A twin-T band stop filter is a specialized notch circuit using two interleaved T-networks of resistors and capacitors to severely attenuate one specific target frequency while passing all others. In a real signal chain, it surgically removes a single interfering tone—like 50Hz or 60Hz mains hum—without rolling off the high or low frequencies of your desired signal the way a low-pass or high-pass filter would. Designers commonly confuse it with standard LC tank notch filters, active biquad state-variable filters, or simple band-pass filters (frequently mixing up the 'stop' and 'pass' terminology). When you need to kill one exact frequency and leave the rest of the spectrum untouched, the twin-T topology is the standard approach.
The Core Mechanics and a 60Hz Numeric Example
The twin-T network consists of two T-shaped RC networks connected in parallel. One network acts as a low-pass filter (two resistors and a shunt capacitor), and the other acts as a high-pass filter (two capacitors and a shunt resistor). At the target notch frequency ($f_c$), the signals passing through both networks arrive at the summing junction exactly 180 degrees out of phase, causing destructive interference and creating a deep null.
The center frequency is determined by the formula:
$f_c = \frac{1}{2\pi RC}$
Let's design a 60.0 Hz notch filter to eliminate AC mains hum from an audio preamp. We will use high-quality WIMA MKS2 film capacitors for the main branch: let $C = 100nF$ (0.1µF).
- Calculate R: $R = \frac{1}{2\pi \times 60 \times 100 \times 10^{-9}} = 26,525\Omega$.
- Practical R selection: Use a 24kΩ 1% metal film resistor in series with a 5kΩ cermet trimpot. This allows you to tune exactly to 26.52kΩ on the bench.
- Calculate the parallel branch (R/2): $26,525 / 2 = 13,262\Omega$. Use a 12kΩ 1% fixed resistor in series with a 2kΩ trimpot.
- Calculate the parallel branch (2C): $2 \times 100nF = 200nF$. Simply place two 100nF film capacitors in parallel.
If you build this passive network and sweep it with a function generator and an oscilloscope, you will see a notch at 60Hz. However, you will also notice a major limitation: the notch is wide, and the null depth is shallow (typically only -20dB to -30dB). This brings us to the Q-factor problem.
Where You Meet This in Practice
You will encounter the twin T band stop filter in three primary domains where single-frequency interference ruins signal integrity:
- Audio Engineering: Removing 50Hz (EU/UK) or 60Hz (US) mains hum from high-gain microphone preamps, guitar pedalboards, and analog synthesizers without dulling the high-frequency 'air' of the audio signal.
- Biomedical Sensors: ECG and EEG amplifiers operate in the microvolt range. The human body acts as an antenna for mains electric fields. A 50/60Hz twin-T notch is mandatory to prevent the AC hum from saturating the ADC.
- Servo Control Loops: In mechanical systems, structural resonances can cause a servo loop to oscillate. A twin-T notch tuned to the exact mechanical resonance frequency (e.g., 145Hz) prevents the controller from amplifying the vibration.
The Q-Factor Problem: Passive vs. Active Topologies
A purely passive twin-T network has a fixed Quality factor (Q) of exactly 0.25. This means the -3dB bandwidth of the notch is four times the center frequency. A 60Hz passive notch will heavily attenuate everything from roughly 15Hz to 240Hz, which is unacceptable for audio or broadband sensor applications.
To fix this, we use active bootstrapping. By buffering the output of the twin-T network with an op-amp voltage follower and feeding a portion of that output back to the common node of the T-networks (the junction where the $R/2$ and $2C$ components meet), we can artificially increase the Q. The feedback effectively 'hides' the loading effects of the network, sharpening the notch.
The Q of an active bootstrapped twin-T is defined by the feedback gain ($K$):
$Q = \frac{0.5}{1 - K}$
If you set the feedback gain to $K = 0.95$ using a precision voltage divider, the Q jumps to 10. The notch becomes incredibly sharp, yielding a null depth of -50dB or better, while leaving frequencies just a few hertz away completely untouched. For a deep dive into active filter topologies, the TI WEBENCH Filter Designer is an excellent tool for simulating these exact Q values before you breadboard them.
Decision Tree: Selecting Your Notch Filter Topology
Do not default to a passive twin-T just because it uses fewer parts. Use this decision matrix to select the right architecture for your specific constraints.
| Application Scenario | Required Topology | Concrete Component Pick |
|---|---|---|
| Fixed 50/60Hz audio hum removal; high fidelity required | Active Bootstrapped Twin-T | OPA1612 (Ultra-low noise, low THD audio op-amp) |
| Biomedical (ECG/EEG); DC-coupled, ultra-low offset drift | Active Bootstrapped Twin-T (Zero-Drift) | OPA2188 (Chopper-stabilized, eliminates 1/f noise) |
| Variable frequency notch (e.g., tunable feedback suppression) | Switched-Capacitor Filter | LTC1062 (Tune the notch via an external clock oscillator) |
| Ultra-low power IoT sensor; wide notch acceptable | Passive Twin-T | TLC27L2 (Low power CMOS, accept Q=0.25 limitation) |
Component Tolerances and Real-World Null Depth
The theoretical null depth of a twin T band stop filter is infinite. In reality, your null depth is entirely dictated by component mismatch. If your $R$ and $R/2$ resistors are off by even 1%, the destructive interference is incomplete, and your -60dB theoretical null degrades to -30dB.
For production environments where trimpots are too expensive, you must specify 0.1% tolerance resistors and 1% tolerance C0G capacitors. Alternatively, implement a digital potentiometer (like the MCP4151) in the feedback path to allow a microcontroller to auto-calibrate the Q and null depth at startup.
For further reading on the mathematical derivation of the twin-T transfer function and component sensitivity analysis, the Electronics Tutorials notch filter guide provides a solid foundational breakdown of the s-domain equations.
Troubleshooting and Frequently Asked Questions
Why is my 60Hz notch only attenuating by 15dB?
Your component ratios are mismatched. Measure the actual resistance of your $R$ and $R/2$ branches with a 4.5-digit multimeter. The ratio must be exactly 2:1. If you are using an active topology, check that your op-amp isn't clipping; a 60Hz hum spike might be driving the op-amp into the rails, destroying the feedback loop's ability to maintain the null.
Can I use a twin-T to block DC offset?
No. A twin-T is strictly an AC band-stop topology. To block DC, use a simple AC-coupling high-pass filter (a series capacitor and a shunt resistor) sized for your lowest frequency of interest.
Does the op-amp bandwidth affect the notch?
Yes. The op-amp's Gain-Bandwidth Product (GBWP) must be at least 100 times higher than your notch frequency. For a 60Hz notch, almost any op-amp works. But if you are designing a 20kHz ultrasonic notch, you need an op-amp with a GBWP of at least 5MHz to prevent phase shift from ruining the high-frequency null depth.
For 90% of bench, audio, and sensor applications targeting fixed mains hum, the active bootstrapped twin-T using an OPA1612, 1% metal film resistors, and WIMA film capacitors is the definitive default choice. It provides the sharpest null, the lowest noise floor, and requires no complex digital clocking.
