Designing a Sallen Key Bandpass Filter: Theory, Math, and Audio Applications

A Sallen-Key bandpass filter is an active analog circuit topology that uses a single operational amplifier combined with a specific resistor-capacitor network to isolate and amplify a targeted range of frequencies while attenuating signals outside that band. Unlike passive LC filters that require bulky, lossy inductors and suffer from insertion loss, this topology replaces the inductor with an op-amp and capacitors, providing signal gain, high input impedance, and low output impedance without loading down the preceding stage. Beginners commonly confuse the single-op-amp Sallen-Key bandpass with the Multiple Feedback (MFB) topology (which inverts the signal and handles high Q-factors better) or assume it requires two op-amps (confusing it with a cascaded Sallen-Key high-pass and low-pass pair).

Topology Showdown: Sallen-Key vs. Multiple Feedback vs. State-Variable

Before soldering components, you must choose the right active filter topology for your application. The Sallen-Key is excellent for low-to-moderate Q applications where non-inverting phase and minimal component count are priorities. However, it is not a universal solution. Below is a data-dense comparison of the three most common active bandpass topologies used in audio and sensor conditioning.

As noted in the Wikipedia entry on Sallen-Key topology, the primary limitation of the single-amplifier Sallen-Key bandpass is that the gain and Q-factor are mathematically coupled. If you need a Q-factor above 10, the spread between your resistor values becomes so extreme that standard 1% component tolerances will detune the filter entirely. For Q > 10, switch to MFB or State-Variable.

The Core Math: Center Frequency, Q-Factor, and Gain

The standard Sallen-Key bandpass configuration utilizes two capacitors and three resistors around a single non-inverting op-amp. To design the circuit, we rely on three primary equations that define the center frequency ($f_0$), the Quality factor ($Q$), and the mid-band voltage gain ($A_0$).

• Center Frequency ($f_0$): The geometric mean of the low and high -3dB cutoff frequencies. Calculated as $f_0 = \frac{1}{2\pi \sqrt{R_1 R_2 C_1 C_2}}$.
• Quality Factor ($Q$): Defines the bandwidth relative to the center frequency ($Q = f_0 / BW$). A higher Q means a narrower, steeper peak.
• Mid-Band Gain ($A_0$): The voltage amplification at exactly $f_0$. In the standard equal-capacitor design, gain is set by the feedback resistors on the non-inverting input.

Worked Numeric Example: 1 kHz Audio Extractor

Let's design a practical bandpass filter to isolate a 1 kHz test tone from a noisy audio signal. We want a moderately narrow peak to reject adjacent frequencies, so we will target a Q-factor of 5. We also want a mid-band gain of 10 (20 dB) to boost the extracted signal for an ADC.

Design Targets:

• $f_0$ = 1,000 Hz
• $Q$ = 5
• $A_0$ = 10

Step 1: Select the Capacitors
For audio frequencies, 10 nF (0.01 µF) is the sweet spot. It avoids the parasitic inductance of large electrolytics and the stray capacitance vulnerabilities of sub-100 pF ceramics. We will set $C_1 = C_2 = 10\text{nF}$.

Step 2: Calculate the Resistors
Using the standard equal-capacitor Sallen-Key bandpass design equations (sourced from the Analog Devices Filter Wizard documentation):

• $R_1 = \frac{Q}{2\pi f_0 C A_0} = \frac{5}{2\pi (1000)(10 \times 10^{-9})(10)} = 7,957\ \Omega$
• $R_2 = \frac{Q}{2\pi f_0 C (2Q^2 - A_0)} = \frac{5}{2\pi (1000)(10 \times 10^{-9})(2(25) - 10)} = 198.9\ \Omega$
• $R_3 = \frac{Q}{\pi f_0 C} = \frac{5}{\pi (1000)(10 \times 10^{-9})} = 159,154\ \Omega$

Step 3: Map to Real-World E24/E96 Values
We cannot buy a 7,957 Ω resistor off the shelf. We map to the nearest standard values:

• R1: 8.06 kΩ (E96 series) or 8.2 kΩ (E24 series). Let's use 8.2 kΩ for wide availability.
• R2: 200 Ω (Exact E24 value).
• R3: 160 kΩ (Exact E24 value).

Step 4: Verify Op-Amp Selection
Applying our GBW rule: $100 \times 1000 \times 5 = 500\text{ kHz}$. The ubiquitous NE5532 audio op-amp has a GBW of 10 MHz and a low noise floor (5 nV/√Hz), making it vastly overqualified for the GBW requirement and perfect for audio. If this were a high-impedance piezo sensor circuit, we would swap to a JFET-input TL072 to minimize input bias current errors.

Where You Meet This in Practice (and Where It Fails)

You will frequently encounter Sallen-Key bandpass filters in analog synthesizers, active guitar pedal EQ stages, and industrial vibration monitoring circuits. In a CNC machine telemetry setup, for example, a Sallen-Key filter tuned to 120 Hz might be used to isolate the specific vibration signature of a spindle motor passing through a bearing defect frequency, rejecting the 60 Hz mains hum and the 2 kHz PWM noise from the VFD.

However, the transition from simulation to breadboard is where most designs fail. Here are the physical realities you must account for:

1. Stray Breadboard Capacitance: A standard solderless breadboard has roughly 2 pF to 5 pF of stray capacitance between adjacent rows. If your design relies on a 10 pF capacitor for a high-frequency filter, the breadboard's parasitics will shift your center frequency by 30%. Always solder high-frequency or high-Q Sallen-Key filters on a PCB or perfboard.
2. Component Tolerance Stacking: If $R_1$ and $R_3$ are both 1% resistors, but they drift in opposite directions due to temperature coefficients, your Q-factor will peak or collapse. For Q > 5, use 0.1% tolerance metal film resistors and C0G/NP0 dielectric capacitors (never X7R or Y5V, which exhibit severe voltage and temperature coefficients).
3. Slew Rate Limiting: If you design a high-gain bandpass filter ($A_0 = 50$) and feed it a 1V peak-to-peak signal at $f_0$, the op-amp must output 50V. Since it is likely powered by ±12V, it will clip. Even if powered by ±15V, the op-amp's slew rate (e.g., 9 V/µs for the NE5532) must be fast enough to handle the $dV/dt$ of the amplified sine wave, otherwise intermodulation distortion will corrupt your signal.