A notch filter is a type of electronic filter that removes a very narrow band of frequencies from a signal while leaving everything above and below that band untouched. Think of it as a surgical tool: instead of cutting out a wide range of frequencies like other filters, it targets one specific frequency (or a tiny range around it) and suppresses it, often by 40 to 70 decibels or more. The frequency it removes is called the notch frequency.
How a Notch Filter Works
Most filters work by drawing a line somewhere in the frequency spectrum. A low-pass filter blocks everything above a cutoff point. A high-pass filter blocks everything below it. A notch filter does something different: it passes low frequencies, passes high frequencies, and carves out a deep, narrow dip at one specific point in between. If you looked at a graph of its frequency response, you’d see a smooth line with a sharp downward spike, like a notch cut into wood.
The depth of that notch determines how thoroughly the unwanted frequency is suppressed. A well-tuned notch filter can achieve attenuation levels beyond 60 dB at the target frequency, meaning the signal at that frequency is reduced to less than one-millionth of its original power. Meanwhile, frequencies even a short distance away pass through with little or no change.
Notch Filters vs. Band-Stop Filters
A notch filter is technically a specialized version of a band-stop filter. The difference comes down to how wide the rejected frequency range is. A band-stop (or band-reject) filter blocks a broader swath of frequencies. A notch filter blocks an extremely narrow band, sometimes just a few hertz wide, with steep sides and a deep center. This selectivity is described by a value called Q: the higher the Q, the narrower and deeper the notch. Notch filters are high-Q designs built to reject a single frequency or a very small cluster of frequencies rather than a whole bandwidth.
Common Uses in Audio
One of the most familiar applications is removing electrical hum from audio recordings. In North America, the power grid runs at 60 Hz, and that frequency often bleeds into microphone cables and recording equipment as an audible buzz. In the UK and most other countries, the mains frequency is 50 Hz. A notch filter set to the appropriate frequency cuts out that hum without noticeably affecting the music or voice around it.
The hum rarely stops at one frequency, though. It tends to produce harmonics, meaning you’ll also see noise spikes at multiples of the fundamental (120 Hz, 180 Hz, 300 Hz, and so on). Audio engineers typically apply the notch filter at the fundamental frequency first, then check a spectrum analyzer to find remaining harmonic spikes, and apply additional notches at those points. For higher harmonics, increasing the Q factor (making the notch narrower) helps avoid artifacts in the surrounding audio. A Q between 2 and 10 works well for mains hum removal in most situations.
Notch filters also handle high-pitched interference. If a recording picks up a constant whistle at, say, 19,000 Hz from nearby electronics, a spectrum plot will show a tall, isolated peak at that frequency. Setting a notch filter to that exact frequency removes the whistle while leaving the rest of the audio intact.
Medical and Biomedical Applications
Electrocardiogram (ECG) machines face the same power-line interference problem as audio equipment, but with higher stakes. The electrical signals from your heart are tiny, and a 50 or 60 Hz hum from nearby power lines can easily overwhelm them on the readout. Notch filters are standard in ECG devices to strip out this interference.
The challenge in medical applications is that heart signals themselves contain frequency components close to the power-line frequency. If the notch is too wide, it distorts the very waveforms doctors need to read, particularly the sharp peaks of the heartbeat and the segments between them. Designers have to balance the notch width carefully: narrow enough to avoid distorting the medical signal, but wide enough to actually catch the interference even if the power-line frequency drifts slightly. Some modern devices use adaptive notch filters that automatically track the exact interference frequency and adjust in real time.
Radio and Wireless Communications
In radio systems, notch filters protect receivers from powerful interfering signals on nearby frequencies. Ultra-wideband (UWB) communication systems, for example, operate across a huge frequency range (3.1 to 10.6 GHz) but at very low power levels. Signals from Wi-Fi networks operating in the 5.7 to 5.85 GHz range can be more than 60 dB stronger than the UWB signal, effectively drowning it out. A notch filter tuned to that Wi-Fi band rejects the interference while allowing the rest of the UWB spectrum through cleanly.
These filters can be built as separate physical components placed before the receiver’s amplifier, or integrated directly onto the same chip as the amplifier. On-chip designs save space but require careful engineering to achieve enough rejection depth. One recent design achieved 35.2 dB of interference rejection at 5.8 GHz while preserving the receiver’s sensitivity across the rest of its operating band.
How Notch Filters Are Built
In analog circuits, one classic design is the Twin-T filter. It uses two T-shaped networks of resistors and capacitors connected in parallel. One T is made from resistors across the top and a capacitor forming the leg to ground. The other T reverses this, with capacitors across the top and a resistor to ground. At one specific frequency, determined by the resistor and capacitor values, the two paths cancel each other out and the signal drops to near zero. The notch frequency occurs when the product of resistance, frequency, and capacitance equals exactly one.
In digital systems, notch filters are implemented in software or on dedicated signal-processing chips. The most common approach uses a structure called an IIR (infinite impulse response) filter, which is computationally efficient because it recycles its own previous output values as part of the calculation. Multiple IIR sections can be cascaded together to notch out a fundamental frequency and its harmonics simultaneously. Adaptive versions use algorithms that continuously adjust the notch frequency to track a moving target, which is useful when the interference frequency isn’t perfectly stable.
Tradeoffs and Limitations
Notch filters involve a fundamental tradeoff between selectivity and stability. Making the notch narrower (higher Q) means less distortion to nearby frequencies, which is desirable. But a narrower notch is also less forgiving: if the interference frequency shifts even slightly, it may fall outside the notch and pass through unfiltered.
There’s also the problem of ringing. When a signal changes abruptly, a high-Q notch filter can produce brief oscillations in its output, like a bell that keeps vibrating after being struck. The narrower the notch, the longer this ringing persists. In applications where the signal contains sudden jumps or transients, this can be a real problem, adding artificial oscillations that weren’t in the original signal.
Phase shift is another consideration. A single notch filter shifts the timing relationship between different frequency components of the signal. This shift is most pronounced near the notch frequency and becomes negligible further away. For audio or communications work, this is often acceptable. For precision measurement applications, engineers sometimes run the filter in both directions (forward and backward through the data) to cancel out the phase distortion entirely, though this only works with recorded data, not live signals.