This guide covers the four simultaneous demands that define IoT PCB requirements (POINT 01), how to select the right board type (POINT 02), RF, antenna, and ground design rules (POINT 03), power noise and test point design for volume manufacturing (POINT 04), and a volume-oriented procurement strategy (POINT 05).
General-purpose and industrial PCBs are typically optimised for one or two primary requirements. IoT hardware, by contrast, must satisfy four constraints simultaneously — and the tension between them is where most design and procurement problems originate.
📏
Miniaturisation
Physical footprint is often non-negotiable
Wearables, sensor nodes, and embedded devices are constrained by housing dimensions that cannot change. This drives double-sided assembly, reduced trace/space, finer-pitch components, and in many cases HDI construction. The PCB must do more in less space than almost any other product category.
🔋
Low Power
The PCB affects power consumption
Battery life targets are set at the system level, but the PCB design contributes through ground plane continuity (which affects switching noise), layer stack-up (which affects decoupling effectiveness), and trace routing (which affects parasitic impedance). These are not PCB-only problems, but bad PCB design makes them worse.
📡
RF Integration
Wireless performance depends on PCB layout
IoT devices transmit. The antenna — whether chip, trace, or module — interacts with the ground plane, nearby copper, and the physical environment. RF performance cannot be treated as a post-layout consideration. Antenna clearance, ground plane geometry, and RF trace routing must be designed in from the start.
📊
Volume Cost
Unit economics scale aggressively
IoT devices are frequently produced at 10,000–1,000,000+ units per year. A ¥1 per board cost difference becomes ¥1,000,000–¥1,000,000,000 over a product's life. PCB design decisions (layer count, via type, surface finish, panel density) that seem minor at prototype scale have significant economic consequences at volume.
The right board type for an IoT design depends primarily on the physical envelope, component density, and whether the board must flex or conform. Choosing the wrong type at the design stage creates problems that cannot be corrected without a full redesign.
Small Rigid FR-4
The default choice for most IoT sensor nodes and connected devices. 1–4 layer FR-4 construction with double-sided SMT assembly. Cost-effective at volume, manufacturable at most PCB facilities, compatible with standard pick-and-place and reflow assembly.
Use when: board dimensions above ~20mm × 20mm, component pitch ≥ 0402, no mechanical flex requirement.
HDI (High Density Interconnect)
Required when component density exceeds what standard via-in-pad and trace routing can accommodate. Uses laser-drilled microvias, build-up layers, and finer trace/space specifications. Typically 4–8 layers. Significantly higher cost and longer lead time than standard rigid.
Use when: ultra-compact designs (smartwatches, medical wearables), fine-pitch BGAs, or devices requiring < 20mm × 20mm board area.
Flexible / Rigid-Flex
Required when the board must conform to a non-planar surface, fit within a curved housing, or tolerate repeated mechanical flexing. Polyimide substrate instead of FR-4. Rigid-flex combines rigid component areas with flexible interconnect, eliminating discrete cable connectors between segments.
Use when: wearable body contouring, hinge-mount devices, or designs where eliminating flex connectors improves reliability.
Module-Based Design
Integrating a pre-certified radio module (Wi-Fi, Bluetooth, LTE-M, LoRa) onto your host PCB simplifies RF design and dramatically reduces certification scope. The module handles the radio frequency design; your board handles MCU, power management, sensors, and application logic.
Use when: RF design resources are limited, rapid market entry is required, or the wireless standard changes frequently across product variants.
The most common over-engineering mistake in IoT PCB selection: specifying HDI for a board that could be implemented in standard 4-layer construction. HDI adds 40–100% to bare board cost and extends lead time by 1–2 weeks. Before committing to HDI, confirm with your PCB design tool that 0.1mm trace/space, 0.2mm finished hole, and via-in-pad are genuinely necessary — not just convenient. In most IoT sensor nodes, they are not.
Wireless performance is one of the most common sources of late-stage failure in IoT product development. RF problems discovered after the PCB layout is finalised are expensive to fix — they typically require a board spin. Addressing them during layout review prevents this.
RULE 01Antenna keep-out zone — follow the module manufacturer's reference design exactly
Every chip antenna, trace antenna, and antenna module specifies a keep-out zone in its datasheet. This zone defines the area around the antenna element where copper — including ground planes, power pours, and component pads — must not appear. Violating the keep-out reduces radiated efficiency, distorts the radiation pattern, and detuning the antenna from its resonant frequency. A 10mm keep-out infringement can reduce effective range by 30–50%.
Practical rule: Copy the antenna placement exactly from the module manufacturer's reference schematic and layout — including the board edge position and ground plane cutout. Confirm antenna performance with a pre-compliance conducted and radiated measurement before releasing the layout to production tooling.
RULE 02RF ground plane — uninterrupted, referenced, and isolated from digital noise
RF circuit performance depends directly on ground plane continuity. A solid, uninterrupted ground plane directly beneath RF signal traces provides the controlled impedance path that RF signals require. Slots, voids, or digital routing cuts through the ground plane create impedance discontinuities, radiate noise, and degrade RF receiver sensitivity. The ground plane must terminate cleanly at the antenna feed point — not extend into the antenna keep-out zone.
Practical rule: Route all RF traces directly above the ground plane without layer transitions. Place stitching vias (0.3–0.5mm drill) around the RF section perimeter at intervals ≤ λ/20 to prevent cavity resonance. Use a separate poured region for the RF ground, connected to system ground at a single star point, on mixed-signal boards where noise coupling is a concern.
RULE 0350Ω trace impedance control — specify it in the fabrication note
RF signal traces from the MCU or radio IC to the antenna must be designed for 50Ω characteristic impedance. The trace width required depends on the substrate material, layer stack-up, copper weight, and distance to the reference plane — typically 2.4GHz Wi-Fi/Bluetooth traces on a standard 2-layer 1.6mm FR-4 board require approximately 1.8–2.2mm width for 50Ω. Your PCB layout tool's impedance calculator will give the exact value for your stack-up.
Practical rule: Include a controlled impedance specification in your fabrication note: "Layer 1 trace width X.Xmm = 50Ω ± 10%, referenced to layer 2 ground plane." Confirm the manufacturer can verify impedance with a TDR (Time Domain Reflectometer) measurement and provide a test coupon report.
⚠ Crystal oscillator placement affects RF performance: On IoT boards with integrated radio, the crystal oscillator (XTAL) for the MCU or radio module is a common source of unwanted harmonic radiation. Keep the crystal, its load capacitors, and connecting traces as close to the IC as possible and away from RF traces and the antenna area. Shield the crystal with a ground guard ring if board space permits. This is a detail that frequently causes FCC/CE pre-compliance test failures that could have been avoided.
Two design decisions that are often treated as secondary — power decoupling and test point placement — have significant consequences for production quality and manufacturing cost at volume.
⚡
Bypass capacitor placement and power rail decoupling
For battery-powered IoT devices, power supply noise affects the noise floor of ADCs, the sensitivity of RF receivers, and sleep-mode leakage current. Place 100nF decoupling capacitors at each IC's power pin within 2–3mm. Add a 10µF bulk capacitor per power domain. For DCDC converters, minimise the switching loop area (VIN → inductor → VOUT → output capacitor → GND) to reduce radiated switching noise. Keep switching converter components away from RF and analogue sections of the board.
🔌
Power and ground plane topology for low-power designs
Split power domains for sleep mode isolation — ensure high-current loads (radio transmit, motor drivers) cannot inject noise into the supply rail of low-current, precision circuits (RTC, low-power ADC). Use separate LDO rails for sensitive analogue and RF sections where cost allows. On 2-layer boards, use wide power traces and a partial ground pour rather than a full back-copper fill, which can create unintended coupling between sections.
🎯
Test point strategy for small IoT boards
Space is limited on small IoT PCBs, but omitting test points makes production testing impossible and field debugging prohibitively slow. Minimum required test pads: all power rails (VCC, VBAT, each LDO output), system GND, SWD/JTAG programming interface (CLK, DATA, RST, VCC, GND — 5 pads), and UART TX/RX for debug output. Pad size minimum 0.9mm diameter, 1.5mm centre-to-centre for bed-of-nails probe access. Place all test pads on the same board side where possible.
📐
Panelisation and DFM for volume production
Design your panel layout in coordination with your production manufacturer — not independently. Target panel size depends on the manufacturer's SMT line capabilities (typically 100mm × 100mm to 250mm × 330mm). Use V-score (V-groove) or tab-routing for singulation. Maximise boards per panel while leaving adequate breakaway tab strength. Include fiducial marks, tooling holes, and a panel barcoding area. Poor panelisation design is a common cause of automatic singulation failures at volume.
Regulatory Certifications Your Board Manufacturer Cannot Obtain for You
The PCB manufacturer produces the substrate. Product-level regulatory certifications are obtained by the product manufacturer (you) through accredited test laboratories. Understanding what each certification covers — and which is covered by using a pre-certified radio module — is essential before you commit to a test schedule and budget.
🇯🇵 TELEC / 技適 — Japan wireless certification
🇺🇸 FCC Part 15 / FCC ID — US intentional radiators
🇪🇺 CE + RED Directive (2014/53/EU) — European radio devices
KC — South Korea
RCM — Australia / New Zealand
SRRC — China wireless
Module certification and host system scope: Using a pre-certified radio module (e.g. ESP32-based, Nordic nRF series, u-blox) means the radio transmitter itself is pre-approved. Your certification scope is then limited to the host system: EMC emissions (unintentional radiation from the rest of your board), electrical safety (LVD if applicable), and any market-specific requirements. This reduces both certification cost and time-to-market. Confirm module modular approval scope with the manufacturer and your certification laboratory before assuming coverage.
Four Procurement Decisions That Determine Your Volume Economics
🏭Qualify your volume manufacturer at prototype stage — not production stage
The single most costly mistake in IoT procurement is prototyping with a quick-turn service and then switching to a production facility when volumes scale. The switch requires re-qualifying the board at the new manufacturer — re-running controlled impedance tests, validating the new stack-up, and in some cases re-running pre-compliance RF testing. Prototype at your intended production manufacturer from the first meaningful run. The unit cost will be higher for small quantities, but the long-term cost of re-qualification is far higher.
📋Negotiate a volume pricing table covering your expected production range
PCB pricing drops significantly with volume — often 30–50% between prototype pricing and 50,000-unit pricing for the same board. Negotiate a formal price schedule covering 1k, 5k, 10k, 50k, 100k, and 250k+ unit tiers before you commit to the manufacturer, even if your initial orders will be small. This prevents renegotiation at scale and gives you predictable landed cost for financial planning. Annual purchase commitments (guaranteeing a minimum annual volume in exchange for a fixed price) are standard practice at volumes above 50k units per year.
🔄Confirm laminate and material EOL policies before committing to a product design
IoT devices frequently have 3–7 year product lifecycles. FR-4 laminates are generally stable, but specific laminate brands and grades can be discontinued or become unavailable — particularly if the manufacturer switches suppliers. Confirm your chosen manufacturer's EOL policy: how much advance notice is provided, whether they maintain approved alternate laminates that match the stack-up specifications, and whether they will guarantee supply for the duration of your product's planned life. This is a risk that is easy to manage proactively and extremely disruptive to manage reactively.
📦Specify packaging for high-volume shipments to reduce incoming defects
At volume, packaging-related damage — moisture ingress, ESD events, board warping from inadequate support — becomes a measurable reject rate, not an occasional incident. Specify in your purchase order: individual vacuum-sealed moisture barrier bags per panel (IPC/JEDEC J-STD-033 compliant), ESD-rated inner wrapping, foam-padded outer cartons, and panel stack orientation to prevent warping under weight. Packaging specifications should be part of the manufacturing order, not a verbal instruction that gets inconsistently applied across production batches.
Summary
IoT PCB procurement requires balancing four simultaneous constraints — miniaturisation, low power, RF integration, and volume cost — from the first design decisions through to the volume supply agreement. Select the simplest board type that meets your physical and density requirements. Follow antenna keep-out and ground design rules precisely at layout stage. Design test points and panel layout for volume production from the start. And qualify your production manufacturer at prototype stage — not after you have committed to a launch timeline that cannot absorb a re-qualification delay.