Smart Speaker PCB and PCBA Design Notes
What Is Inside a Smart Speaker
A smart speaker packs an audio system, a radio, a microphone array, and a processor into a small enclosure. The main board carries the system-on-chip that runs the assistant, a codec for analog conversion, a class-D amplifier, the Wi-Fi and Bluetooth module, memory, and the power conversion that turns the adapter voltage into the rails the rest of the design needs.
Around it sit smaller boards: a microphone board at the top of the enclosure where the microphones can hear the room, an LED board for the indicator ring, and often a separate power board to keep the switching conversion away from the analog and radio sections. The finished assembly is a PCBA problem as much as a PCB problem, because most of the failures that reach the customer originate in layout or in the assembly process rather than in the silicon.
Board Architecture
Splitting the electronics across boards is a deliberate choice. The microphone board is small and analogue, and its layout is dominated by keeping the traces short and the ground clean. The LED board is a simple driver board that is physically separate for mechanical reasons. The power board handles the high-current switching and the thermal load, and putting it on its own substrate keeps its noise and heat away from the audio path.
The main board carries everything else, and its layout is a negotiation between the radio, the audio, and the digital sections. Each of those wants a quiet reference and short connections, and the three of them cannot all have the centre of the board.
Audio Layout
The analog signal path is where the sound quality is decided. It runs from the codec output to the amplifier input, and every millimetre of that path is an opportunity to pick up interference. Keeping it short, away from switching regulators, and separated from the digital section is the basic discipline; running it over a continuous ground plane is the other half.
Ground return is the most common source of audible hum. The amplifier’s high-current return must flow in a defined path back to the supply, not across the reference that the input signal uses, which is why the input ground and the output ground are joined at the point where the currents converge rather than allowed to mix. Star arrangements at the amplifier, or careful plane partitioning, both work if the intent is explicit.
The class-D amplifier is itself a switching device, so its output traces are noisy by nature. Keeping them short and confining them to their own area, with the feedback path taken directly from the output terminals rather than from a remote point, is what keeps the switching noise out of the input and off the radio.
Thermal design follows the power level. A small speaker may dissipate a few watts from a package with a thermal pad, and that pad needs vias to a copper area rather than the laminate alone.
Radio Layout
The Wi-Fi and Bluetooth radio shares the board with the audio chain, and it is both a victim and a source. Its antenna needs a clear keep-out region, an impedance-controlled feed line, and a ground plane that extends beneath the module but stops short of the radiating element. Placing the antenna inside a metal grille or beside the power supply will cost range that no firmware change recovers.
On the receive side, the radio has to survive the noise the speaker creates. The switching regulator, the class-D output, and the processor’s memory bus all radiate, and the quietest solution is physical separation and a continuous ground plane between them rather than reliance on filtering alone. The module’s supply needs local decoupling that can deliver the transmit current burst without letting the rail sag into the audio circuitry.
The Defects That Show Up in Products
Manufacturers who are new to this combination tend to produce the same set of problems. A ground design that does not separate audio and digital returns produces audible hum or hiss. A stackup that puts the radio on the wrong side of a noisy layer causes desensitisation, which the user experiences as poor range or a speaker that drops off the network.
Solder defects on the microphone or the codec show up as low or intermittent pickup, which is easy to mistake for a software problem. Insufficient decoupling or a marginal regulator causes resets when the amplifier draws a transient, which looks like a random reboot. Each of these is a design or process issue in the board, and each is cheaper to fix before the design is released than after a thousand units have been built.
Stackup and Materials
Four layers is the practical minimum for a smart speaker, with a signal layer, a ground plane, a power plane, and a second signal layer. Six layers appear where the radio requires controlled impedance or where the audio section needs its own ground region. The ground plane should be solid under the radio, the audio path, and the processor in turn, and it should not be perforated by a grid of via holes any denser than the design requires.
High-Tg FR-4 is the usual material, selected for dimensional stability through assembly. ENIG is the common finish because it presents a flat surface for the ball grid array and quad flat no-lead devices beneath the processor and the module, and because it survives storage without losing solderability. Where the radio layer requires a specific dielectric height, the stackup is agreed with the fabricator so that the impedance can actually be held.
Manufacturing Requirements
The processor and the radio module push the fabrication requirement towards fine line and, in the smallest products, HDI with laser-drilled microvias. Impedance control is needed on the antenna feed and on the high-speed interface between the processor and memory. The board also has to hold a reasonable copper weight for the audio and power section, which means the fabricator has to manage fine features and heavy copper on the same panel.
Assembly and Test
Assembly is a mixed-technology process: fine-pitch surface mount for the processor and passives, a module that may be soldered or socketed, and through-hole connections for the speaker and power connectors. The microphone and the acoustic ports require care so that no residue or contamination enters the opening, and the board is usually cleaned before coating.
Testing after assembly is where the product is actually proven. Automated optical inspection and X-ray cover the joints that cannot be seen. Functional test then checks the audio path with a known signal and measures distortion and output level, verifies the radio at several channels and power levels, and confirms microphone sensitivity. A burn-in period catches the early failures before the product ships, and thermal testing confirms that the amplifier does not throttle in a sealed enclosure.
Compliance work sits alongside: RoHS and REACH for materials, IPC-A-610 for assembly acceptability, EMC testing for emissions and immunity, and the radio certification required by each market. A manufacturer that handles PCB assembly and the functional test that goes with it keeps those constraints inside one process rather than spread across suppliers.
What Drives the Cost
A four-layer smart speaker board is inexpensive in volume, and a six-layer board with controlled impedance is noticeably more, but both are small items next to the assembly cost. The components, the fine-pitch placement, and the test time dominate the finished PCBA price.
Prototype quantities are expensive because the setup is spread over a handful of boards, and the price falls steeply with volume only if the design is stable. Changing a stackup or a component after tooling is expensive in both money and schedule, which is why the radio and audio layout reviews belong before the first panel is ordered. For related reading, see our notes on PCB manufacturing and PCBA testing.
FAQ
Why does a smart speaker need a separate microphone board? The microphones must sit at the top of the enclosure to hear the room, and keeping that board small and analog keeps its traces short and its ground clean.
What causes audible hum in these products? Ground return paths that mix the amplifier’s high-current return with the input reference, which is why the two are joined only where the currents converge.
How is poor Wi-Fi range related to the PCB? Antenna keep-out, feed-line impedance, and the placement of noisy switching circuits near the radio all affect range directly, and none of them can be corrected in firmware.
Which finish is used? ENIG is common because it is flat for the fine-pitch parts and remains solderable through storage.
Is HDI necessary? Only in the smallest products. Most smart speakers use a conventional four- or six-layer board with fine-line capability.
Conclusion
A smart speaker board is an exercise in coexistence: an audio path that must stay clean, a radio that must be heard, and a processor and amplifier that generate noise and heat in a sealed enclosure. Grounding, physical separation, controlled impedance on the radio, and a stackup that supports all three decide whether the product sounds right and stays connected. For related topics, see our notes on PCB design and layout and PCB capabilities.
