Short notes

Bilinear Transform Method Short Notes

When a Butterworth analog low-pass prototype with cutoff at 1 rad/s is transformed into a digital IIR filter with cutoff at 1 kHz using a 8 kHz sampling rate, the bilinear transform maps the entire imaginary axis of the s-plane onto the unit circle of the z-plane — but it compresses the frequency axis logarithmically, causing the digital frequency to be a warped version of the analog frequency. This frequency warping is not a bug; it is managed by prewarping the critical frequency before design, and the technique completely eliminates the aliasing problem that plagues the impulse invariance method.

ECE, EI

How it works

Bilinear transform substitution: s = (2/T)·(1−z⁻¹)/(1+z⁻¹), where T is the sampling period. This maps the left-half s-plane entirely into the interior of the unit circle in the z-plane, guaranteeing stability. The frequency relationship: Ω = (2/T)·tan(ω/2), where Ω is the analog frequency and ω is the digital frequency (0 to π). Prewarping: design the analog prototype at Ωpre = (2/T)·tan(ωc/2) instead of the desired digital cutoff frequency ωc, so that after transformation the digital filter has its −3 dB point exactly at ωc. For a 1st-order Butterworth H(s) = 1/(s+1): substituting s → (2/T)·(1−z⁻¹)/(1+z⁻¹) yields H(z) directly by algebraic simplification.

Key points to remember

Bilinear transform preserves filter order — an Nth-order analog prototype maps to an Nth-order digital filter. No aliasing occurs because the entire jΩ axis maps to the unit circle, unlike impulse invariance where only the strip −π/T to π/T is mapped. The frequency compression is most severe near ω = π (digital Nyquist), where a very large range of analog frequencies all map to π — this is why stopband edges warp more than passband edges for high-order filters. Prewarping is absolutely essential for band-pass and high-pass digital filters designed via bilinear transform; failure to prewarp shifts the actual cutoff from the specified value. The transformation H(z) = H_a(s)|_(s = (2/T)(1−z⁻¹)/(1+z⁻¹)) is applied by substituting and then rationalising to find the z-domain polynomial coefficients.

Exam tip

Every Anna University DSP paper asks you to design a first-order digital Butterworth LPF at a given cutoff frequency using the bilinear transform — show the prewarping step, substitute into H(s), simplify to H(z), and state that no aliasing occurs because the entire jΩ axis maps to the unit circle.

More Digital Signal Processing notes