It is convenient to name the series impedance and shunt admittance per unit length, \(Z=R+j\omega L\) and \(Y=G+j\omega C\), so the pair reads \(dV_s/dz=-Z I_s\) and \(dI_s/dz=-Y V_s\). Voltage and current are now locked together, each feeding the other along the line.

Section 23-4

Wave Equation and Propagation Constant

Differentiate one telegrapher equation and substitute the other to decouple them. Each variable then obeys a one-dimensional wave equation identical in form to the field equations of Part 5:

The line wave equation
\[ \frac{d^2V_s}{dz^2}=\gamma^2 V_s, \qquad \frac{d^2I_s}{dz^2}=\gamma^2 I_s, \qquad \gamma=\sqrt{ZY}=\sqrt{(R+j\omega L)(G+j\omega C)} \]

The complex propagation constant splits into a real attenuation constant \(\alpha\) (nepers per metre, how fast the wave decays) and an imaginary phase constant \(\beta\) (radians per metre, how fast its phase advances):

Attenuation and phase constants
\[ \gamma=\alpha+j\beta, \qquad \alpha=\text{Re}\,\gamma\ (\text{Np/m}), \qquad \beta=\text{Im}\,\gamma\ (\text{rad/m}) \]

The general solution is the sum of two travelling waves — one moving in \(+z\), one in \(-z\) — just as on a plane-wave problem at a boundary:

Forward and backward voltage waves
\[ V_s(z)=V_0^{+}e^{-\gamma z}+V_0^{-}e^{+\gamma z} \]
~ source Z_L V₀⁺ e^(−γz) → ← V₀⁻ e^(+γz) z →
The line solution is a forward wave plus a backward wave; the load decides how much energy returns
Section 23-5

Characteristic Impedance

Substituting the voltage solution back into a telegrapher equation gives the current, and the ratio of voltage to current within a single travelling wave turns out to be a constant of the line — its characteristic impedance:

Characteristic impedance
\[ Z_0=\sqrt{\frac{Z}{Y}}=\sqrt{\frac{R+j\omega L}{G+j\omega C}}=R_0+jX_0 \]

With it, the current wave mirrors the voltage wave, the backward term carrying a minus sign because it propagates the other way:

The companion current wave
\[ I_s(z)=\frac{V_0^{+}}{Z_0}e^{-\gamma z}-\frac{V_0^{-}}{Z_0}e^{+\gamma z} \]

The meaning of \(Z_0\) is physical, not just algebraic: it is the impedance a forward wave "sees" as it advances, and the ratio of voltage to current it sets up. A line terminated in a load equal to \(Z_0\) launches only a forward wave — there is nothing to reflect — which is why \(Z_0\) is the target value for matching, the subject of the next two chapters. The propagation constant \(\gamma\) and the characteristic impedance \(Z_0\) together are the line's secondary constants, derived from the four primary ones, and between them they describe everything the line does to a signal.

Section 23-6

The Lossless Line

At high frequencies a good line has negligible loss: \(R\ll\omega L\) and \(G\ll\omega C\), so set \(R=G=0\). The secondary constants then become beautifully simple — the attenuation vanishes and the characteristic impedance is purely real:

Lossless line
\[ \alpha=0,\quad \beta=\omega\sqrt{LC},\quad u=\frac{\omega}{\beta}=\frac{1}{\sqrt{LC}},\quad Z_0=\sqrt{\frac{L}{C}}\;(\text{real}),\quad \lambda=\frac{2\pi}{\beta} \]

Because \(LC=\mu\varepsilon\), the phase velocity \(u=1/\sqrt{LC}=1/\sqrt{\mu\varepsilon}\) is exactly the speed of a plane wave in the same dielectric — the line carries the wave at the medium's own velocity, slowed from \(c\) only by the relative permittivity. For an air-filled coaxial line, combining \(L\) and \(C\) from the table gives the often-quoted design formula \(Z_0=\frac{1}{2\pi}\sqrt{\mu/\varepsilon}\,\ln(b/a)\approx 60\ln(b/a)\,\Omega\). Almost all practical RF and microwave work assumes the lossless line; the small real loss is added back afterward as a perturbation.

Section 23-7

Distortionless Lines

A line distorts a signal when its different frequency components travel at different speeds or fade by different amounts, so the pulse shape spreads as it goes. Heaviside found the condition that prevents this on a lossy line: arrange the parameters so that the series and shunt branches lose in the same proportion,

Heaviside distortionless condition
\[ \frac{R}{L}=\frac{G}{C} \]

Under this single condition the propagation constant and characteristic impedance collapse to the clean forms below — the attenuation \(\alpha\) is a constant independent of frequency, and the velocity \(u=1/\sqrt{LC}\) is the same for every frequency, so every component arrives together and the waveform is preserved (merely attenuated):

Distortionless line
\[ \alpha=\sqrt{RG}=R\sqrt{\frac{C}{L}},\quad \beta=\omega\sqrt{LC},\quad u=\frac{1}{\sqrt{LC}},\quad Z_0=\sqrt{\frac{L}{C}}=\sqrt{\frac{R}{G}}\;(\text{real}) \]

Real cables have too little series inductance to satisfy the condition naturally, so early telephone engineers added it deliberately — "loading" lines with series coils at intervals to raise \(L\) until \(R/L\) dropped to meet \(G/C\). It was the practical triumph that made long-distance telephony intelligible. Note that a lossless line is just the special distortionless case with \(R=G=0\).

Line typeConditionαβZ₀
General (lossy)\(\text{Re}\sqrt{ZY}\)\(\text{Im}\sqrt{ZY}\)\(\sqrt{Z/Y}\) (complex)
Distortionless\(R/L=G/C\)\(R\sqrt{C/L}\)\(\omega\sqrt{LC}\)\(\sqrt{L/C}\) (real)
Lossless\(R=G=0\)\(0\)\(\omega\sqrt{LC}\)\(\sqrt{L/C}\) (real)
Section 23-8

Worked Examples

1 Lossless line constants

Problem. A lossless line has \(L=0.25\,\mu\text{H/m}\) and \(C=100\,\text{pF/m}\). Find \(Z_0\), the phase velocity \(u\), and at \(100\,\text{MHz}\) the phase constant \(\beta\) and wavelength \(\lambda\).

Solution. Use \(Z_0=\sqrt{L/C}\), \(u=1/\sqrt{LC}\), \(\beta=\omega/u\), \(\lambda=u/f\):

Working
\[ Z_0=\sqrt{\tfrac{0.25\mu}{100p}}=50\,\Omega,\quad u=\tfrac{1}{\sqrt{LC}}=2\times10^{8}\,\text{m/s},\quad \beta=\tfrac{\omega}{u}\approx 3.14\,\text{rad/m},\quad \lambda=2\,\text{m} \]
2 Propagation constant of a lossy line

Problem. At the operating frequency a line has \(Z=R+j\omega L=3+j4\ \Omega/\text{m}\) and \(Y=G+j\omega C=(0.3+j0.4)\,\text{mS/m}\). Find \(\gamma\) and \(Z_0\).

Solution. Write each in polar form: \(Z=5\angle53.13^\circ\), \(Y=0.5\angle53.13^\circ\,\text{mS}\). Then \(\gamma=\sqrt{ZY}\) and \(Z_0=\sqrt{Z/Y}\):

Working
\[ \gamma=\sqrt{2.5\times10^{-3}\angle106.26^\circ}=0.05\angle53.13^\circ=0.03+j0.04 \]

So \(\alpha=0.03\ \text{Np/m}\), \(\beta=0.04\ \text{rad/m}\), and \(Z_0=\sqrt{10^4\angle0^\circ}=100\,\Omega\) (real here, since \(Z\) and \(Y\) share a phase angle).

3 Designing a 50 Ω air coax

Problem. Find the ratio \(b/a\) for an air-filled coaxial line of characteristic impedance \(50\,\Omega\).

Solution. For air, \(Z_0=60\ln(b/a)\), so invert:

Working
\[ \ln\frac{b}{a}=\frac{Z_0}{60}=\frac{50}{60}=0.833 \;\Rightarrow\; \frac{b}{a}=e^{0.833}\approx 2.30 \]
4 Making a line distortionless

Problem. A line has \(R=0.1\ \Omega/\text{m}\), \(L=0.4\,\mu\text{H/m}\), \(C=160\,\text{pF/m}\), \(G=0\). What \(G\) makes it distortionless, and what are \(\alpha\) and \(Z_0\) then?

Solution. The condition \(R/L=G/C\) fixes \(G\); then \(Z_0=\sqrt{L/C}\) and \(\alpha=R/Z_0\):

Working
\[ G=\frac{RC}{L}=\frac{0.1(160p)}{0.4\mu}=40\,\mu\text{S/m},\quad Z_0=\sqrt{\tfrac{L}{C}}=50\,\Omega,\quad \alpha=\frac{R}{Z_0}=2\times10^{-3}\,\text{Np/m} \]
5 Low-loss attenuation in dB

Problem. A nearly lossless \(50\,\Omega\) line has \(R=0.5\ \Omega/\text{m}\) and negligible \(G\). Estimate \(\alpha\) in Np/m and dB/m.

Solution. For a low-loss line \(\alpha\approx R/(2Z_0)\); convert with \(1\ \text{Np}=8.686\ \text{dB}\):

Working
\[ \alpha\approx\frac{R}{2Z_0}=\frac{0.5}{100}=5\times10^{-3}\,\text{Np/m}=0.0434\,\text{dB/m} \]
6 Velocity factor of a filled line

Problem. A coaxial line is filled with PTFE (\(\varepsilon_r=2.1\)). Find the phase velocity and the wavelength on the line at \(1\,\text{GHz}\).

Solution. Since \(u=c/\sqrt{\varepsilon_r}\) (non-magnetic dielectric):

Working
\[ u=\frac{3\times10^8}{\sqrt{2.1}}\approx 2.07\times10^{8}\,\text{m/s}\ (0.69c),\quad \lambda=\frac{u}{f}\approx 0.207\,\text{m} \]
Review

Chapter Summary

Distributed model

When \(\ell\gtrsim\lambda/10\), use \(R,L,G,C\) per unit length, not lumped elements.

Telegrapher's equations

\(dV/dz=-(R+j\omega L)I\), \(dI/dz=-(G+j\omega C)V\).

Propagation constant

\(\gamma=\sqrt{ZY}=\alpha+j\beta\); \(\alpha\) attenuates, \(\beta\) phases.

Characteristic impedance

\(Z_0=\sqrt{Z/Y}\); match the load to \(Z_0\) for no reflection.

Lossless line

\(\alpha=0\), \(\beta=\omega\sqrt{LC}\), \(u=1/\sqrt{LC}\), \(Z_0=\sqrt{L/C}\).

Distortionless

\(R/L=G/C\Rightarrow\alpha=\sqrt{RG}\), real \(Z_0\), shape preserved.

Practice

Problems

For each item, decide first whether the line is lossless, distortionless, or fully lossy — that sets which formula for \(\gamma\) and \(Z_0\) applies. Difficulty rises down the list.

  1. A lossless line has \(L=0.5\,\mu\text{H/m}\) and \(C=200\,\text{pF/m}\). Find \(Z_0\) and \(u\).
  2. At what line length (in wavelengths) does the lumped approximation start to fail, by the \(\lambda/10\) rule?
  3. An air coax has \(b/a=3.5\). Find its characteristic impedance.
  4. What \(b/a\) gives a \(75\,\Omega\) air-filled coaxial line?
  5. A lossless line operates at \(300\,\text{MHz}\) with \(u=2\times10^8\,\text{m/s}\). Find \(\beta\) and \(\lambda\).
  6. Show that for every line in the parameter table, \(LC=\mu\varepsilon\).
  7. A line has \(Z=4+j3\ \Omega/\text{m}\) and \(Y=(0.4+j0.3)\,\text{mS/m}\). Find \(\gamma\) and \(Z_0\).
  8. A line has \(R=0.2\ \Omega/\text{m}\), \(L=0.5\,\mu\text{H/m}\), \(C=50\,\text{pF/m}\). Find the \(G\) for distortionless operation and the resulting \(\alpha\).
  9. For a \(50\,\Omega\) low-loss line with \(R=1\ \Omega/\text{m}\) and \(G=0\), find \(\alpha\) in dB/m.
  10. A polyethylene-filled line has \(\varepsilon_r=2.25\). Find its velocity factor and the wavelength at \(2\,\text{GHz}\).
  11. Explain physically why \(Z_0\) is real for a lossless line but generally complex for a lossy one.
  12. Explain why adding series "loading" inductance reduced distortion on early telephone lines, in terms of the Heaviside condition.
Tip: two derived numbers run all of transmission-line theory — the propagation constant \(\gamma=\sqrt{ZY}=\alpha+j\beta\) and the characteristic impedance \(Z_0=\sqrt{Z/Y}\). Get them from the four primary constants and everything else follows. For most RF work the line is treated as lossless, where \(Z_0=\sqrt{L/C}\) is a real number, \(\beta=\omega\sqrt{LC}\), and the wave glides at the dielectric's own speed \(1/\sqrt{LC}\). Loss, when it matters, enters through \(\alpha\); and the Heaviside condition \(R/L=G/C\) is the one arrangement that lets a lossy line attenuate without distorting. Hold on to \(Z_0\) especially, because the next chapter is entirely about what happens when the load does not equal it — Chapter 24 takes up wave propagation, reflection, and standing waves on transmission lines.