Look inside any modern, ultra-compact audio device, and you are almost guaranteed to find a Class D amplifier. These highly efficient power stages have revolutionized audio engineering, enabling high-output sound systems to fit into smartphones, active monitors, and portable headphone DACs. However, the efficiency of Class D comes with a major caveat: the output stage does not output a continuous analog waveform. Instead, it outputs a high-voltage, high-frequency square wave. To reconstruct the music we actually hear, the amplifier must rely on an output low-pass filter to demodulate the signal. Whether you are shopping for the latest audio gear on our Homepage or studying amp designs, understanding this filtration process is crucial to appreciating modern audio performance.
The Nature of Class D Switching
Traditional Class A or Class AB amplifiers operate in the linear region, meaning their output transistors act as variable resistors. This results in smooth waveforms but high heat dissipation, as a significant portion of the power is wasted as thermal energy. In contrast, Class D amplifiers operate as switching amplifiers. The output transistors (usually MOSFETs) are driven fully on or fully off, acting like rapid electronic switches.
To convert an analog audio signal into a binary switching signal, Class D amplifiers use Pulse Width Modulation (PWM). A comparator circuit compares the input audio signal with a high-frequency triangular carrier wave. The output of the comparator is a series of square pulses where the width (duty cycle) corresponds to the instantaneous amplitude of the audio signal. The switching frequency of these pulses is typically between 250 kHz and 1.2 MHz—far above the 20 kHz limit of human hearing.
However, if this raw high-frequency square wave were fed directly into a loudspeaker or headphone driver, the results would be disastrous. The voice coils would act as heaters for the high-frequency switching energy, leading to thermal overload, massive electromagnetic interference (EMI) radiating from the speaker cables, and severe distortion. The amplifier needs a mechanism to extract the original audio waveform from the high-frequency pulses. This is the job of the low-pass LC filter.
Demodulating the Signal: The Low-Pass LC Filter
The demodulation of the PWM signal is accomplished using a second-order passive low-pass filter, which consists of an inductor (L) and a capacitor (C) for each channel. This configuration is widely detailed in our audio engineering blog. Unlike active filters that require power, a passive LC filter stores and releases energy to smooth out the switching waveform without consuming significant power themselves.
- The Inductor (L): Placed in series with the speaker load, the inductor acts as a low-frequency gateway. The inductive reactance increases linearly with frequency ($X_L = 2\pi f L$). As a result, the inductor presents negligible impedance to audio frequencies (20 Hz – 20 kHz), allowing them to pass through freely, while presenting high impedance to the switching frequency (e.g., 400 kHz), blocking the switching noise.
- The Capacitor (C): Placed in parallel with the speaker load, the capacitor acts as a high-frequency shunt. The capacitive reactance decreases as frequency rises ($X_C = \frac{1}{2\pi f C}$). For the audio band, the capacitor presents high impedance, forcing the audio current through the speaker. For the high-frequency switching noise, the capacitor presents a low-resistance path, shunting the switching carrier signal directly to ground.
Together, the inductor and capacitor form a resonant tank that stores energy during the “on” phase of the square wave and releases it during the “off” phase. This continuous integration of energy averages out the pulses, effectively reconstructing the original analog wave. The table below outlines typical component configurations for various speaker impedances and switching frequencies.
| Target Load (Ω) | Switching Freq (kHz) | Inductance L (µH) | Capacitance C (µF) | Cutoff Freq fc (kHz) | Damping Characteristic |
|---|---|---|---|---|---|
| 4 | 400 | 10 | 1.0 | 50.3 | Butterworth (Optimized) |
| 8 | 400 | 22 | 0.47 | 49.4 | Critically Damped |
| 4 | 600 | 6.8 | 0.68 | 74.0 | Overdamped (Smooth roll-off) |
| 8 | 600 | 15 | 0.33 | 71.6 | Butterworth (Flat response) |
Designing for the Cutoff Frequency and Load Impedance
Designing a Class D output filter involves a delicate balance of electrical formulas. The nominal cutoff frequency ($f_c$) of a second-order LC filter is calculated using the following formula:
$$f_c = \frac{1}{2\pi \sqrt{L \cdot C}}$$
Typically, engineers target a cutoff frequency between 30 kHz and 50 kHz. This ensures the filter passes the 20 kHz audio band with minimal phase shift and attenuation, while providing significant attenuation (often 30 dB or more) at the switching carrier frequency. However, the cutoff frequency is only part of the story. The behavior of the filter is highly dependent on the Quality Factor ($Q$), which is determined by the load impedance ($R$):
$$Q = R \sqrt{\frac{C}{L}}$$
Because the speaker or headphone driver is the resistor $R$ in this circuit, the damping of the filter is directly tied to the connected load. This load-dependency is a primary challenge in Class D amplifier design, as we can visualize in the frequency response graph below.
Filter Frequency Response Under Varying Loads
As illustrated in the chart, if an amplifier is connected to a load with too low resistance (like a 4 ohm load on a filter designed for 8 ohms), the Q factor drops, resulting in an overdamped response. The highs roll off early, causing a loss of treble detail. Conversely, if the load has too high resistance or is run open-circuit (no load), the Q factor skyrockets. The filter becomes underdamped, causing a massive resonance peak near the cutoff frequency. This peak can boost high-frequency noise, create harshness in the upper treble, and even cause amplifier instability.
The Challenge of Real-World Headphone Loading
In the real world, headphones and speakers do not behave like simple resistors. Their electrical impedance changes dynamically depending on the frequency. For example, many multi-driver balanced armature in-ear monitors (IEMs) have complex crossover networks that cause their impedance to swing wildly from 8 ohms to over 100 ohms across the audible spectrum. Because the passive LC filter sits directly before the output terminals, these impedance swings directly alter the filter’s frequency response. This is why some older Class D headphone amplifiers sounded bright and thin with certain headphones, but dark and muddy with others. This challenge is highly relevant for high-performance audiophile headphones where high-frequency noise can degrade the soundstage.
To overcome this limitation, modern high-fidelity Class D amplifiers utilize Post-Filter Feedback (PFFB). PFFB loops take the analog feedback signal directly from the output of the LC filter, rather than before it. By comparing the filtered analog output to the original input signal, the amplifier’s internal feedback loop can dynamically compensate for any impedance-induced frequency deviations. PFFB results in a flat frequency response regardless of the load, while also dramatically lowering the amplifier’s output impedance and total harmonic distortion (THD).
Critical Component Selection: Toroids and Film Caps
Because the output filter handles the full power of the amplifier, the physical characteristics of the inductors and capacitors are crucial. Engineers must carefully select components to prevent them from introducing distortion of their own.
- Inductor Requirements: The inductors must have extremely low DC resistance (DCR) to prevent power loss and heat generation. More importantly, they must have a high saturation current rating. If the current through the inductor exceeds its saturation limit, the core material loses its magnetic properties, causing the inductance to drop sharply. This causes immediate clipping and high-frequency distortion. Metal alloy powder cores and toroidal inductors are widely preferred because of their soft saturation characteristics and low electromagnetic radiation.
- Capacitor Requirements: The capacitors must handle significant high-frequency currents without overheating. Standard multi-layer ceramic capacitors (MLCCs) are generally avoided in high-performance Class D output filters because of their voltage coefficient—their capacitance changes dynamically with the voltage across them, introducing significant non-linear harmonic distortion. Instead, engineers specify metalized polyester or polypropylene film capacitors, which offer excellent linearity, low dielectric loss, and outstanding stability over temperature and voltage.
Conclusion
The Class D output filter is a masterclass in electrical efficiency. Rather than burning off excess switching energy as heat, the inductor and capacitor work together to smoothly integrate the high-frequency pulses back into a clean analog waveform. From mobile devices to high-end home theater systems, the LC filter is what makes high-efficiency audio playback possible without sacrificing fidelity. By selecting high-quality magnetic components, managing load-dependent Q factors, and implementing post-filter feedback, modern designers have turned Class D into a true high-fidelity amplification standard.
Discuss more about this, FAQ, Announcements and Miscellaneous, over on our community.
