A Sallen and Key filter is a second-order active analog filter topology that uses a single operational amplifier, two resistors, and two capacitors to shape frequency response without requiring inductors. In a real circuit, it changes the game by replacing bulky, lossy, and expensive passive inductors with cheap, precise RC networks while simultaneously providing a low-impedance output buffer to drive downstream stages. Beginners and even some intermediate designers commonly confuse it with the Multiple Feedback (MFB) topology—which inverts the signal and is highly sensitive to op-amp bandwidth limits—or mistakenly assume that simply cascading two first-order passive RC filters achieves the same result. (It doesn't; cascaded passive RC stages suffer from -3dB interaction and cannot achieve a Butterworth Q-factor of 0.707 without active buffering between them).
The Core Topology: How the RC Network Shapes the Signal
The standard Sallen and Key low-pass topology routes the input signal through a two-stage RC network (R1-C1 and R2-C2) into the non-inverting input of an op-amp configured as a voltage follower (unity gain) or a non-inverting amplifier. The magic that elevates this from a sluggish, overdamped passive filter to a sharp, tunable active filter is the feedback path.
Think of the feedback capacitor (often connected from the op-amp output back to the junction of R1 and R2) like a bouncer at a club. At low frequencies, the bouncer does nothing, and the signal passes through the resistors unimpeded. As the frequency approaches the cutoff threshold, the bouncer grabs energy from the output and shoves it back into the input junction, artificially boosting the signal right at the knee. This 'boost' is what allows the filter to achieve a maximally flat Butterworth response (Q = 0.707) or a resonant Chebyshev response (Q > 0.707) rather than rounding off prematurely.
Worked Numeric Example: 1 kHz Butterworth Low-Pass Design
Let's design a 2nd-order Butterworth low-pass Sallen and Key filter with a cutoff frequency (fc) of 1 kHz. We will use the equal-capacitor, fixed-gain variant because sourcing matched 1% capacitors is harder than sourcing 1% resistors.
Step 1: Choose the Capacitors
Select a standard capacitor value that yields practical resistor sizes (between 1kΩ and 100kΩ to minimize noise and loading). Let's choose C1 = C2 = 10 nF.
Step 2: Calculate the Resistors
For the equal-capacitor Sallen-Key topology, the resistor formula simplifies to:
R = 1 / (2 × π × fc × C)
R = 1 / (2 × 3.14159 × 1000 × 10 × 10-9) = 15,915 Ω
Step 3: Round to Real-World E96 Values
15.915kΩ isn't a standard value. Using the 1% E96 resistor series, the closest match is 16.0 kΩ (or you can use a 15kΩ and 1kΩ E24 resistor in series). Let's use R1 = R2 = 16.0 kΩ (1%).
Step 4: Set the Op-Amp Gain
For a Butterworth response, the required gain is K = 3 - (1/Q) = 3 - 1.414 = 1.586.
The non-inverting gain formula is K = 1 + (Rf / Ri).
Let's set Ri = 10.0 kΩ.
Rf = (K - 1) × Ri = 0.586 × 10,000 = 5,860 Ω.
The closest 1% E96 resistor is 5.90 kΩ.
Where You Meet This in Practice
You won't just find this topology in textbooks; it is the backbone of modern mixed-signal and audio hardware.
- ADC Anti-Aliasing: Placed immediately before a 12-bit or 16-bit SAR ADC (like the ADS1115) to brutally attenuate high-frequency RF noise and switching harmonics that would otherwise fold back into your Nyquist band and corrupt your DC measurements.
- Audio Crossovers and Subwoofers: Used in active subwoofer preamplifiers to roll off everything above 80 Hz or 120 Hz before the signal hits the power amplifier stage, ensuring the driver only reproduces deep bass.
- Sensor Signal Conditioning: Strain gauges, thermocouples, and load cells output millivolt signals riding on a sea of 50/60 Hz mains hum and high-frequency EMI. A low-pass Sallen and Key filter strips the noise while the op-amp buffers the high-impedance sensor for the microcontroller.
Sallen-Key vs. Multiple Feedback (MFB) Decision Matrix
Designers frequently debate whether to use Sallen-Key or the Multiple Feedback (MFB) topology. While both use one op-amp and achieve a second-order response, their behavior at the limits of silicon physics is vastly different.
| Criteria | Sallen and Key Filter | Multiple Feedback (MFB) |
|---|---|---|
| Signal Phase | Non-inverting (0° shift) | Inverting (180° shift) |
| Sensitivity to Op-Amp GBW | Low (Tolerates slower, cheaper op-amps) | High (Requires high GBW to maintain Q at high frequencies) |
| High-Frequency Bypass | Poor (At extreme HF, the op-amp dies and the RC network passes noise) | Excellent (The feedback capacitor inherently shunts HF noise to ground even if the op-amp rails) |
| Component Spread | Requires tight capacitor matching or specific gain ratios | Requires wide resistor ratios for high-Q designs |
IF you are designing an audio filter, a low-frequency sensor conditioner (<10 kHz), or need a non-inverting signal path → Pick Sallen-Key. Use the TL072 (for ±15V analog audio) or the MCP6002 (for 3.3V MCU ADC anti-aliasing).
IF you are designing an RF-adjacent anti-aliasing filter (>100 kHz), require high attenuation of out-of-band EMI, or need an inverted signal → Pick MFB. Use a high-speed voltage feedback op-amp like the OPA838.
Component Selection: Op-Amps and Capacitor Dielectrics
The math only gets you 80% of the way to a working filter. The physical components you select dictate whether your filter performs as simulated or turns into a noisy, drifting mess. For a deep dive into active filter theory and component sensitivities, the Active Low Pass Filter guide on Electronics Tutorials provides excellent foundational math, but bench experience reveals two critical hardware gotchas.
1. The Capacitor Dielectric Trap (X7R vs. C0G)
Never use X7R, X5R, or Y5V multilayer ceramic capacitors (MLCCs) in the frequency-determining network of a Sallen and Key filter. X7R capacitors exhibit severe voltage coefficient of capacitance (a 10 µF X7R cap might drop to 3 µF at its rated voltage) and are highly microphonic, converting mechanical board vibration into piezoelectric noise.
The Fix: Always specify C0G (also known as NP0) dielectric capacitors for C1 and C2. C0G is a Class I dielectric with near-zero temperature drift and no voltage coefficient. If your required capacitance is too large for C0G (which tops out around 100 nF in small SMD packages), switch to WIMA polypropylene film capacitors for audio, or Panasonic PPS film for precision DC applications.
2. The Op-Amp Gain Bandwidth Product (GBW) Rule
An op-amp's open-loop gain drops as frequency increases. If your op-amp runs out of gain near your filter's cutoff frequency, the Q-factor will collapse, and your Butterworth filter will turn into a sloppy, overdamped curve. According to Analog Devices' filter design literature, you must select an op-amp where the GBW satisfies the following rule of thumb:
GBW > 100 × fc × Q
For our 1 kHz Butterworth example (Q = 0.707), the minimum GBW is 100 × 1000 × 0.707 = 70.7 kHz. A cheap LM358 (GBW ≈ 1 MHz) easily clears this hurdle. However, if you were designing a 100 kHz Butterworth filter, you would need an op-amp with a GBW of at least 7.07 MHz, pushing you away from general-purpose parts and toward dedicated high-speed amplifiers like the OPA211.
Frequently Asked Questions
Can I use a Sallen and Key filter for high-pass applications?
Yes. The topology is identical, but the positions of the resistors and capacitors are swapped. The same design equations apply, though high-pass Sallen-Key filters are more susceptible to high-frequency op-amp slew-rate limiting if the input contains large, fast transients.
Why not just use a digital FIR/IIR filter in my microcontroller?
Digital filters are excellent, but they require the analog signal to be digitized first. If you don't use an analog Sallen and Key filter to band-limit the signal before the ADC, any high-frequency noise (like a nearby 2.4 GHz WiFi module or switching regulator) will alias into your digital baseband and cannot be removed by software after the fact. Analog filters protect the ADC; digital filters process the data.
What happens if I use 5% tolerance resistors and 10% capacitors?
Your cutoff frequency will shift, and your Q-factor will deviate. A 10% capacitor variance can easily shift a Butterworth response (Q=0.707) into a Bessel (Q=0.57) or Chebyshev (Q>0.8) response, causing passband ripple or a sluggish roll-off. For anything beyond hobbyist audio tone controls, 1% resistors and 5% (or better) C0G capacitors are mandatory.
