Module or chip-down: making the radio decision

Every connected product faces the same early choice: buy the radio as a pre-certified module, or put the radio chip on your own board. Plan for compliance from day one summarizes the trade-offs. This article goes further: how to weigh each factor, where module certification stops helping, and what a chip-down design needs to succeed. It is not legal advice; the rules depend on the product and market, and they change.

Start with volume, but think in total cost

A module costs more per unit than the radio chip and the parts around it. A chip-down design costs more up front: RF design, antenna tuning, extra prototype spins, and full radio certification in each market. The question is where those lines cross.

Pre-certified module
Chip-down from the start
Module first, chip-down cost-down later

Total cost as volume grows. A module is cheaper to start; a chip-down design costs more up front but less per unit, so past the break-even point it costs less overall. Illustrative, not to scale.

To estimate the crossing point for your product, gather:

The per-unit difference. Compare the module price with the chip, crystal, matching parts, antenna, any shield can, extra board area, and production RF test time. The gap can be smaller than the first quote suggests.
The one-time cost of chip-down. Engineering, extra spins, test equipment or lab time, radio certification for every market on your list, and the schedule risk of a later launch.
Realistic volume over the product’s life, not the optimistic launch forecast.

If break-even lies beyond any volume you can forecast with confidence, the module wins on cost. If you expect to pass it within the first production run or two, chip-down deserves a serious look. But cost is rarely the only factor.

The factors that can override cost

Size and antenna placement

A module fixes the footprint and, more importantly, the antenna position. Its certification assumes a particular antenna, ground plane, and clearance. In a small or oddly shaped enclosure, that keep-out may not fit where the antenna needs to be. Chip-down lets you place the radio and antenna where the product needs them.

RF performance and range

A module’s antenna works well in the vendor’s test fixture. In your enclosure, near batteries, displays, metal, and the user’s hand, it may detune and lose range. With chip-down you choose the antenna and tune the matching network to it as installed. That only pays off if someone does the tuning.

The surprise is often not metal but plastic, potting, and coatings. Any dielectric close to the antenna, in its near field, loads it: the resonance shifts down in frequency, the impedance moves, and a lossy material adds loss on top. Potting compound around a board, a conformal coating over the antenna, or an enclosure wall that sits tight against it can each do this, and I have seen it many times. The match measured on a bare board no longer applies. The fix is measurement and RF work: measure the antenna on finished assemblies, potted or coated and in the enclosure, then retune the matching network for what is actually there.

Control and power

Many modules run the vendor’s firmware and expose the radio through AT commands or a fixed interface. That is fine until you need a sleep mode, timing behavior, protocol feature, or update path the vendor did not expose. For battery products, power often decides it: the lowest-power operation usually needs direct control of the radio’s states and power rails (see Battery-powered design: budgeting and measuring power). Some modules let you run your own firmware on the radio’s processor, which removes much of this limit, provided your firmware keeps the radio within the power, channel, and region settings the grant covers. Check before you assume either way.

Supply chain and lifecycle

A module adds a supplier between you and the radio chip. If the module is discontinued or changed, you may need a new module and new integration testing even if the chip inside is still in production. Chip-down depends directly on the chip vendor. Either way, check lifecycle status and second sources before you commit, and read the vendor’s change-notification policy.

Where module certification stops helping

In the US, a module can hold a modular approval under 47 CFR 15.212; the FCC’s guidance on modules and their integration is in KDB 996369. A module that cannot meet every modular requirement on its own, for example because it has no shield, may instead hold a limited modular approval, valid only under the conditions in its grant. Canada has a similar modular approval under ISED’s RSP-100, with one notable difference: a handheld or wearable host needs its own certification unless it contains a single transmitter module whose RF exposure is exempt or very low, and always does if it contains more than one.

Any modular approval is conditional. The grant and the integration instructions list what you must keep: antenna type and gain, trace design, RF exposure conditions, and limits on transmitting alongside another radio. Typical ways to fall outside them:

An antenna of a different type, or with higher gain than the grant lists. Trace antennas must follow the module’s documented design exactly.
Placing the antenna within 20 cm of the user’s body when the grant assumed at least 20 cm. That changes the device from “mobile” to “portable” in FCC terms, and portable devices may need a SAR evaluation.
Adding a second radio that transmits at the same time, unless the grant or integration instructions already cover that combination.

Only the grant holder can make permissive changes. If your use falls outside the module’s grant, the usual paths are to ask the module vendor to file a Class II permissive change, or, with the module vendor’s permission, to move the approval to your own FCC ID through a change in ID (47 CFR 2.933), after which you can file the permissive change yourself. Both take time and lab work, so find out early.

The EU has no modular approval. Under the Radio Equipment Directive, the finished product is the radio equipment, and you, as its manufacturer, issue the Declaration of Conformity and apply the CE mark. The module’s test reports can support your assessment, but you remain responsible for the whole product. Compliance by market covers the approvals in each region.

In every market, the host product still needs its own emissions testing as a digital device, its own labeling, and any safety and cybersecurity requirements that apply. A module reduces radio testing, but does not eliminate it, and it does not remove compliance work.

Radio law is not the only approval. Bluetooth products need a Bluetooth SIG listing, and cellular products need carrier and industry certification, such as PTCRB in North America. That makes chip-down cellular impractical for most teams, and most cellular products use a certified module.

Timing and changes after certification

With a module, radio testing can be small and late, but confirm early that your design fits the grant. With chip-down, radio certification is on the critical path: book the lab well ahead, and freeze the RF design, antenna, and radio firmware settings before formal testing. After certification, changes to the radio circuit, the antenna, or the transmit power, including firmware that changes power or channels, can mean retesting or a new certification. Under FCC rules, changes to frequency-determining circuitry, modulation, or maximum power need a new grant, except software changes to a certified software-defined radio, which the grantee can file as a Class III permissive change. Route every post-certification change past whoever owns compliance.

The middle paths

You do not have to choose once.

Start with a module, plan a cost-down. Launch on a module to reach the market and learn what customers use. If volume grows, redesign with the same radio chip on your board. Choosing a module built on a chip you can buy directly keeps the firmware and protocol stack reusable. The cost-down needs full radio certification, so plan it as its own project.

Use the chip vendor’s reference design. Most radio chip vendors publish a reference layout with a tested matching network and antenna. Copying it closely, including its stackup and keep-outs, removes much of the risk of a first chip-down design. It does not certify your product.

What chip-down done well requires

A chip-down radio is routine when it is treated as RF work from the first schematic. I design chip-down radios regularly and take them through certification. The pieces that matter:

RF layout discipline. Short RF traces over solid ground, plenty of stitching vias, decoupling exactly as the chip vendor shows, and noisy circuits kept away.
A defined stackup. Controlled-impedance traces depend on dielectric thickness and material. Agree on the stackup with the fabricator before layout and put it on the fabrication drawing.
The reference crystal. Use the tolerance and load capacitance the chip vendor specifies, keep it close to the chip, and measure the carrier frequency error on prototypes. Many chips have a trim setting that can be calibrated per unit if needed.
Matching and antenna tuning. Copy the chip vendor’s matching and filter network exactly; it sets harmonic levels. Between it and the antenna, leave a separate adjustable matching network, and tune that on assembled boards, in the enclosure, with a vector network analyzer. A good match alone does not prove the antenna radiates well, so check radiated performance too, in a chamber or at least with comparative range tests.
Conducted test access. A switched test connector, or pads for a coax pigtail, on the RF path let you and the lab measure power, frequency, and spurious emissions directly. Labs usually want both conducted and radiated samples, so plan to build both.
An early test plan. List the radio standards for each market and the test modes the firmware must provide, such as continuous transmit on a chosen channel. Labs need those modes.
Pre-scans. Measure emissions on early prototypes, and book pre-compliance time before formal testing.
Production RF test and calibration. Decide how each unit’s RF path is checked on the line, and whether frequency or power needs per-unit calibration. Build it into the production test fixture; From prototype to production covers how that fixture develops across the builds, and Designing for production test covers conducted RF access and test limits.
The right equipment. A spectrum analyzer, a vector network analyzer, and near-field probes, or access to someone who has them.

If those are not available in-house, that is a reason to bring in help or choose a module, not to skip them.

The short version

Estimate the break-even volume honestly, then check whether size, RF performance, control, or power force the decision anyway. Read the module’s grant and integration instructions before you design, and remember that the EU has no modular approval. If you go chip-down, treat it as RF work from day one.

If you are weighing this decision for a product and want a second opinion, get in touch.