A class d audio amplifier circuit achieves over 90% efficiency by using pulse-width modulation (PWM) to switch output transistors fully on or off, rather than operating them in their linear region. If you need 50W of audio output from a 24V supply without bolting a massive aluminum heatsink to your enclosure, Class D is your only practical choice. This guide breaks down the half-bridge topology, selects real-world component values for a 100W design, and shows you how to safely bench-test the switching node without blowing up your oscilloscope probe.
Why Choose a Class D Audio Amplifier Circuit?
Before we lay out copper, we need to justify the complexity. Class AB amplifiers are simpler and have inherently lower electromagnetic interference (EMI), but they waste massive amounts of power as heat. Here is how a modern Class D topology stacks up against the legacy Class AB standard for a 50W per channel audio design.
| Parameter | Class AB (e.g., LM3886) | Class D (e.g., TPA3116D2) |
|---|---|---|
| Typical Efficiency (at 50W) | 50% - 60% | 90% - 93% |
| Heat Dissipation (Heatsink) | Requires ~3°C/W massive finned sink | Requires minimal or no heatsink |
| Quiescent Current | ~50mA - 100mA | ~10mA - 20mA |
| Output Filter Required? | No | Yes (LC Low-Pass) |
| EMI / RF Noise | Negligible | High (requires careful PCB layout) |
The Verdict: Choose Class AB for ultra-low-noise preamp stages or headphone amplifiers where milliwatts matter. Choose a class d audio amplifier circuit for main speakers, subwoofers, battery-powered boomboxes, or any application where thermal management is a constraint.
Topology Breakdown: Nodes, Switches, and Filters
Most modern DIY and commercial Class D designs use a Bridge-Tied Load (BTL) topology, which eliminates the need for a bulky DC-blocking output capacitor. Let's trace the signal path through the critical nodes of a single BTL channel:
- IN+ / IN- (Differential Input Node): Where the analog audio signal enters. Usually AC-coupled via a 1µF film capacitor to block upstream DC offset.
- PWM_GEN (Internal Logic Node): The analog signal is compared against a high-frequency triangle wave (typically 300kHz to 600kHz) to generate a PWM duty cycle.
- HO / LO (Gate Drive Nodes): High-side and low-side gate signals. These drive the gates of the N-channel MOSFETs. A critical 'dead-time' (usually 10ns-20ns) is inserted here to prevent shoot-through.
- SW+ / SW- (Switch Nodes): The junction between the high and low MOSFETs. This node violently swings between GND and PVCC (e.g., 0V to 24V) at 400kHz. This is where your PCB layout makes or breaks the design.
- LC_FILTER (Filter Node): The PWM square wave passes through an inductor and capacitor to strip the high-frequency carrier, leaving only the amplified audio envelope.
- OUT+ / OUT- (Load Nodes): The cleaned analog audio connects directly to the speaker voice coil.
Design Walkthrough: Picking Real Component Values
Let's design a 2x50W stereo channel using the ubiquitous Texas Instruments TPA3116D2. We are targeting a 24V DC supply and a 4Ω speaker load.
1. Setting the Switching Frequency
The switching frequency ($f_{sw}$) dictates your LC filter size and EMI profile. We want 400kHz to stay above the audible band but below the threshold where MOSFET switching losses become severe. On the TPA3116D2, the FREQ pin sets this via a resistor to GND.
- Component: $R_{FREQ} = 20\text{k}\Omega$ (1% tolerance, 0402 SMD). This locks the internal oscillator to exactly 400kHz.
2. Sizing the LC Low-Pass Filter
The filter must pass 20kHz audio while aggressively attenuating the 400kHz carrier. We target a cutoff frequency ($f_c$) of around 60kHz.
- Inductor ($L$): $10\mu\text{H}$. You must use a shielded ferrite core rated for at least 6A saturation current. If the inductor saturates during bass transients, inductance drops to near zero, sending raw 24V 400kHz PWM straight into your tweeter, melting the voice coil.
- Capacitor ($C$): $680\text{nF}$ (0.68µF). Rated for at least 50V DC. Use a WIMA MKP polypropylene film capacitor or a high-voltage C0G ceramic.
Math Check: $f_c = \frac{1}{2 \pi \sqrt{L \cdot C}} = \frac{1}{2 \pi \sqrt{10\mu\text{H} \cdot 680\text{nF}}} \approx 61\text{kHz}$. Perfect.
Behavior Matrix and Extreme Failure Modes
Understanding how a class d audio amplifier circuit reacts to component drift or catastrophic failure is what separates a working prototype from a reliable product. Here is what happens when elements shift or break.
| Element Changed / Failed | Resulting Circuit Behavior |
|---|---|
| Increase $L$ to $22\mu\text{H}$ | Cutoff frequency drops to ~41kHz. Better EMI suppression, but phase shift at 20kHz increases, potentially causing high-frequency audio harshness. |
| Decrease $C$ to $220\text{nF}$ | Cutoff frequency rises to ~107kHz. More high-frequency carrier bleeds into the speaker, wasting power and increasing tweeter heating. |
| Open Inductor (Wire breaks) | No audio output. The SW node loses its snubbing path and will ring to >60V due to parasitic capacitance, likely breaching the MOSFET $V_{DS}$ rating and destroying the IC. |
| Shorted Output Capacitor | Massive overcurrent. The IC's internal over-current protection (OCP) should trip, but repeated shorts will eventually degrade the internal MOSFETs. |
| No Load (Speaker disconnected) | The LC filter Q-factor spikes. The OUT node will ring violently at the resonant frequency, potentially generating voltage spikes that arc across PCB traces. |
Breadboard Testing: Step-by-Step Verification
Warning: Do not attempt to build a 100W, 400kHz class d audio amplifier circuit on a standard solderless breadboard. The ~1nH per millimeter parasitic track inductance and lack of a ground plane will cause massive voltage ringing, resulting in immediate shoot-through and a shattered IC.
However, you can and should breadboard the logic and low-power filter concepts using a 5V, 3W Class D module like the PAM8403 to verify your LC filter math and measure PWM duty cycles safely.
- Power the Logic Stage: Insert the PAM8403 DIP module into the breadboard. Connect 5V to VCC and GND. Keep the power leads short and place a 100µF electrolytic decoupling capacitor directly across the module's power rails.
- Inject a Test Tone: Connect a function generator set to a 1kHz sine wave, 1V peak-to-peak, into the Left Audio Input. Tie the generator ground to the breadboard ground.
- Probe the Switch Node: Clip your oscilloscope probe to the module's raw SW+ pin (before the inductor). Set the scope to 10ns/div. You should see a 5V square wave switching at roughly 140kHz. As you increase the audio input amplitude, watch the PWM duty cycle widen from 50% toward 90%.
- Verify the LC Filter: Move the scope probe to the speaker output side of the inductor. The 140kHz square wave should now be completely gone, replaced by a clean, amplified 1kHz sine wave. If you still see 'fuzz' on the peaks of the sine wave, your breadboard parasitics are ruining the filter response—time to move to a copper-clad perfboard.
Frequently Asked Questions
Do I need an output LC filter for a class d audio amplifier circuit?
It depends on your cable length and load. Many modern ICs support 'filterless' operation, relying on the speaker coil's own inductance to average the PWM. This works fine if the speaker is mounted less than 10cm from the IC (like in a TV or soundbar). If you are running speaker wire across a room, the wire acts as an antenna, radiating the 400kHz carrier and failing FCC/CE EMI compliance. For any DIY project with external speaker terminals, always include the LC filter.
Why does my class d audio amplifier circuit interfere with AM radio?
Class D amplifiers generate massive $dV/dt$ (voltage change over time) edges at the switch node. If your PCB layout has a large loop area between the PVCC bypass capacitor, the high-side MOSFET, and the low-side MOSFET, that loop acts as a magnetic dipole antenna. To fix this, place a 1nF to 4.7nF ceramic capacitor (rated for >50V) directly across the SW+ and SW- nodes to act as a high-frequency snubber, and ensure your input signal cables are shielded and twisted.
How do I calculate the LC cutoff frequency for a class d audio amplifier circuit?
Use the standard resonant frequency formula: $f_c = \frac{1}{2 \pi \sqrt{L \cdot C}}$. Your goal is to set $f_c$ at least one octave above the highest audio frequency you want to reproduce (so, >40kHz for 20kHz audio), but at least one decade below your PWM switching frequency to ensure adequate carrier attenuation. For a 400kHz switcher, a 60kHz cutoff is the engineering sweet spot.
