Filters

Filters pass wanted frequencies and reject others. Learn low-pass, high-pass, band-pass and notch responses, passive RC/LC and active op-amp filters, filter order and roll-off, Butterworth, Chebyshev and Bessel approximations, Sallen–Key design, and anti-aliasing filters for ADCs.

Filters are everywhere: removing 50 Hz hum from ECG signals, anti-aliasing before an ADC, separating bass and treble in speakers, selecting a radio channel. Active filters use op-amps to achieve sharp responses without inductors.

An analogy

A filter is like a sieve for frequencies: the mesh size (cut-off) decides which frequencies pass and which are caught.

Filter types

Low-pass (keep lows), high-pass (keep highs), band-pass (keep a band), band-stop/notch (remove a band, e.g. 50/60 Hz hum). Passive RC filters are simple but load-sensitive; active filters add gain and buffering.

Order and approximations

Higher order = steeper roll-off. Butterworth: maximally flat passband. Chebyshev: steeper but with passband ripple. Bessel: best phase (pulse shape), gentle roll-off. Elliptic: steepest with ripple in both bands.

Sallen–Key and MFB

Second-order sections using one op-amp; cascade sections for higher orders (each with a specific Q from tables). Multiple-feedback (MFB) topologies suit band-pass and inverting designs.

Anti-aliasing and digital alternatives

Before an ADC sampling at f_s, attenuate everything above f_s/2 below 1 LSB. Oversampling ADCs relax the analog filter. Digital FIR/IIR filters (DSP module) complement analog ones.

Formulas

Worked example: Sallen–Key Butterworth

Given: f_c = 1 kHz, R = 10 kΩ, unity gain, Q = 0.707

Find: C₁ and C₂.

Answer: C₁ ≈ 22 nF, C₂ ≈ 11 nF (use 22 nF and 10 nF + 1 nF)

Worked example: Attenuation at 2 f_c

Given: Butterworth n = 2 and n = 4

Find: Gain at 2 kHz.

Answer: −12.3 dB vs −24.1 dB

Common mistakes

Applications

Key takeaways

Going further