A bandstop notch filter is a circuit that severely attenuates a specific, narrow range of frequencies while allowing all lower and higher frequencies to pass through unchanged. In a real circuit, it changes the signal by surgically excising a single interfering frequency—like 60Hz mains hum or a specific RF carrier—without rolling off the high-frequency content the way a low-pass filter would. Beginners commonly confuse a narrow notch filter with a wide band-reject filter (which cuts entire octaves of bandwidth) or a low-pass filter (which acts like a lawnmower cutting everything above a certain height, whereas a notch filter is a scalpel removing one specific weed).

The Bottom Line: If you need to kill a single, persistent, narrow-band interference without degrading the rest of your signal's bandwidth, you need a notch filter. If you need to roll off all high-frequency noise, you need a low-pass filter.

The Anatomy and Q-Factor of a Notch Filter

Every notch filter is defined by two primary parameters: the center frequency ($f_c$) and the Quality factor ($Q$). The center frequency is the exact point of maximum attenuation. The Quality factor dictates how "sharp" or narrow the notch is.

Quality Factor Formula: $Q = f_c / BW$
Where $BW$ is the -3dB bandwidth of the rejected frequencies. A high $Q$ (e.g., $Q > 10$) yields a very narrow, deep spike of attenuation. A low $Q$ yields a wide, shallow dip.

In passive designs, achieving a high $Q$ is nearly impossible without introducing massive insertion loss. The signal energy is absorbed by the resistive elements, resulting in a shallow notch that might only drop the interfering signal by 10dB or 15dB. To achieve the 40dB to 60dB of attenuation required to actually "kill" a hum or interference spike, we use active topologies that incorporate operational amplifiers to buffer the signal and inject positive feedback, artificially sharpening the $Q$ without destroying the passband gain.

Worked Numeric Example: Designing a 60Hz Hum Killer

Let’s design an active Twin-T notch filter to eliminate 60Hz mains hum from an audio preamplifier. The Twin-T network uses two interlaced T-networks: one low-pass (R-R-C) and one high-pass (C-C-R).

Step 1: Choose your capacitor.
Capacitors come in fewer standard values than resistors, so we start here. Let’s select C = 100nF (0.1µF) film capacitors. Film is mandatory here; ceramic capacitors introduce microphonic noise and voltage coefficient distortion in audio paths.

Step 2: Calculate the main resistor value.
The formula for the Twin-T center frequency is $f_c = 1 / (2 \pi R C)$.
Rearranging for R: $R = 1 / (2 \pi f_c C)$
$R = 1 / (2 \times \pi \times 60 \times 100 \times 10^{-9})$
$R = 26,525.8 \Omega$

Step 3: Select standard 1% values and build the sub-networks.
The closest standard E96 1% resistor is 26.7kΩ.
The Twin-T topology requires specific ratios: two $R$ resistors, one $R/2$ resistor, two $C$ capacitors, and one $2C$ capacitor.

  • R branch: Two 26.7kΩ 1% metal film resistors.
  • R/2 branch: Two 26.7kΩ resistors in parallel (yielding 13.35kΩ), or a single 13.3kΩ 1% resistor.
  • C branch: Two 100nF film capacitors.
  • 2C branch: Two 100nF film capacitors in parallel (yielding 200nF).

Step 4: Dial in the exact frequency.
Because 26.7kΩ shifts our $f_c$ slightly to 59.6Hz, a fixed resistor build will miss the 60.0Hz target by a fraction of a Hertz, reducing notch depth. On the bench, replace one of the 26.7kΩ resistors in the main branch with a 20kΩ fixed resistor in series with a 10kΩ multi-turn cermet trimpot. Inject a 60Hz sine wave from your function generator, monitor the output on an oscilloscope, and tweak the trimpot until the waveform hits minimum amplitude.

Where You Meet This in Practice

You will encounter and need to deploy notch filters across several distinct engineering domains:

  • Audio Engineering & Live Sound: Guitar amplifiers and studio preamps often pick up 50Hz (EU/UK) or 60Hz (US) electromagnetic interference from mains wiring. A high-Q active notch filter removes this hum without dulling the high-frequency "crispness" of a cymbal or acoustic guitar, which a low-pass filter would destroy.
  • Biomedical Instrumentation: Electrocardiogram (ECG) and Electroencephalogram (EEG) machines measure millivolt-level biological signals. The human body acts as an antenna for mains hum. Medical front-ends use ultra-high-Q notch filters (often digital or switched-capacitor) to strip 50/60Hz noise without clipping the critical ST-segment data of the heartbeat.
  • RF Front-Ends and SDR: If you are operating a Software Defined Radio (SDR) near a high-power local FM broadcast tower (e.g., 97.1 MHz), that strong signal will desensitize your receiver's low-noise amplifier (LNA) through intermodulation distortion. An LC or SAW notch filter tuned precisely to 97.1 MHz is placed at the antenna input to reject the local tower while passing the rest of the RF spectrum.

Decision Tree: Picking the Right Notch Topology

Choosing the wrong topology will result in excessive phase shift, insertion loss, or complete failure to attenuate the target frequency. Use this matrix to select your approach.

Application Scenario Frequency Range Recommended Topology Concrete Part / Implementation Pick
Audio Mains Hum (Guitar, Preamps, Synths) 50Hz - 60Hz Active Twin-T (Op-Amp buffered) TL072 or OPA1612 with 1% metal film resistors and WIMA film caps.
Biomedical / Precision Sensor Front-End 50Hz - 60Hz Switched-Capacitor Filter LTC1062 or MAX274 (Clock-tuned, eliminates resistor-matching headaches).
RF Interference (SDR, Ham Radio, IoT) 1 MHz - 3 GHz LC Trap / SAW Filter Mini-Circuits BFCN Series (Ceramic chip) or custom air-core LC trap.
Digital Audio / DSP Processing Any (Programmable) FIR / IIR Digital Filter ADAU1701 SigmaDSP or STM32 running CMSIS-DSP library.
Default Recommendation: If you are debugging a bench prototype or building an audio pedal and need to kill 60Hz hum, do not overthink it. Build an active Twin-T filter using a TL072 dual op-amp. It requires no specialized clocking, runs cleanly off a standard ±9V or ±15V split supply, and costs under $2 in components.

Component Tolerances: Why Your Notch is Shallow

The most common failure mode when building an analog notch filter on the bench is a "shallow" notch. You expect 60dB of attenuation, but your spectrum analyzer only shows a 15dB dip. This is almost never a topology error; it is a component tolerance error.

The depth of a passive or active Twin-T notch relies on the exact mathematical cancellation of the two T-networks. If the resistors and capacitors are not perfectly matched, the signals do not cancel completely at $f_c$. According to Analog Devices' MT-223 Tutorial on Notch Filters, the relationship between component tolerance and maximum theoretical attenuation is brutal:

  • 5% Tolerance Components: Yields a maximum notch depth of roughly 15 dB to 20 dB. (Useless for serious hum removal).
  • 1% Tolerance Components: Yields a maximum notch depth of roughly 40 dB. (Acceptable for most audio applications).
  • 0.1% Tolerance Components: Yields a maximum notch depth of 60+ dB. (Required for precision instrumentation).

Furthermore, as noted in Texas Instruments' Op Amps for Everyone design guides, the op-amp's own open-loop gain and common-mode rejection ratio (CMRR) will cap your maximum attenuation. A standard LM741 will bottleneck your notch depth to about 30dB regardless of how perfect your resistors are. This is why we specify modern, high-slew-rate, low-noise op-amps like the TL072 (for general audio) or the OPA1612 (for high-fidelity studio gear) when designing active topologies.

Always buy 1% metal film resistors (like the Yageo MFR-25 series) and 5% or better film capacitors. If you are using 5% capacitors because they were in your junk bin, you must use a multi-turn trimpot in the R/2 branch to manually balance the network and force the cancellation deeper, trading off exact center-frequency accuracy for notch depth.