The Core Decision: Active Sallen-Key vs. Passive RLC vs. MFB
Choosing the right topology prevents you from fighting your own components later. Here is the decision path that terminates in a concrete pick based on your application constraints.
| Application Constraint | Recommended Topology | Concrete Part Pick |
|---|---|---|
| Audio/Sensor (20Hz - 20kHz), low component count, unity gain | Active Sallen-Key | NE5532 Op-Amp, C0G Caps |
| High Q (>10), narrow passband, requires high gain | Active Multiple Feedback (MFB) | TL072 Op-Amp, 1% Film Resistors |
| RF / UHF (>1MHz), high power, no power rails available | Passive RLC (LC Tank) | Air-core inductor, NP0 Caps |
| Ultra-low power, battery-operated IoT sensor node | Switched-Capacitor Filter IC | Texas Instruments MF10 |
Anatomy of the Sallen-Key Band Filter Circuit
The Sallen-Key bandpass relies on a specific arrangement of two resistors, two capacitors, and a non-inverting op-amp buffer. Unlike the Multiple Feedback (MFB) topology, which routes feedback to the inverting input and can become unstable at high gains, the Sallen-Key feeds the signal into the non-inverting input, making it inherently stable at unity gain.
Topology and Node Labels
- Vin: AC input signal source.
- Node A: The primary summing junction. Connects to C1 (from Vin), R2 (to GND), and R3 (feedback from Vout).
- Node B: The non-inverting input (+) of the op-amp. Connects to R1 (from Node A) and C2 (to GND).
- Vout: Op-amp output. Fed back through R3 to Node A to create the resonance peak.
- GND: System ground. R2 and C2 terminate here.
Why Sallen-Key over MFB? According to Analog Devices Tutorial MT-223, the Sallen-Key topology exhibits lower sensitivity to the op-amp's open-loop gain roll-off at higher frequencies. MFB is superior if you need a high Q (>5) and high gain simultaneously, but for Q values under 5, Sallen-Key requires less precise component matching and won't oscillate if your op-amp's phase margin dips.
Component Behavior and Failure Modes at the Extremes
Understanding how each component shapes the transfer function is critical for tuning. More importantly, knowing what happens when a component fails (or is soldered incorrectly) saves hours of debugging.
| Component | Effect on Center Freq (f0) | Effect on Q (Bandwidth) | Failure Mode: Open | Failure Mode: Short |
|---|---|---|---|---|
| C1 (Input Coupling) | Inversely proportional | Minor shift | No signal passes (Dead circuit) | Passes DC; circuit becomes a low-pass filter |
| R2 (Node A to GND) | Inversely proportional | Lowers Q (wider band) | No DC bias path; op-amp output saturates to rail | Node A grounded; no signal reaches op-amp |
| R1 (Node A to B) | Inversely proportional | Lowers Q (wider band) | No signal reaches non-inverting input | Node A and B merge; alters Q drastically |
| C2 (Node B to GND) | Inversely proportional | Minor shift | Passes high-freq unattenuated; becomes high-pass | Node B grounded; output is flat zero |
| R3 (Feedback) | No direct effect | Primary Q control | Q drops to <0.5; very wide, flat response | Q approaches infinity; circuit oscillates |
Design Walkthrough: 1 kHz Audio Bandpass (Real Values)
Let's design a band filter circuit with a center frequency ($f_0$) of 1 kHz and a Q factor of 2 (which gives a bandwidth of 500 Hz). We will use the simplified unity-gain Sallen-Key equations where $C1 = C2 = C$ and $R1 = R2 = R$.
Step 1: Select the Capacitor Dielectric
Do not use X7R or Y5V ceramic capacitors for audio or precision sensor filters. These dielectrics exhibit severe voltage coefficients (capacitance drops as voltage rises) and microphonics (they act as piezoelectric microphones, injecting noise). Always select C0G/NP0 ceramics or polypropylene film capacitors. We will choose 10nF (0.01µF) C0G as it is a standard, easily sourced value.
Step 2: Calculate the Resistors
The center frequency formula for this simplified topology is:
$$f_0 = \frac{1}{2 \pi R C}$$
Rearranging to solve for R:
$$R = \frac{1}{2 \pi f_0 C} = \frac{1}{2 \pi (1000) (10 \times 10^{-9})} = 15,915 \Omega$$
Step 3: Map to E24 Standard Values
The nearest standard E24 resistor value is 16kΩ. Let's verify the actual center frequency with 16kΩ:
$$f_0 = \frac{1}{2 \pi (16000) (10 \times 10^{-9})} = 994.7 \text{ Hz}$$
This is well within the 1% tolerance of standard 10nF C0G capacitors. For the feedback resistor R3, which sets the Q factor, the formula for unity gain Sallen-Key Q is $Q = \frac{1}{3 - K}$, where K is the feedback ratio. To achieve Q=2, we calculate R3 to be approximately 27kΩ (E24 value).
For deeper multi-stage cascading or complex pole-zero placement, the TI Webench Filter Designer is the industry-standard tool to automate these E-series mappings and simulate the Bode plot before you cut a PCB.
Breadboard Testing Protocol
Testing an active band filter circuit on a solderless breadboard introduces parasitic capacitance (typically 2pF to 5pF between adjacent rows). While negligible at 1 kHz, this will skew your results if you are designing for >100 kHz. Follow this exact sequence to verify your build.
- Power and Bias Check: With no AC input, power the NE5532 with ±9V (or a single 9V with a virtual ground rail at 4.5V). Measure the Vout pin with a multimeter. It must read 0V (dual supply) or 4.5V (single supply). If it reads near the rails, your DC bias path (R2) is broken.
- Inject Mid-Band Signal: Connect a function generator to Vin. Set it to a 1 kHz sine wave, 1Vpp. Connect an oscilloscope to Vout. You should see approximately 1Vpp (unity gain). If the amplitude is wildly different, check R3 (feedback).
- Sweep the Lower Cutoff: Slowly decrease the function generator frequency. Note the frequency where the output amplitude drops to 0.707V (-3dB point). For our design, this should be around 750 Hz.
- Sweep the Upper Cutoff: Increase the frequency past 1 kHz. Note the -3dB point on the high side. It should be around 1.35 kHz. The difference between the upper and lower -3dB points is your Bandwidth (approx 600 Hz, yielding a measured Q of ~1.6 to 1.8, accounting for breadboard parasitics and component tolerance).
- Check for High-Frequency Roll-off: Push the frequency to 100 kHz. The NE5532 has a slew rate of 9V/µs and a gain-bandwidth product of 10MHz. The signal should be heavily attenuated. If you see high-frequency noise or oscillation on the scope, your ground return paths are too long; move the op-amp's decoupling capacitors (100nF MLCC) directly across pins 4 and 8.
The Default Recommendation
Stop guessing component values and wrestling with inductor parasitics. If you need a band filter circuit for an audio, vibration, or general-purpose sensor application, build the unity-gain Sallen-Key topology.
- Op-Amp: NE5532 (Dual, low noise, high slew rate) or TL072 (if running on higher voltage rails >±12V).
- C1, C2: 10nF (0.01µF) 50V C0G/NP0 Ceramic Capacitors.
- R1, R2: 16kΩ 1/4W 1% Metal Film Resistors.
- R3 (Feedback): 27kΩ 1/4W 1% Metal Film Resistor.
- Decoupling: 2x 100nF X7R MLCCs (one from V+ to GND, one from V- to GND, placed within 5mm of the op-amp pins).
This exact combination guarantees a stable 994 Hz center frequency with a Q of roughly 1.8, requiring no trimmer potentiometers or complex tuning on the bench.
For foundational theory on how the passband and stopband interact, the Electronics Tutorials guide on Band Pass Filters provides excellent Bode plot visualizations to cross-reference with your oscilloscope readings. Stick to the C0G dielectrics, keep your ground nodes tight, and your filter will perform exactly as the math predicts.






