Module or chip-down?
The crossover volume where designing the radio in stops costing more than buying it pre-certified — using your numbers, not a rule of thumb.
Recommendation
—
—
The arithmetic
module_total = V · C_module
chipdown_total = V · (C_bom + C_area) + NRE + cert + spins + delay_cost
V_crossover = (NRE + cert + spins + delay_cost) / (C_module − C_bom − C_area)Two straight lines. The module line passes through the origin with a steep slope; the chip-down line starts high and climbs slowly. Where they cross is the volume at which the up-front investment has paid for itself. Everything else on this page is about getting honest values into those two slopes.
Where the estimates usually go wrong
- Comparing a module against bare silicon. The module contains a transceiver, crystal, matching network, shield and often an antenna. Compare it against all of those or the delta is flattered by several euros.
- Certification scoped for one market. RED for Europe is one bill; FCC and ISED for North America is another; other regions again. A module usually arrives with several of these already held.
- Assuming one spin. First-pass radio designs from a team without recent RF experience rarely clear EMC. Budgeting one extra spin is optimistic, not pessimistic.
- Pricing engineering at salary cost. The relevant figure is what that engineer is not doing instead.
- Leaving schedule out entirely. Three months of delayed margin is frequently larger than the whole certification bill.
What the numbers cannot tell you
Reasons to take the module anyway
- Nobody in the building has taken a radio through certification recently. The schedule estimate is then a guess, and guesses run long.
- The product is a first version and the volume forecast is a hope rather than a purchase order.
- Certification is on the critical path to a tender or a season.
- You need several regional approvals quickly.
Reasons to design it in anyway
- Board area or height genuinely will not accommodate a module.
- The module vendor has revised or discontinued parts on you before. Single-sourcing a certified module is a real supply risk, and an EOL notice forces re-certification on their timetable.
- You need control of the RF front end — output power, antenna diversity, an unusual band plan.
- Volumes are far enough past crossover that the recurring saving dwarfs everything else.
Where this stops being accurate
- Both routes are modelled as linear. Real BOM pricing steps at volume breaks, and both curves have discontinuities this does not capture.
- Certification cost varies widely by lab, by market and by how many standards apply. Get a quote rather than trusting a default.
- No allowance for module EOL and forced redesign, which is the tail risk of the module route.
- Ongoing maintenance is excluded — under the CRA both routes now carry a multi-year support obligation.
Design guidance, not a manufacturing instruction. Verify every result against your fabricator's stack-up and the applicable standard before release. Closed-form models are approximations; the fab's field solver and process window are the authority.
Questions
What people ask about this
What is the honest rule of thumb?
Below roughly two to three thousand units a year, a pre-certified module almost always wins. Above ten thousand, chip-down usually does. Between those, it depends on how much the schedule is worth to you and how much radio experience is already in the building. The point of this tool is that "roughly" is not good enough when the gap is tens of thousands of euros, and your numbers are not the industry average.
What does chip-down actually cost that a module does not?
Radio certification is the big one — full RED testing across EN 300 328, EN 301 489, EN 62368-1 and now EN 18031 runs to five figures, and FCC/ISED again if you sell in North America. Then antenna design and tuning, which is specialist work plus chamber time. Then the EMC iterations you did not plan, because the first spin rarely passes. A module carries all of that already done, which is what the unit premium is buying.
Is the module premium really the whole difference?
No, and this is where the arithmetic usually goes wrong. Chip-down also needs the passives, the crystal, the matching network, the antenna and the extra board area — so the true delta is smaller than "module price minus chip price". This tool asks for the full chip-down BOM rather than just the silicon, because comparing a module against a bare transceiver flatters chip-down by several euros a unit.
What about the risks that are not costs?
Two matter. A module ties you to one vendor: if they revise or discontinue it, you re-certify anyway, and you had no control over the timing. Chip-down ties you to your own competence: if nobody in the building has taken a radio through certification, the schedule estimate is fiction. Neither is a number, and both should move the decision. Below the calculator we list what tips it either way.
Why does time to market appear as a cost?
Because for most products it is the largest number on the page and it usually goes unrecorded. If a product earns margin once shipped, three extra months of development is three months of that margin not earned, plus the risk of missing a season or a tender. Setting it to zero is fine if the launch date genuinely does not move — but be honest about whether that is true.
Want the answer checked against a real design?
We do this analysis properly — with the certification scope your market actually needs and an NRE figure based on the silicon you are considering. Often the answer is 'buy the module', and we will say so.