Passive: just R, L and C doing the dividing
A passive filter is built from nothing but resistors, capacitors and inductors — no power supply, no amplifier, no chip. It is really just the voltage divider you already know, except one of the legs has an impedance that changes with frequency. In an RC low-pass, the resistor R is the top leg and the capacitor C is the bottom leg to ground; the output is taken across C. At low frequencies the cap's reactance is huge, so almost all the voltage appears across it and the signal passes; at high frequencies the cap's reactance shrinks toward a short, the output collapses, and the signal is shunted away to ground.
Swap the two legs and you get the opposite behaviour: put C on top and R to ground, take the output across R, and now low frequencies are blocked by the cap while highs pass — a high-pass. Same two parts, swapped order, opposite filter. The corner frequency is the same friendly formula either way: f = 1 / (2 times pi times R times C). Pick R = 1.6 k and C = 100 nF and you land at f = 1 / (2 times pi times 1600 times 100 times 10^-9) which is about 1 kHz — the textbook starting point.
RC LOW-PASS RC HIGH-PASS
R C
Vin o-[===]--+--o Vout Vin o-||--+--o Vout
| |
=== C [===] R
| |
GND GND
low f : C blocks, V on C low f : C blocks, no V on R
-> output passes -> output small
high f: C ~ short high f: C ~ short
-> output shunted -> full V on R passes
f_cutoff = 1 / (2 x pi x R x C) (both, same formula)Passive filters are cheap, rugged and need no power, which is why they live everywhere from a tweeter's crossover to the snubber across a relay. Their honest limits, though, are exactly what the rest of this guide is about: a single RC gives only a gentle 6 dB-per-octave roll-off (a first-order slope), it always loses a little signal in the passband, and — most painfully — it sags the moment you connect a real load to its output.
The catch with passive filters
Here is the trap. That tidy f = 1 / (2 times pi times R times C) assumes nothing is hanging off the output. But the next stage has its own input resistance, and it sits in parallel with your C-to-ground (or in parallel with R). This is the loading effect: the load quietly joins your divider and changes both the gain and the corner you so carefully chose. A high-impedance load barely matters; a low-impedance one can drag the cutoff and flatten the passband badly.
Cascading two RC sections to get a steeper slope makes it worse, not simply better. The second section loads the first, so the two corners interact and smear instead of stacking cleanly; the combined roll-off near the corner is mushier than the textbook 12 dB-per-octave you hoped for, and the passband droops more. With only resistors and caps you cannot get a sharp, well-behaved knee — and you certainly cannot get Q greater than the gentle 0.5 of a plain double-RC, because there is nothing in the circuit to feed energy back and create a resonant peak.
The classic escape used to be the inductor. An LC filter pairs a cap with an inductor, and because the two store energy in opposite ways — the cap resists voltage change, the inductor resists current change — they can hand energy back and forth and produce a real resonant peak, a genuine Q above 0.5 and a sharp second-order corner with no op-amp at all. That is why LC still rules at radio frequencies. At audio and below, though, the inductors you would need are big, heavy, expensive, and pick up hum like an antenna — which is precisely the gap the active filter was invented to fill.
Active: add an op-amp, win the loss back
An active filter keeps the resistors and capacitors that do the frequency-shaping but adds an op-amp (and therefore a power supply) to the circuit. The op-amp earns its keep in three ways at once, and it helps to see them as three separate gifts rather than one vague improvement.
- Buffering. The op-amp's huge input resistance reads the RC network without loading it, and its near-zero output resistance drives the next stage hard. The corner you designed stays put no matter what hangs off the output — the loading effect simply vanishes.
- Gain. A passive filter can only ever lose signal (it has no power source of its own). The op-amp can pass the passband at unity or boost it — a low-pass with a gain of 5 in one stage, say — so filtering and amplifying happen together instead of needing a separate amp.
- Synthesized resonance. Through feedback the op-amp can make the circuit behave as if an inductor were present, producing a controllable Q and a sharp second-order corner — all from cheap resistors and caps, with no bulky coil.
That third gift is the magic trick worth pausing on. A capacitor and a resistor can never resonate on their own. But wrap an op-amp around them with the right feedback path and the circuit develops a peak that looks for all the world like an LC resonance — a synthetic inductor made of feedback. This is the whole idea behind the topologies you will study next, and it is why active filters dominate everything from DC up to roughly a megahertz, where op-amps are fast enough to keep up.
Two workhorse topologies
Almost every second-order active filter you meet is one of two arrangements. The Sallen-Key wires the op-amp as a non-inverting buffer (or a small gain) and feeds one capacitor back from the output to create the resonance. It is the friendliest to design — for unity gain you can even use equal resistors and equal caps — and it keeps the signal in phase. Its honest weakness is that at high gain or high Q it gets touchy about op-amp imperfections and can peak more than you intended.
The multiple-feedback (MFB) topology instead wires the op-amp as an inverter with two feedback paths. It inverts the signal (a 180-degree phase shift) and tends to behave better at higher Q, with less sensitivity to the part values, which is why it is often preferred for sharp band-pass sections. The trade is that its component values are less intuitive and it can demand more from the op-amp's gain at high frequency. Neither is universally best — choosing between them is a real design decision you will make in guide 5.
What active filters cost you, and where they earn their place
Active does not mean free. The op-amp brings its own honest limits, and ignoring them is how a paper-perfect filter misbehaves on the bench. The op-amp needs a power supply and adds quiescent current. Its gain-bandwidth product is finite, so as your filter's corner climbs the op-amp runs out of loop gain and your nice response sags or peaks unexpectedly — a 1 MHz part is no good building a sharp filter at 500 kHz. It adds its own noise and offset, it can only swing between its supply rails, and like any feedback amplifier it can ring or even oscillate if the loop carries too much phase shift.
Keep one truth from guide 1 firmly in mind no matter which family you choose: there are no brick walls. Neither a passive nor an active filter can pass everything below a corner untouched and kill everything above it instantly. Every real filter has a finite roll-off, a passband that is never perfectly flat, and a phase response that delays different frequencies by different amounts. Making the corner sharper always costs something — passband ripple, or phase distortion, or a touchier circuit — which is the whole subject of the Butterworth, Chebyshev and Bessel choice in guide 4.
With all that said, active filters earn their keep across an enormous range of everyday jobs. An active low-pass is the anti-alias filter that guards an analog-to-digital converter, and the smoothing reconstruction filter that turns a DAC's steps back into a clean curve. They carve out audio bands in a graphic equalizer, scrub mains hum from a sensor signal, and shape the tone in countless instruments. The shaping is still done by humble R and C — the op-amp just makes those parts behave the way you actually wanted them to.