Transfer Functions in Circuit Analysis

What a Transfer Function Means

A transfer function $H(s)$ is the single most compact description of how a linear circuit responds to an input. It is defined as the ratio of the Laplace transform of the output to the Laplace transform of the input, evaluated with all initial conditions set to zero:

The complex frequency $s = \sigma + j\omega$ carries two pieces of information at once: the imaginary part $\omega$ describes steady-state sinusoidal behavior, while the real part $\sigma$ describes growth or decay. Setting $s = j\omega$ recovers the familiar frequency response — magnitude and phase versus frequency — that you plot on a Bode diagram. Because a transfer function encodes the circuit's entire dynamic personality in one rational expression, it is the natural output of any symbolic solver and the starting point for filter and control design. Once you have $H(s)$ you can read off the steady-state gain at any frequency, predict the transient settling behavior, and judge stability — all without ever returning to the original differential equations. That economy is why engineers reach for the transfer function as the very first description of a new circuit.

Why Everything Becomes Algebra in the s-Domain

The reason transfer functions are so powerful is that the Laplace transform converts the differential equations of capacitors and inductors into ordinary algebra. Each storage element gets a frequency-dependent impedance, so the whole circuit can be treated as if it were purely resistive:

• Resistor: $Z_R = R$
• Capacitor: $Z_C = \dfrac{1}{sC}$
• Inductor: $Z_L = sL$

With these impedances in hand you can apply nodal analysis verbatim, simply replacing each conductance $1/R$ with an admittance $Y(s)$. The node-voltage equations become functions of $s$, and the symbolic solution of the system is the transfer function. This is exactly the path a tool like Lcapy takes internally.

Deriving H(s) From Nodal Analysis

Consider the canonical RC low-pass filter: a resistor $R$ from the input node $V_{in}$ to the output node $V_{out}$, and a capacitor $C$ from $V_{out}$ to ground. Writing KCL at the output node, the current entering through $R$ must equal the current leaving through $C$:

Grouping the output terms gives $V_{out}\left(\frac{1}{R} + sC\right) = \frac{V_{in}}{R}$, and dividing through yields the transfer function:

This single first-order expression is the complete description of the filter. Notice how the nodal equation produced it with one line of algebra — no differential equations were ever solved by hand. The same derivation generalizes immediately: add a second resistor and capacitor and the denominator becomes quadratic in $s$, producing two poles and the possibility of a sharper rolloff. The structure of the answer always reflects the number and arrangement of the energy-storage elements in the circuit.

Poles, Zeros, and What They Do

Every transfer function is a ratio of polynomials in $s$. The roots of the numerator are zeros (frequencies where the output vanishes); the roots of the denominator are poles (frequencies where the response would become infinite for a lossless system). Poles set the natural time constants and the bandwidth; zeros carve notches and boost slopes. The RC low-pass above has a single pole and no finite zeros:

The pole lies on the negative real axis, which guarantees a stable, decaying response. Its distance from the origin is the corner (cutoff) frequency in radians per second. The table summarizes how pole and zero locations shape behavior.

Worked Example: Evaluating the RC Filter

Let $R = 10\,\text{k}\Omega$ and $C = 10\,\text{nF}$. The time constant is $\tau = RC = 10^4 \times 10^{-8} = 10^{-4}\,\text{s}$, so the pole sits at $s_p = -1/\tau = -10^4\,\text{rad/s}$. The corner frequency in hertz is:

Now evaluate the response at three frequencies by substituting $s = j\omega = j 2\pi f$ into $H(s) = 1/(1 + j\omega RC)$.

• At DC ($f = 0$): $H = 1$, a gain of $0\,\text{dB}$ and zero phase — the signal passes untouched.
• At the corner ($f = f_c$, so $\omega RC = 1$): $H = \frac{1}{1 + j}$, giving $|H| = 1/\sqrt{2} \approx 0.707$ ($-3\,\text{dB}$) and a phase of $-45^{\circ}$.
• A decade above ($f = 10 f_c$, $\omega RC = 10$): $|H| = 1/\sqrt{101} \approx 0.0995$, which is $-20.04\,\text{dB}$, with phase $\approx -84.3^{\circ}$.

These three points confirm the textbook signature of a first-order low-pass: flat passband, a $-3\,\text{dB}$ point at the corner, and a clean $-20\,\text{dB}$ per decade slope thereafter, asymptotically approaching $-90^{\circ}$ of phase lag. The single real pole accounts for all of it. For a deeper treatment of this exact circuit, see the RC low-pass filter tutorial.

Reading a Bode Plot From the Factored Form

The real value of putting a transfer function into pole-zero form is that you can sketch its Bode plot without evaluating a single complex number. Rewrite the denominator so each pole appears as a factor like $(1 + s/\omega_p)$. Below a pole's corner frequency that factor is approximately one and contributes nothing; above the corner it grows like $s$, pulling the magnitude down at twenty decibels per decade and the phase toward $-90^{\circ}$. Zeros do the mirror image, lifting the magnitude and advancing the phase. Because logarithms turn products into sums, you simply add the straight-line contribution of each factor. A second-order system with two well-separated real poles, for instance, rolls off at twenty decibels per decade after the first corner and steepens to forty after the second. Memorizing these asymptotic slopes turns a page of algebra into a quick freehand sketch, which is exactly the instinct that distinguishes a fluent analyst from one who must compute every point.

From Transfer Function Back to Time Response

The pole also predicts the transient response. For a step input, the output of this RC filter rises as $V_{out}(t) = V_{in}\left(1 - e^{-t/\tau}\right)$, reaching about $63\%$ of its final value after one time constant $\tau = 0.1\,\text{ms}$. The same number, $1/\tau$, that set the cutoff frequency in the frequency domain governs the decay rate in the time domain — a reminder that poles unify both views.

Common Mistakes

1. Confusing rad/s with Hz. The pole magnitude $1/RC$ is in radians per second; divide by $2\pi$ before quoting a cutoff in hertz.
2. Ignoring loading. A transfer function derived with an open-circuit output changes the moment you attach a finite load; the load resistance must be included in the nodal equation.
3. Dropping initial conditions improperly. $H(s)$ assumes zero initial conditions. Nonzero starting states add a separate term that is not part of $H(s)$.
4. Misreading the slope. Each pole adds $-20\,\text{dB}$ per decade and each zero adds $+20\,\text{dB}$ per decade; counting them wrong gives the wrong asymptote.
5. Forgetting stability requires left-half-plane poles. A pole with positive real part means a growing, unstable response, no matter how benign the magnitude plot looks.

Related Tutorials

• RC Low-Pass Filter — the worked circuit behind this article, with Bode plots.
• Nodal Analysis Step by Step — the method used to derive $H(s)$.
• RLC Resonant Circuit — where complex-conjugate poles create resonance.