Designing a Butterworth 2nd Order Filter: Sallen-Key Topology and Component Math

A Butterworth 2nd order filter is an analog signal processing circuit that provides a maximally flat frequency response in the passband while attenuating unwanted frequencies at a rate of -40 dB per decade. When you insert this into a real circuit, it changes the signal chain by cleanly stripping out high-frequency noise, switching ripple, or aliasing artifacts without introducing the amplitude 'ripple' or 'peaking' that other filter topologies inject into your passband. If you are reading raw sensor data or building an audio crossover, this flat passband is exactly what you need to preserve signal integrity before it hits your ADC or amplifier.

The Core Math and Sallen-Key Topology

To build an active Butterworth 2nd order filter, the industry standard is the Sallen-Key topology. It requires only one operational amplifier, two resistors, and two capacitors. The defining characteristic of a Butterworth response is its damping factor, which translates to a Quality factor (Q) of exactly 0.707 (or $1/\sqrt{2}$). This specific Q value is the mathematical sweet spot that yields the flattest possible passband right up to the cutoff frequency ($f_c$), with no peaking at the corner.

In a unity-gain Sallen-Key configuration (where the op-amp output is tied directly to its inverting input), the Q factor is determined entirely by the ratio of the two capacitors, assuming the two resistors are equal ($R_1 = R_2 = R$). The governing equations are:

• Cutoff Frequency: $f_c = \frac{1}{2\pi R \sqrt{C_1 C_2}}$
• Quality Factor: $Q = \frac{1}{2}\sqrt{\frac{C_2}{C_1}}$

To achieve the Butterworth Q of 0.707, the capacitor ratio $C_2 / C_1$ must be exactly 2. This makes the math incredibly forgiving for bench work, as you only need to source a capacitor pair where one is roughly double the value of the other.

Worked Numeric Example: 1 kHz Low-Pass Design

Let's design a low-pass filter to clean up a noisy 1 kHz sensor signal before it enters a microcontroller. We will target a cutoff frequency ($f_c$) of 1,000 Hz.

1. Choose the Resistors: We want impedances high enough to not load the source, but low enough to avoid excessive thermal noise and op-amp bias current errors. 10 kΩ is the standard bench default. So, $R_1 = R_2 = 10\text{ k}\Omega$.
2. Calculate Ideal Capacitors: Using the $f_c$ formula and substituting $C_2 = 2C_1$, we get $f_c = \frac{1}{2\pi R C_1 \sqrt{2}}$. Solving for $C_1$ yields approximately 11.25 nF. Therefore, $C_2$ should be 22.5 nF.
3. Map to Real E24 Components: You cannot buy an 11.25 nF capacitor off the shelf. We step to the nearest standard E24 values: $C_1 = 10\text{ nF}$ and $C_2 = 22\text{ nF}$.
4. Verify the Real-World Result: Plugging 10 nF and 22 nF back into the equations gives us an actual cutoff frequency of 1,074 Hz and a Q factor of 0.74.

Where You Meet This in Practice

You will rarely see a Butterworth 2nd order filter used for power supply filtering (where simple LC pi-filters dominate), but it is ubiquitous in signal conditioning. Here is where it earns its keep on the workbench:

• ESP32 and Arduino ADC Anti-Aliasing: The SAR ADCs inside microcontrollers like the ESP32 are notoriously noisy and susceptible to aliasing. A 2nd order Butterworth low-pass filter set just below the Nyquist frequency of your sampling rate prevents high-frequency RF and switching noise from folding back into your DC measurements.
• Audio Subwoofer Crossovers: In active speaker builds, you need to roll off the high frequencies going to the subwoofer amplifier. The Butterworth response ensures the bass frequencies aren't artificially boosted (peaked) right before the cutoff, preserving a natural, flat low-end response.
• Load Cell and Thermocouple Amplification: When amplifying millivolt signals from a Wheatstone bridge, high-frequency EMI from nearby motors or VFDs can easily swamp the signal. A 2nd order Butterworth filter at the front end of an instrumentation amplifier rejects this noise without distorting the slow-moving DC weight or temperature data.

Common Confusions: Butterworth vs. Chebyshev vs. Bessel

The most common mistake hobbyists make is assuming 'filter' just means 'removes high frequencies.' The shape of the removal matters immensely. People frequently confuse the Butterworth with the Chebyshev or Bessel responses, leading to ruined audio phase or poorly timed control loops.

If you are filtering digital pulses or square waves where preserving the exact timing of the rising edge is critical, a Butterworth filter will cause the edge to ring and overshoot. In that specific scenario, you must switch to a Bessel filter. For further reading on selecting the right active topology, the Analog Devices ADIsimPE Filter Wizard is an excellent tool for visualizing these phase differences before you solder a single component.