Designing a Class D Amplifier Schematic Step-by-Step Guide

schematic diagram amplifier class d

Start with a synchronous buck converter topology when designing a switching audio stage. Use MOSFETs like the IRFB4110 or IPP075N10N3 for low RDS(on) (<10 mΩ) and fast switching (rise/fall times <20 ns). Pair them with a gate driver such as the IR2110 or UCC27424, configured for dead-time control to prevent shoot-through–target 20–50 ns dead-time for 400 kHz switching.

Implement a two-pole, two-zero compensation network in the feedback loop. Use an op-amp like the OPA2277 with a Type III compensator (C1=1 nF, C2=10 nF, R1=10 kΩ, R2=1 kΩ) to stabilize the modulator at a 50–100 kHz crossover frequency. Ensure the LC output filter (L=10 µH, C=1 µF) has a resonant frequency below half the switching rate to minimize aliasing.

For modulation, use a hysteretic or ΣΔ PWM controller (e.g., TAS5731M or discrete solution with TLV3501). Keep the carrier frequency above 350 kHz to reduce EMI and improve linearity. Add a small-signal pre-filter (R=1 kΩ, C=100 pF) at the input to reject HF noise without degrading phase margin.

Power the gate drivers from a separate, regulated 12V rail to avoid ground bounce. Use ceramic capacitors (X7R, 1 µF) placed within 2 mm of each MOSFET’s source/drain pads for decoupling. For thermal management, mount the FETs on a 2 oz copper PCB with a minimum of 50 mm² pad area per device, calculated for a 3°C/W junction-to-ambient thermal resistance.

Test stability with a load-step response: apply a 1A to 5A step at 1 kHz and verify settling within 10 µs with <2% overshoot. Measure distortion at 1W into 4Ω–target <0.05% THD+N (20 Hz–20 kHz) with a bridged output stage to reject power supply noise. If using a Class BD modulator, add a floating charge pump for the high-side driver to ensure reliable bootstrapping.

Key Design Principles for High-Efficiency Switching Audio Stages

Begin with a synchronous MOSFET configuration for the output stage–half-bridge or full-bridge–depending on power demands. For compact designs under 50W, half-bridge suffices; full-bridge doubles output voltage swing, ideal for 100W+ systems. Select low RDS(on) FETs (sub-50mΩ) like Infineon BSC014NE2LS or Texas Instruments CSD18502Q5B to minimize conduction losses.

Implement a two-phase modulation strategy: fixed-frequency PWM for steady loads, spread-spectrum for EMI-critical applications. Use a 300–500kHz carrier frequency to balance switching losses (proportional to fSW) and filter size. For ultra-low THD designs, add a small hysteresis band (±2% of full scale) to suppress idle-mode distortion.

Critical Component Selection

Component Recommended Part Key Spec Design Impact
Gate driver DRV8601 2A sink/source, 6–16V Minimizes shoot-through with 10ns dead-time
Output filter Murata LQW32CN470M52 47µH, 3.5A, DCR 120mΩ Reduces ripple to
Feedback cap C0G/NP0 100nF 50V, ±5%, X7R Eliminates phase lag in feedback loop
Bootstrap diode BAS16 1A, 100ns reverse recovery Prevents ghost switching at 500kHz+

Route high-current traces (>5A) on 2oz copper with minimum 3mm width; use vias only where unavoidable. Keep the gate driver’s ground separate from the audio ground until the output filter inductor to prevent ground bounce. Terminate gate traces with a 1–5Ω series resistor to dampen ringing from parasitic inductance.

For thermal management, employ a copper pour on the PCB’s backside, sized to handle 1°C/W per 1W dissipation. Thermal vias (12–15 per FET) should have a 0.3mm diameter to ensure efficient heat transfer to inner planes. Avoid exceeding 130°C junction temperature–FET efficiency drops 0.2%/°C above this threshold.

Calibrate the feedback network carefully: a 1% mismatch between the integrator resistors causes 0.5dB gain error. Use 0.1% tolerance resistors and adjust the integrator capacitor to match the PWM carrier frequency–typical ratio is 1nF per 100kHz. Test THD at 1W output: anything above 0.05% indicates parasitic oscillation in the feedback path.

Debugging Common Pitfalls

schematic diagram amplifier class d

If idle current exceeds 50mA, check for false switching–add a 1kΩ pull-down resistor to the PWM input. Audible whine at

Key Components of a Switching Power Stage Layout

Prioritize low-inductance traces for the output stage to minimize ringing and electromagnetic interference. Use wide, parallel copper pours for the H-bridge or half-bridge configuration, keeping switching node connections under 5 mm in length. Select MOSFETs with sub-20 ns rise/fall times and on-resistance below 10 mΩ for high-current applications–pair them with gate drivers capable of sourcing/sinking at least 2 A to prevent shoot-through. Decouple each power transistor with a ceramic capacitor (100 nF–1 µF, X7R dielectric) positioned within 2 mm of the drain/source pins, supplemented by a bulk electrolytic (22–100 µF) nearby.

  • Gate drive circuit: Isolate gate traces from power loops using a dedicated ground plane for the driver IC; route signals with controlled impedance (50–100 Ω) to avoid overshoot. Ferrite beads (1–10 MHz range) on the driver’s supply rails suppress high-frequency noise without degrading response time.
  • PWM generator: Place the comparator or digital controller adjacent to the feedback network, ensuring analog ground separates from switching ground. Use a low-pass filter (cutoff 20–50 kHz) on the modulation input to reject aliasing artifacts.
  • Output filter: Design the LC network with inductor values between 1–10 µH (saturation current 1.5× target load) and film capacitors (1–10 µF) for THD below 0.1%. Keep capacitor leads shorter than 10 mm to reduce ESR-induced distortion.

Thermal vias (0.3–0.5 mm diameter, 1 mm spacing) beneath MOSFET pads enhance heat dissipation when paired with a 2–4 oz copper pour on the PCB backside. For multi-layer boards, dedicate an inner layer to a continuous ground plane, stitching it to the top/bottom planes with vias at 1 cm intervals to prevent loop currents. Validate the layout using a time-domain reflectometer to verify trace impedance mismatches remain under 5%.

Step-by-Step PWM Modulation Block Construction

schematic diagram amplifier class d

Begin with a high-speed comparator IC like the LM311 or TLV3501–these deliver sub-50ns rise times critical for clean edge transitions. Pair the non-inverting input with a 1MHz triangular waveform generator (e.g., a CLC1100 op-amp in a classic integrator configuration), ensuring amplitude spans 0.5V to 4.5V to align with standard 5V logic thresholds. Keep trace lengths under 15mm to minimize stray inductance.

Feed the audio signal into the comparator’s inverting input via a 10kΩ resistor, followed by a 100pF capacitor to ground–this forms a low-pass filter cutting off at ~160kHz to suppress aliasing while preserving 20kHz bandwidth. Offset the audio by 2.5V using two matched 10kΩ resistors to center the modulation range around the triangle wave’s midpoint.

Introduce hysteresis by adding a 5kΩ resistor between the comparator’s output and non-inverting input. This creates a ~50mV deadband, preventing spurious switching from noise without sacrificing THD below 0.1%. Verify stability by checking for consistent duty cycle symmetry at 50% modulation depth with an oscilloscope trigger set to the carrier frequency.

Use a dedicated high-side driver IC (e.g., IRS2092S) to interface the comparator output with the power stage. The driver’s dead-time should be adjusted via external 10kΩ potentiometers targeting 50ns to avoid shoot-through while maximizing efficiency at >90% for 8Ω loads. Bypass VCC with 1μF ceramic capacitors at each IC supply pin–place them within 2mm of the pad to suppress ringing.

For fault protection, route the comparator’s output through a NOR gate (e.g., 74AHC1G02) before the driver. One gate input monitors the over-current comparator (set via a 0.01Ω sense resistor and LM393), pulling the gate low if current exceeds 10A for more than 1μs. The other input connects to a thermal sensor (e.g., MCP9700) mounted on the heatsink, disabling switching if temperature surpasses 85°C.

Implement bootstrap circuitry for the high-side MOSFET using a UF4007 diode and a 1μF capacitor. The diode must handle reverse recovery times

Terminate PWM lines with 22Ω series resistors at the MOSFET gates to dampen reflections and reduce EMI. For layouts, keep the triangle wave generator on a separate ground plane from the power stage to prevent coupling, then stitch the planes together at a single point near the comparator’s ground pin. Use 4-layer boards with 1oz copper for inner layers and 2oz for power traces to handle >20A peak currents.

Validate the modulation block by injecting a 1kHz sine wave at 0dBFS. Measure output with a spectrum analyzer–spurious noise should remain below -60dBc up to 1MHz, and THD+N under 0.2% at full-scale. Adjust the comparator’s input hysteresis if sidebands appear asymmetrical, indicating improper timing synchronization.

Power Supply Specifications and Noise Suppression for Switching Audio Stages

Select a low-ESR bulk capacitor with at least 2,200 µF per 100 W of continuous output power, placed within 20 mm of the half-bridge MOSFETs. A 105 °C rated Sanyo OS-CON or Panasonic SP series outperforms standard electrolytic types by reducing ripple current by 40 % at 250 kHz.

Add a 1 µF X7R ceramic capacitor directly between the DC input pins of the gate driver IC and the nearest ground plane. Thermal cycling stability requires 0603 or larger case size; 0402 parts may crack under 15 A/s switching edges.

Implement a π-filter using a 22 µH ferrite bead (e.g., Murata BLM18PG121SN1) followed by two 100 nF C0G ceramics. The ferrite bead must exhibit 200 Ω at 100 MHz to block both low-frequency supply ripple and high-frequency gate-drive interference without saturating.

Keep the ground return of the pre-regulator and logic circuits separate from the high-current switching return. Use a single star-point connection to the main reservoir capacitor negative terminal; PCB traces between the star-point and individual grounds should not exceed 10 mm in length to prevent 80 mV peak-to-peak common-mode noise from coupling into the feedback loop.

When operating above 100 W, split the DC feed into two parallel 2 oz copper pours, each carrying half the current. Thermal simulations show this arrangement lowers trace temperature rise from 110 °C to 65 °C under 12 V, 15 A conditions, reducing copper loss by 35 %.

Mountable Schottky diodes (ON Semi MBR1045) on the input of each half-bridge leg prevent reverse current during dead-time. Position them adjacent to the switching devices; a 3 mm separation reduces stray inductance to

Bypass the analog section of the PWM modulator with a 10 nF COG capacitor and a 4.7 µF tantalum polymer capacitor (Kemet T530 series) in parallel. This combination yields noise floor improvement from –92 dB to –108 dB at 1 kHz, reference 1 Vrms

For 24 V rail designs, add a 1.2 A poly-fuse on the DC input. The fuse must clear within 5 ms at 2× rated current (2.4 A) to prevent catastrophic failure of the reservoir capacitors under fault conditions; Littelfuse 243 series meets this criterion.