Smart Speaker PCB: Design, Cost and Manufacturing Guide

What the Board Has to Do

A smart speaker packs a surprising amount of conflicting electronics into a small space. It has a processor running voice recognition, an array of microphones listening for a wake word, a class D amplifier driving a speaker at several watts, a Wi-Fi and Bluetooth radio, an LED ring for feedback and a power supply that has to run continuously. The microphone array needs an extremely quiet electrical and acoustic environment, while the amplifier switches several amps in the same enclosure and the radio transmits in bursts. Making all of that coexist is the central design problem, and the layout, not the schematic, decides whether it works.

Audio Amplifier Layout

A class D amplifier is efficient but electrically noisy, because its output switches at hundreds of kilohertz with fast edges. The switching loop between the amplifier’s supply pins and its decoupling has to be as small as possible, or it will radiate into the microphones and the radio. The output filter and the speaker traces should be kept short and routed as a pair, away from the sensitive analogue area, and the amplifier ground should return directly to the supply ground rather than through the analogue ground. Where the design allows, physical separation on the board between the amplifier block and the microphone block is worth more than any amount of shielding later.

smart speaker PCB class D amplifier layout

Microphone Array and Analogue Front End

Modern smart speakers use MEMS microphones that output a digital PDM or I2S stream, which removes the traditional analogue signal path but introduces its own requirements. The clock has to be clean, because jitter on the microphone clock directly degrades the audio. The supply to each microphone must be well filtered and free of the switching noise from the amplifier and the power supply, and the traces from the processor to the microphones should be short and matched so that the array stays phase aligned. Mechanically, each microphone needs a sealed acoustic port with a gasket, because any leak between the microphone and the outside world, or any internal path to the speaker, creates echo and degrades the wake word performance. The acoustic design and the board layout are therefore linked: the microphone positions and the port geometry are set by the mechanical design, and the board has to match.

Wireless Coexistence

Wi-Fi and Bluetooth share the same 2.4 GHz band and often the same antenna, so the layout has to respect a generous keep-out around the antenna area with no copper, no traces and no components underneath. The antenna feed should be a controlled impedance line with a solid ground reference, and the ground plane under the radio should be continuous, with via stitching around the radio block. Coexistence between the radio and the class D amplifier is also a layout problem: the amplifier’s switching harmonics can land in the receive band, so keeping the amplifier output and its filter away from the antenna and the radio input is essential. In a small enclosure, that may mean placing the amplifier and the antenna on opposite ends of the board.

smart speaker PCB wireless keep out

Thermal and Power

A smart speaker is always on, so standby power matters as much as peak power, and the power supply design has to keep quiescent consumption low while still delivering the amplifier’s peak current. The amplifier and the processor are the heat sources, and in a sealed enclosure with no fan the heat leaves through the copper and the mounting points. That means a thermal via array under the processor, generous copper on the internal planes and a mechanical path to the enclosure or a small heatsink. Thermal management also affects audio quality: an amplifier that thermally throttles will reduce output, and a processor that throttles will slow the voice response.

Cost Structure

The board itself is rarely the largest cost in a smart speaker. The processor or SoC, the microphone array and the Wi-Fi module dominate the bill of materials, followed by the amplifier and the power supply components. The board is typically four to six layers, with the layer count driven by the need for a solid ground plane under the radio and the processor rather than by component density. Assembly cost is set by the fine pitch processor, the module and the connector count. What distinguishes smart speaker manufacturing from other consumer products is the test: each unit should be tested acoustically, with the microphones verified and the speaker response measured, and that test station is a significant fixed investment.

A smart speaker board mixes a noisy power stage, a sensitive microphone array and a radio, so the layout has to be planned with the mechanical and acoustic design from the start. Review how PCB manufacturing builds controlled impedance and mixed signal boards, apply the audio and RF rules in your PCB design and layout, and check the design and manufacturing considerations before release. A prototype PCB assembly run with acoustic and radio testing proves the design before tooling is committed.

FAQ

How many layers does a smart speaker board need? Four to six is typical, with the extra layers providing ground planes for the radio and the processor.

Why is the amplifier layout so critical? Because its fast switching edges radiate, and the microphones and the radio are both sensitive to that noise.

Do MEMS microphones eliminate analogue design? They remove the analogue signal path but still need a clean clock, a filtered supply and a sealed acoustic port.

What is the biggest cost in a smart speaker? The processor, the microphone array and the wireless module, not the bare board.

Conclusion

A smart speaker PCB is a coexistence problem: a switching amplifier, a precision microphone array and a radio on one board inside a sealed box. Keep the amplifier loop short and physically separated, give the microphones a clean clock and supply with sealed acoustic ports, respect a full keep-out around the antenna, and design the thermal path before the enclosure is fixed. Done that way in 2026, the product hears the wake word and plays back cleanly at the same time.

Leave A Comment