Notch frequency is the exact center frequency where a band-stop filter attenuates a signal the most, effectively 'notching' it out of the frequency spectrum. When you are building an audio preamp or conditioning a sensor signal, broadband low-pass filters often destroy the high-frequency data you actually need. A notch filter acts like a highly selective bouncer at a club: it lets the low frequencies and high frequencies walk right in, but specifically blocks one exact frequency (and a narrow band around it) from passing. This targeted rejection changes how a circuit handles interference, allowing you to surgically remove 60 Hz mains hum or a specific switching regulator ripple without rolling off your entire upper bandwidth or introducing the phase-shift penalties of a steep low-pass filter.
The Math and Mechanics: Calculating a 60 Hz Notch
To see how notch frequency works on the bench, let's design a classic passive Twin-T network to eliminate 60 Hz US mains hum from an audio signal chain. The fundamental formula for the notch frequency ($f_n$) of a symmetric Twin-T filter is:
$f_n = \frac{1}{2 \pi R C}$
Let's target exactly 60 Hz. First, we select a standard capacitor value. A 270 nF (0.27 µF) capacitor is common, but to get closer to true 60 Hz with standard 1% resistors, we can parallel a 220 nF and a 47 nF capacitor to yield 267 nF. Now, we solve for R:
$R = \frac{1}{2 \pi \times 60 \text{ Hz} \times 267 \times 10^{-9} \text{ F}} \approx 9,934 \ \Omega$
The nearest standard E96 1% resistor value is 10.0 kΩ. Using 10.0 kΩ and 267 nF, our actual notch frequency shifts slightly to 59.6 Hz, which is perfectly adequate for catching 60 Hz hum.
However, a Twin-T network requires specific shunt components to balance the bridge and create a deep null. The topology requires:
- Series arms: Two 10.0 kΩ resistors and two 267 nF capacitors.
- Shunt arms: One R/2 resistor (4.99 kΩ standard 1% value) and one 2C capacitor (534 nF, achieved by paralleling 470 nF and 68 nF).
Never use X7R or Y5V ceramic capacitors for precision notch filters. These dielectrics exhibit severe voltage coefficients (capacitance drops as signal voltage rises) and microphonic effects, which will shift your notch frequency under load and limit your maximum attenuation to 20 dB or less. Always use C0G/NP0 ceramics or polypropylene film capacitors for the timing elements to maintain a stable, deep null.
Reference Table: Common Notch Frequencies and Filter Topologies
Different interference sources require different circuit topologies. The table below maps common real-world noise problems to their target notch frequencies, the best filter topology for the job, and the critical component constraints you must respect.
| Application | Target Notch Frequency | Typical Topology | Required Q-Factor | Key Component Constraint |
|---|---|---|---|---|
| Audio Mains Hum (US) | 60 Hz | Active Twin-T / Fliege | High (>20) | C0G/NP0 Caps, 1% Metal Film Resistors |
| Audio Mains Hum (EU/UK) | 50 Hz | Multiple Feedback (MFB) | Medium (10-15) | Low-leakage Film Capacitors |
| Switching Regulator Ripple | 100 kHz - 2 MHz | LC Trap / Passive Notch | Low (1-3) | High-Q Inductors, Low-ESR Ceramic Caps |
| RF Interference (e.g., FM) | 88 - 108 MHz | SAW Filter / Cavity | Very High (>50) | Shielded Enclosures, 50-Ω Impedance Matching |
| Sensor Excitation Rejection | 1 kHz - 10 kHz | State-Variable | Adjustable | Precision Op-Amps with high GBWP (>10 MHz) |
Where You Meet Notch Frequency in Practice
You will rarely build a notch filter just for the academic exercise; they are almost always deployed to solve a specific noise-floor problem that software or simple RC filters cannot fix.
Audio Signal Chains and Mains Hum
In analog audio, 60 Hz (or 50 Hz) ground loop hum is the most common adversary. While digital FIR/IIR filters can handle this in the DSP domain, analog-to-digital converters (ADCs) will clip if the 60 Hz hum is too large before digitization. Placing an active notch filter right before the ADC preserves the dynamic range. According to Electronics Tutorials, buffering a passive Twin-T with a high-impedance op-amp like the OPA1612 prevents the subsequent stage from loading the network and destroying the Q-factor.
Load Cells and Strain Gauges
When amplifying the millivolt signals from a Wheatstone bridge load cell, the excitation voltage or nearby AC machinery can induce narrowband noise. If your DC measurement is being corrupted by a 60 Hz pickup, a notch filter removes the AC component without introducing the group delay that a low-pass filter would add, ensuring your PID control loops remain stable.
The Fliege Filter Upgrade
While the Twin-T is famous, it is a nightmare to tune because the notch frequency and the bandwidth (Q-factor) are mathematically coupled. If you need an adjustable notch, look into the Fliege filter topology. In a Fliege design, the notch frequency is set by one pair of matched resistors, while the Q-factor is set by a completely independent resistor. This allows you to use a potentiometer to dial in the exact notch frequency on the bench without accidentally widening the rejection band and eating into your desired signal. For deep-dive analog design theory, All About Circuits provides excellent breakdowns of these advanced active topologies.
Notch Frequency vs. Cutoff and Resonant Frequencies
A common mistake among hobbyists and junior engineers is confusing the notch frequency with other filter corner frequencies. Here is how they differ in behavior and application.
| Parameter | Definition | Amplitude Behavior | Primary Use Case |
|---|---|---|---|
| Notch Frequency ($f_n$) | The center point of maximum attenuation in a band-stop filter. | Deep dip (e.g., -40 dB to -60 dB). | Surgical removal of a single interfering tone (hum, ripple). |
| Cutoff Frequency ($f_c$) | The -3 dB point where a low-pass or high-pass filter begins to roll off. | Gradual slope (e.g., -20 dB/decade). | Broadband bandwidth limiting and anti-aliasing. |
| Resonant Frequency ($f_r$) | The center point of maximum gain in a band-pass filter or LC tank. | Sharp peak (gain > 0 dB). | Radio tuning, oscillator feedback networks, wireless power. |
Think of cutoff frequency as a hill you walk down, resonant frequency as a mountain peak you climb, and notch frequency as a narrow trench you dig to bury a specific problem.
Frequently Asked Questions
Can an analog notch filter completely eliminate a frequency (infinite attenuation)?
No. In the real world, component tolerances limit your maximum attenuation. Even with 1% resistors and 2% C0G capacitors, a passive Twin-T will typically bottom out around -35 dB to -45 dB of attenuation. To achieve -60 dB or better, you must use an active topology with a summing amplifier to inject a phase-inverted version of the noise back into the signal path, or use digitally trimmed potentiometers.
Why does my active notch filter circuit oscillate when I power it on?
This usually happens when you design for a very high Q-factor (a very narrow, deep notch) using an op-amp with an insufficient Gain Bandwidth Product (GBWP). The op-amp's internal phase shift at high frequencies compromises the feedback loop's phase margin. As a rule of thumb, your op-amp's GBWP should be at least 50 to 100 times higher than the notch frequency to maintain a stable, non-oscillating null.
Does a notch filter affect the phase of the frequencies that pass through it?
Yes. While the amplitude of the frequencies far outside the notch band remains untouched, the phase response still shifts. A standard Twin-T notch filter introduces a phase shift that approaches 180 degrees across the transition bands. If your application is highly sensitive to phase alignment (like stereo audio imaging or multi-sensor array processing), you must account for this group delay or use a linear-phase digital filter instead.
