A fourth order Butterworth filter is an electronic signal processing circuit that attenuates frequencies beyond its cutoff point at a steep rate of -24 dB per octave (or -80 dB per decade) while maintaining a perfectly flat amplitude response in the passband. What this changes in a real circuit or installation is the ability to aggressively strip out high-frequency noise, eliminate PWM switching artifacts, or prevent ADC aliasing without introducing the passband ripple that distorts sensitive audio or precision sensor signals. Beginners and intermediate builders commonly confuse it with the Chebyshev filter (which trades passband ripple for an even steeper roll-off) or the Bessel filter (which sacrifices roll-off steepness to preserve a linear phase response and clean step transient).
The Math and Mechanics: Why 'Maximally Flat' Matters
The defining characteristic of any Butterworth filter is its 'maximally flat' magnitude response. In the passband, the gain does not ripple; it stays as close to 0 dB (unity) as mathematically possible before beginning its descent at the -3 dB cutoff frequency. A fourth order implementation achieves this by utilizing four poles in its transfer function. In the s-plane, these poles are spaced evenly along a semicircle in the left half-plane, ensuring stability while dictating the damping factors required for each cascaded stage.
Think of a 1st order filter like a single porous baffle in a wind tunnel that gradually slows down high-frequency air turbulence. A fourth order Butterworth is like a precisely tuned series of four acoustic chambers that let low-frequency rumbles pass perfectly uncolored, but completely choke out high-frequency hisses the exact moment they cross the threshold.
Because a fourth order filter rolls off at -24 dB/octave, a noise signal sitting exactly one octave above your cutoff frequency will be attenuated to roughly 6.3% of its original voltage. By two octaves, it is down to 0.4%. This makes it the gold standard for anti-aliasing and crossover networks where out-of-band energy must be killed rapidly.
Worked Numeric Example: 1 kHz Active Low-Pass Design
Let's design a practical 1 kHz low-pass fourth order Butterworth filter using the Sallen-Key topology. We will use the 'equal-component' variant (where R1=R2 and C1=C2 in each stage) because it simplifies sourcing and tuning, relying on the op-amp's non-inverting gain to set the required damping factors (Q).
For a 4th order Butterworth, the two cascaded 2nd-order stages require specific Q values: Stage 1 needs a Q of 0.707, and Stage 2 needs a Q of 1.307. In a Sallen-Key equal-component design, the gain (A) dictates the Q via the formula A = 3 - (1/Q).
- Stage 1 Gain (A1): 3 - (1 / 0.707) = 1.586
- Stage 2 Gain (A2): 3 - (1 / 1.307) = 2.235
We will target a cutoff frequency ($f_c$) of roughly 1 kHz. Using the formula $R = 1 / (2 \pi f_c C)$, if we select a standard 10 nF capacitor, our ideal resistor value is 15.91 kΩ. The closest standard 1% resistor is 15.8 kΩ, yielding a true cutoff of 1007 Hz.
| Component | Stage 1 (Low Q) | Stage 2 (High Q) | Notes / Tolerance |
|---|---|---|---|
| R1, R2 (Timing) | 15.8 kΩ | 15.8 kΩ | 1% Metal Film |
| C1, C2 (Timing) | 10 nF | 10 nF | 5% Polypropylene or C0G |
| Rg (Gain Set) | 10.0 kΩ | 10.0 kΩ | 1% Metal Film |
| Rf (Feedback) | 5.90 kΩ | 12.4 kΩ | 1% Metal Film (Yields A=1.59 & A=2.24) |
| Op-Amp | Texas Instruments OPA2134 (Dual) | FET-input, 8 MHz GBW, low THD | |
By cascading the output of Stage 1 directly into the input of Stage 2, the combined transfer function yields the target fourth order Butterworth response. The OPA2134 provides ample Gain Bandwidth Product (GBW) to ensure the op-amp's own internal roll-off does not interfere with the 1 kHz filter math. For budget audio builds, a TL072 is an acceptable substitute, though it will introduce slightly higher noise.
Where You Meet This in Practice
You will rarely see a fourth order Butterworth filter built with discrete passive components (inductors and capacitors) at low frequencies, as the required inductor values become physically massive and suffer from parasitic resistance. Instead, active implementations dominate specific high-stakes applications:
- ADC Anti-Aliasing: When feeding an analog sensor into a microcontroller's ADC (like the 12-bit SAR ADC on an STM32 or an external I2S ADC for an ESP32), Nyquist theory dictates you must filter out frequencies above half the sampling rate. A 4th order Butterworth ensures flat sensor data in the passband while brutally suppressing high-frequency EMI that would otherwise fold back into your digital readings as aliasing noise.
- Active Audio Crossovers: In bi-amplified studio monitors, a Butterworth response is often used to protect tweeters from low-frequency excursion damage. (Note: Many high-end speakers use the Linkwitz-Riley alignment, which is actually created by cascading two 2nd-order Butterworth filters to achieve a 4th-order slope with a -6 dB crossover point for perfect phase summation).
- Subwoofer Plate Amplifiers: The input stage of a powered subwoofer uses a low-pass Butterworth filter to strip out midrange and treble frequencies before they reach the class-D amplification stage, ensuring the driver only reproduces deep bass.
For deeper mathematical modeling and automated component selection, engineers frequently rely on the Texas Instruments WEBENCH Filter Designer, which calculates exact pole placements and suggests real-world standard component values.
Frequently Asked Questions
Why use a fourth order Butterworth filter instead of a second order?
A second order filter rolls off at -12 dB per octave. If your noise source is close to your signal frequency (e.g., a 2 kHz noise spike on a 1 kHz audio signal), a 2nd order filter will only attenuate that noise by about 7 dB—barely noticeable. A fourth order filter doubles that attenuation to roughly 14 dB at the same frequency gap, and the gap widens exponentially as frequency increases. You choose 4th order when the noise floor must be aggressively suppressed immediately after the cutoff point.
What is the difference between a fourth order Butterworth and a Linkwitz-Riley crossover?
While both exhibit a -24 dB/octave roll-off, their phase and amplitude behaviors at the crossover frequency differ. A standard 4th-order Butterworth filter is -3 dB down at the cutoff frequency. If you use a Butterworth LPF and HPF at the same frequency and sum their outputs, you get a +3 dB peak at the crossover point due to phase interaction. A Linkwitz-Riley (LR4) is designed to be -6 dB down at the crossover point, resulting in a perfectly flat 0 dB summed response and in-phase acoustic output. LR4 is essentially two cascaded 2nd-order Butterworth filters.
How do I cascade two second-order filters to make a fourth order Butterworth?
You cannot simply build two identical 2nd-order Butterworth filters (both with Q=0.707) and wire them in series. Doing so creates a 4th-order filter, but it will not be a Butterworth response; it will have a drooping passband and a steeper initial roll-off. To achieve a true 4th-order Butterworth response, the two cascaded 2nd-order stages must have different damping factors (Q values). As shown in the worked example above, Stage 1 must have a Q of 0.707, and Stage 2 must have a Q of 1.307. For more on active topologies, All About Circuits provides an excellent primer on Sallen-Key cascading.
Does a fourth order Butterworth filter introduce phase shift at the cutoff frequency?
Yes. Every pole in a filter introduces up to -90° of phase shift. A fourth order filter has four poles, meaning the maximum phase shift deep in the stopband approaches -360°. Exactly at the -3 dB cutoff frequency, a 4th-order Butterworth filter introduces a phase shift of -180°. If phase coherence is your primary concern (such as in digital communications or square-wave pulse processing), you should abandon the Butterworth topology and use a Bessel filter, which maintains a near-linear phase shift across the passband at the cost of a much gentler roll-off.
