Transfer Function of a Notch Filter: Math, Circuits, and IC Selection

The transfer function of a notch filter is a mathematical equation, typically expressed in the Laplace domain as H(s), that defines the exact ratio of output voltage to input voltage across all frequencies, specifically characterized by a sharp drop (zero) at a target center frequency. In a physical circuit, this function dictates the attenuation depth (how much of the signal is killed), the bandwidth (how wide the rejected 'notch' is), and the phase shift introduced to the surrounding frequencies. Hobbyists and students commonly confuse a notch filter's sharp, narrow band-stop behavior with a broad low-pass filter, or they mistakenly assume the ideal mathematical model perfectly predicts the real-world response, ignoring the hard limits imposed by op-amp gain-bandwidth product (GBW) and component tolerances.

The Core Math: Deriving the Transfer Function of a Notch Filter

To design a filter on the bench, you first need to understand the s-domain model. For a standard active notch (band-stop) filter, the transfer function H(s) is expressed as:

Here is what those variables actually mean when you are selecting components:

• s: The complex frequency variable (jω). When you want to see the filter's behavior at a specific AC frequency, you substitute s with j2πf.
• ω₀: The center (notch) frequency in radians per second. This is where the numerator becomes zero, theoretically driving the output to absolute zero volts.
• Q (Quality Factor): This defines the sharpness of the notch. A high Q (e.g., Q > 10) creates a very narrow, surgical cut. A low Q creates a wide, shallow dip. In the denominator, the term (ω₀/Q) represents the damping; less damping means a sharper notch.

When you evaluate this function at the center frequency (where s = jω₀), the numerator becomes (-ω₀² + ω₀²) = 0. This mathematical zero is the 'notch'. However, as we will see in the real-world section, physical components prevent this from ever being a perfect zero.

Worked Numeric Example: Designing a 60 Hz Hum Rejector

Let's design a discrete Twin-T active notch filter to eliminate 60 Hz AC mains hum from a sensitive audio preamplifier. We need to translate the transfer function into physical resistors and capacitors.

For a standard Twin-T network, the center frequency formula is f₀ = 1 / (2πRC). Let's target exactly 60 Hz.

1. Pick the Capacitor (C): Capacitors have fewer standard values than resistors, so we start here. Let's choose a high-quality film capacitor (avoid ceramics for audio due to piezoelectric noise and voltage coefficient). We pick 100 nF (0.1 μF).
2. Calculate the Ideal Resistor (R): R = 1 / (2 × π × 60 × 100 × 10^-9) = 26,525 Ω.
3. Select the Real Resistor: You cannot buy a 26,525 Ω resistor off the shelf. Looking at the E96 (1% tolerance) series, the closest standard value is 26.7 kΩ.
4. Recalculate Actual f₀: Plugging 26.7 kΩ back into the formula yields an actual center frequency of 59.61 Hz.

At 1 kHz (well outside the notch), the transfer function magnitude approaches 1 (0 dB gain). But at our 59.61 Hz target, the ideal math says the output is 0V (-∞ dB). In a real SPICE simulation using 1% resistors and 5% film capacitors, the mismatch prevents perfect cancellation, yielding a real-world attenuation of roughly -40 dB instead of infinite rejection. That is still enough to reduce a 100 mV hum spike down to an inaudible 1 mV.

Where You Meet This in Practice

You will encounter the need to manipulate a notch filter transfer function in several specific hardware domains:

• Biopotential Amplifiers (ECG/EEG): The human body acts as an antenna for 50/60 Hz mains noise. An active notch filter is placed right before the ADC to strip out this interference without destroying the low-frequency biological signals (which can be as low as 0.5 Hz).
• Audio Engineering and PA Systems: Parametric EQs use state-variable notch filters to hunt down and eliminate acoustic feedback (microphone squeal) at specific resonant room frequencies, typically between 1 kHz and 4 kHz.
• Motor Drive Signal Conditioning: When reading current shunt sensors on a BLDC motor, the PWM switching frequency (often 16 kHz to 20 kHz) creates massive noise. A high-Q notch filter tuned to the PWM frequency cleans the feedback signal for the microcontroller's PID loop.

Ideal vs. Real-World: What the Transfer Function Doesn't Tell You

The Laplace equation assumes perfect, infinite-gain operational amplifiers and flawless passive components. On the workbench, three parasitic realities will degrade your notch depth:

Component Tolerances and Drift: As shown in the 60 Hz example, a 1% mismatch in the Twin-T resistors limits your maximum attenuation to about -40 dB. If you use 5% carbon film resistors, your notch depth might shrink to -20 dB. Furthermore, temperature coefficients (tempco) mean a 50 ppm/°C resistor will shift your ω₀ as the board heats up, causing the notch to 'miss' the target frequency.

Dielectric Absorption: Using cheap Y5V or X7R ceramic capacitors introduces dielectric absorption (memory effect), which smears the phase response and rounds out the sharp bottom of the notch. Always use C0G/NP0 ceramics or polypropylene film capacitors for precision filter networks.

Decision Path: Choosing Your Notch Filter Topology and IC

Building a discrete notch filter with 4 op-amps and 12 matched resistors is a great learning exercise, but a poor engineering choice for a production or reliable bench tool. Use this decision matrix to select the right approach.

The Default Recommendation: For 90% of DIY, bench, and sensor-conditioning projects operating between 10 Hz and 100 kHz, the Texas Instruments UAF42 is the definitive pick. By integrating the state-variable topology and laser-trimming the internal components to 0.1% tolerance, it guarantees the deep notch depth that the ideal transfer function promises, saving you hours of hand-matching discrete 1% resistors.

Frequently Asked Questions