Request a firmware review
Menu

RF certification test firmware: what the lab will ask for

Reader question: What test firmware and radio test modes will the FCC or EU test lab need from us, and does using a pre-certified module remove that work?

Micrograph-style illustration of a clear six-sided quartz crystal point growing from rock, with the tip tinted amber.

An FCC or EU radio test lab needs firmware it can command to hold the radio in fixed states: transmit continuously on a chosen channel at a set power and data rate, transmit an unmodulated carrier, and receive on a chosen channel while counting packets. Some EU tests also need the product to keep a normal connection busy with traffic. Chip vendors supply most of this, either as a test mode in the radio stack, such as Bluetooth LE Direct Test Mode, or as separate test firmware. Your job is to build it with your product’s radio settings, make it controllable in the lab, and keep it out of shipping units. A pre-certified module cuts the transmitter testing in the US, but the finished product still needs its own emissions compliance, and in the EU its own conformity assessment. This article is not legal or regulatory advice; your accredited lab interprets the rules for your product.

Why a lab cannot test your shipping firmware

Shipping firmware transmits when the protocol says so: a short advertising burst, a connection event, a packet every few seconds. A lab needs the opposite: the same signal, on a frequency it chooses, for as long as a measurement takes. The rules make that concrete.

For intentional radiators, 47 CFR 15.31 sets how the FCC measures. Paragraph (m) says measurements are made with the device operating at one, two or three frequencies per band depending on its width; for a band wider than 10 MHz, “1 near top, 1 near middle and 1 near bottom.” Your firmware has to put the radio on each of those channels on command. Paragraph (e) adds that measurements of how the input power or the radiated level of the fundamental varies are made “with the supply voltage varied between 85% and 115% of the nominal rated supply voltage”, and that battery-operated equipment is tested “using a new battery” (47 CFR 15.31, eCFR, checked November 2026). The transmitter has to hold steady while the lab turns a knob on the supply.

Chip vendors describe the same need from their side. Texas Instruments’ application note SWRA601 says regulatory testing typically requires configuring “the radio for static receiver and transmitter operation at certain frequencies.” Espressif’s RF test documentation calls the same thing a non-signaling or “fixed frequency test” that “directly controls the device to transmit specific signals without establishing a data connection.”

The procedures themselves live in standards your lab owns. Section 15.31 names ANSI C63.10-2020 for intentional radiators and ANSI C63.4-2014 for unintentional ones. In the EU, the lab applies harmonised standards; for short-range devices in the 2.4 GHz band, TI’s note lists EN 300 328 for radio, EN 301 489 for EMC, EN 62479 for human exposure, and EN 50566 with EN 62209-2 for handheld and body-worn devices. You do not need to read those standards to prepare firmware. You need to give the lab the controls it uses while applying them.

The test modes a lab typically asks for

The states below cover what the rules and the vendor notes above call for, each selectable by channel, power and data rate. Build every one your radio needs before the lab date, and confirm the list with the lab when you book.

Mode What the radio does Used for Controls the lab needs
Continuous modulated transmit Sends packets back to back, duty cycle near 100% Transmitter certification measurements Channel, power level, data rate or PHY, payload pattern
Unmodulated carrier (tone) Sends a single carrier with no data Single-carrier tests Channel, power level
Packet transmit at a set duty cycle Sends packets with gaps Transmit performance checks Channel, power, duty cycle
Receive with packet count Listens and counts good packets Receiver tests, including blocking in the EU Channel, data rate; a way to read the count
Bluetooth LE Direct Test Mode Standard test commands over a UART Bluetooth RF-PHY tests; also on Espressif’s CE list The DTM interface itself
Normal link with traffic Connects to the lab’s access point or tester and sends data EU adaptivity tests for Wi-Fi Network credentials, traffic generator
Worst-case normal operation Runs the product with everything active Unintentional emissions and EMC A mode that keeps every peripheral busy

Espressif’s test tool shows how fine-grained these controls get. Its Wi-Fi transmit test offers “TX packet” with a duty cycle under 50%, “TX continues” with a duty cycle “close to 100%, used for certification tests”, and “TX tone” for single-carrier tests, with rate, bandwidth, channel and attenuation in 0.25 dB steps (Espressif Wi-Fi non-signaling test). Its Bluetooth test selects a power level from 0 to 7, a channel and a payload pattern such as PRBS9 (Espressif Bluetooth non-signaling test). For the ESP32, its DTM test page maps those levels from -12 dBm at level 0 to 9 dBm at level 7.

Receive modes matter more in the EU than in the US. Under 47 CFR 15.101(b), receivers “operating above 960 MHz or below 30 MHz, except for radar detectors and CB receivers, are exempt from complying with the technical provisions” of Part 15, though still subject to its general conditions in 15.5. A 2.4 GHz receiver therefore carries little US receiver testing. In the EU, TI’s note records that EN 300 328 version 2.1.1, released in November 2016, “introduced receiver categories and a receiver blocking test”, which needs the receive mode and a way to read its packet count.

What the EU adds differs by radio. Espressif lists non-signaling, adaptivity and blocking tests for CE (Espressif CE certification), while its FCC page lists only the non-signaling tests. For Bluetooth LE at up to 10 dBm, TI’s note concludes that “all required testing for CE/RED can be performed with continuous (static) RF output such as for FCC testing.” Wi-Fi products are different: Espressif’s adaptivity test connects the device to an access point and runs UDP traffic for a long time, so the lab needs a build that associates and streams, not only a test mode.

How chip vendors deliver test modes

Vendors provide test modes in three forms: a standard interface built into the Bluetooth stack, vendor-specific commands you call from your own code, and separate test images run by a vendor tool. Which one you use decides how much firmware work is yours.

Bluetooth LE Direct Test Mode

Bluetooth LE has a standard test interface. TI’s note explains that RF-PHY testing places the device “in Direct Test Mode (DTM)”, where an external Bluetooth tester “can issue standardized HCI test commands to the DUT over a 2-wire universal asynchronous receiver or transmitter (UART) interface.” The interface is defined in the Bluetooth Core Specification, Volume 6, Part F.

Two details from TI’s note shape your hardware and firmware. First, “DTM is only intended for lab test purposes. When DTM is activated, normal Bluetooth Low Energy protocol communication is not permitted.” Second, if a product has no UART for DTM, it is “usually acceptable to modify the product to expose these pins provided the modification does not interfere with or alter RF or antenna performance.” Put test pads for a UART on the board now, while the layout is open.

For products whose application does not expose a UART, TI describes a Production Test Mode: the UART pins are reserved, but “the DTM interface is only activated by the application in response to a software API function call.” That is the pattern to copy on any stack. The test interface lives in the product’s own firmware, with the product’s radio configuration, behind a deliberate trigger.

Vendor commands called from your application

DTM covers Bluetooth’s own test cases. For regulatory testing, TI’s stack adds vendor-specific commands, HCI_EXT_ModemTestTxCmd and HCI_EXT_ModemTestRxCmd, to hold the transmitter or receiver at chosen frequencies, and lists three ways to invoke them: over a UART using TI’s network processor project, through Production Test Mode, or from “a stand-alone test sequence or test mode in the embedded application”, for example “controlled by a timer or push button interface.”

TI’s note also warns against shortcuts. Its SmartRF Studio tool drives the radio directly over JTAG without product firmware, but “for some regulatory testing purposes, the actual protocol stack firmware can be required.” And its network processor build may stand in for the product only if “both the protocol stack and radio configuration are identical” to the product’s.

Separate test images run by a vendor tool

Espressif takes the other route. Its EspRFTestTool package bundles a test tool, a download tool and a power-limit tool, and the package “contains all the necessary firmware for RF Test Items” (Espressif RF test guide). You flash the vendor’s non-signaling image over UART and drive it from the tool or a serial console.

A vendor image is quick to use, but it has to run with your product’s power settings, not the defaults. Espressif handles that with per-country power tables, generated by its PowerLimitTool into a phy_init_data file. The guide’s own example shows why the tables matter: it notes that the FCC table supports only Wi-Fi channels 1 to 11, and recommends keeping channels 12 and 13 at the channel 11 value rather than deleting them. Whatever power the lab measures has to be the power your shipping firmware will use.

Test PC vendor tool or script USB-to-UART adapter Shield box Device under test test build with the product's radio settings channel, power, rate, mode UART Bench supply varied from 85% to 115% of nominal power RF cable at antenna feed RF tester or spectrum analyzer tester control
Figure 1. A conducted RF test setup and the controls the test build must expose. Arrangement after Espressif's RF test guide; supply range from 47 CFR 15.31(e). Radiated tests use the unmodified product in a chamber instead of the cable. Checked November 2026.

Designing the test build

Build the test firmware from the same code base and the same radio configuration as the shipping firmware, add a small command interface and a cable-free fallback, and make it print exactly what it is. The aim is that every number the lab records describes the product you will ship.

One radio configuration, shared by both builds

Keep every radio setting in one place that both builds include: the power table per channel and data rate, crystal trim, any calibration values, the regulatory domain and the antenna configuration. If the test build carries its own copy, the two will drift, and your lab report will describe a product you do not sell.

The EU makes this a documentation duty as well. The Radio Equipment Directive’s technical documentation must include “versions of software or firmware affecting compliance with essential requirements” (Directive 2014/53/EU, Annex V). Record the radio configuration’s version alongside the firmware version in the test report, so a later change to the power table is visible as a change.

A command interface the lab can drive

A plain text console over UART is enough. Keep the command set small, documented on one page, and stable from the first pre-scan to the final test:

version                      -> fw 1.0.0-rftest, radio cfg r7, hw rev D
tx_mod <ch> <pwr> <phy> <pat> -> continuous modulated transmit
tx_cw  <ch> <pwr>            -> unmodulated carrier
rx     <ch> <phy>            -> receive; rx_count prints good packets
stop                         -> radio idle

If your stack supports DTM, expose it as well, on the UART pins the lab’s Bluetooth tester expects. Print the firmware version, radio configuration version and hardware revision at every boot, and again with the version command, so the lab can photograph it into the report.

Have the test build echo every command it accepts, with the resulting channel, power setting and mode, before it starts transmitting. When a measurement looks wrong, the first question is whether the radio was in the state everyone thought it was. A one-line echo answers that in the lab, and a saved console log answers it again when you read the report weeks later.

A way to run without a cable

Radiated tests happen in a chamber with the product closed. A UART cable that is not part of the product adds a conductor the shipping product does not have, so give the test build a cable-free path: hold a button at power-up, or set a value in flash before the session, and the unit boots straight into a fixed mode. A simple scheme is one boot per mode, cycling through bottom, middle and top channel at full power on successive power-ups, with an LED pattern showing which mode is active.

Behavior across the supply range

Section 15.31(e) varies the supply between 85% and 115% of nominal while the lab watches the transmitter. Check that nothing in your firmware switches off inside that range. A low-battery cutoff set for good battery life can sit above 85% of nominal: for example, on a hypothetical product with a 3.0 V nominal supply, the lab will go down to 2.55 V, so a firmware cutoff at 2.7 V stops the test. Disable the cutoff in the test build, or set it below the test range, and say so in the lab notes.

Also turn off anything that interrupts a long transmission. Sleep timers, connection supervision, periodic sensor reads and the watchdog all need handling in test modes. Keep the watchdog on, but feed it from the test loop rather than disabling it, so a hang still resets the unit.

A worst-case normal mode for emissions

Unintentional emissions and EMC tests look at the whole product, not just the radio. Where a product has external accessories, Section 15.31(i) has them attached and the device “fully exercised”, configured “in a manner that tends to produce maximized emissions within the range of variations that can be expected under normal operating conditions.” Give the lab a mode that runs every clock, bus and peripheral the product uses, at the highest rate it uses in the field, with the radio doing its normal job. Ask the lab which radio state it wants during those scans.

Conducted samples need hardware work too

Conducted tests connect the tester by cable to the antenna feed. Espressif’s guide describes the work on a module with a printed antenna: “cut the trace that connects to the PCB antenna feed point and solder the RF connection cable”, with the cable’s shield soldered to ground close to the feed. It also notes a firmware-relevant trap for its chips: the ESP32 calibrates its radio at power-on, so “The RF connection cable must be connected to the tester before the DUT is powered on for testing.” Check whether your chip calibrates at boot, and write the order of operations into the lab notes.

Does a pre-certified module remove the work?

A pre-certified module removes most transmitter testing in the US, provided you integrate it exactly as its grant allows. It does not remove the finished product’s unintentional emissions compliance, the labeling, or RF exposure in the final configuration. In the EU, there is no modular approval procedure in the Directive itself, and the finished product needs its own conformity assessment.

What modular approval means in the US

47 CFR 15.212 sets what a module must have to receive a modular approval: its own shielding, buffered data inputs, its own power supply regulation, an antenna that is permanently attached or uses a unique coupler, and testing “in a stand-alone configuration.” “All single or split modular transmitters are approved with an antenna,” so a different antenna puts you outside the grant. A limited modular approval can be granted when a module does not meet every requirement, and then the applicant must state “how control of the end product into which the module will be installed will be maintained.”

The rule then places duties on your product. If the module’s FCC ID label is hidden inside, “the outside of the device into which the module is installed must also display a label referring to the enclosed module”, with wording such as “Contains FCC ID”. The module “must comply with any specific rules or operating requirements that ordinarily apply to a complete transmitter”, and the module maker must give instructions explaining them. And it “must comply with any applicable RF exposure requirements in its final configuration.”

TI’s note summarizes the practical condition: products “can reuse a modular radio pre-certification provided that the module manufacturer’s layout and placement guidelines are strictly followed, including display of the module’s FCC ID.”

What the host product still needs

Your product is also an unintentional radiator, because it has clocks and digital logic. Under 47 CFR 15.101, “Other Class B digital devices & peripherals” need a Supplier’s Declaration of Conformity or certification before marketing, and 15.31(b) expects the measurement procedure and test equipment to be kept with the data for an SDoC. Collecting that data on the finished product needs the worst-case normal mode above and, depending on what the lab asks, a radio state from the test build.

Firmware can also take a module outside its grant. If your code sets the module’s transmit power, channel list or duty cycle, it can configure the radio beyond what was tested. Section 15.15(b) requires that “the adjustments of any control that is readily accessible by or intended to be accessible to the user will not cause operation of the device in violation of the regulations.” Lock the power table and channel list in firmware, and keep them within the module maker’s integration instructions.

Chip-down designs and split modules

If you put the radio chip on your own board instead of buying a module, there is no module grant to lean on: the finished product is the transmitter, and it goes through the full intentional-radiator tests with your test build. Avoiding that is the main thing a module buys you.

Section 15.212 also covers split modular transmitters, where “a radio front end with antenna” and “a transmitter control element” are approved together. Here the rule reaches into firmware directly: “The transmitter module must not operate unless it has verified that the installed transmitter control elements and radio front end have been authorized together,” using means that include “coding in hardware and electronic signatures in software.” If your design pairs a control processor with a separate radio front end under such an approval, the pairing check is a firmware requirement, and the test build must run it too.

What changes in the EU

The Radio Equipment Directive sets essential requirements for the radio equipment placed on the market, including that it “effectively uses and supports the efficient use of radio spectrum in order to avoid harmful interference” (Article 3(2)). The manufacturer carries out the conformity assessment, which must take “into account all intended operating conditions” and, where the equipment can take different configurations, confirm compliance “in all possible configurations” (Article 17(1)).

Where you apply harmonised standards published in the Official Journal for Article 3(2), you may use internal production control (Annex II). Where you apply them only in part, or none exist, Article 17(4) sends you to a notified body under Annex III or Annex IV. A module maker’s test reports can be strong evidence for your assessment, but the Directive’s text has no modular approval of its own; decide with your lab what the finished product must be retested for.

Question US, with a modular approval EU, under the Radio Equipment Directive
Is transmitter testing repeated? Largely not, if the grant’s conditions are followed The final product’s assessment decides; module reports may support it
Antenna Must be the approved antenna Part of the equipment assessed
Labeling “Contains FCC ID” on the outside if the module’s label is hidden CE marking and the manufacturer’s details on the product
Digital emissions of the host SDoC or certification under 15.101 EMC under Article 3(1)(b); TI’s note lists EN 301 489
Firmware that sets radio parameters Must stay within the grant and 15.15(b) Firmware versions affecting compliance go in the technical documentation

The Directive’s cybersecurity requirements, under Article 3(3)(d), (e) and (f), are a separate set with their own standards, and they do not change what the radio lab needs from your firmware.

Keeping test modes out of shipping units

A test mode that can transmit continuously, at full power, on any channel is a compliance problem if a customer can reach it. Remove it from shipping firmware at build time, or lock it permanently at the end of the production line, and prove it is gone before the first unit ships.

Section 15.15(b) is the rule behind this: controls a user can reach must not make the device operate outside the regulations. A hidden serial command is still reachable by anyone who opens the case and finds the pads. TI’s own Production Test Mode keeps DTM dormant until the application calls an API, which is safe only if nothing in the shipping application ever makes that call.

Three patterns work:

  1. Separate image. The test build is a different image, never installed on customer units. If your factory uses it, the station erases it and programs the shipping image before the unit leaves; the factory provisioning article covers that end-of-line order.
  2. Compile-time exclusion. One code base, with every test command behind a build flag that the release configuration never sets. A CI check fails the release if test command strings or symbols appear in the shipping binary.
  3. Permanent lock. The shipping image contains the test mode but disables it for good when a one-time flag is written at the end of the line. This is the weakest of the three, because the code is still present.

Add one more barrier through the update path. Sign test builds with a development key that production units refuse, so a test image cannot be installed on a customer unit over the air. That needs a bootloader that verifies signatures, and a production key kept separate from development keys, both of which belong in the release process anyway.

Manufacturing test firmware is a related but different build. The factory may need a quick transmit check on every unit, which is a radio test mode too. The manufacturing firmware work covers that build, and the same rule applies: whatever the factory can switch on, a customer must not be able to.

When to build the test firmware

Build the test firmware during DVT, run your own pre-scan with it, then freeze it and the radio configuration before the formal lab session. A change to the power table after formal testing is a change to the tested product.

The production readiness checklist puts this in the DVT gate: pre-certification radio tests run with the test firmware the lab needs. That timing matters for three reasons. The board must be close to final, because layout and antenna changes move the results. The radio configuration must be close to final, because it is what the lab measures. And a pre-scan with a rented spectrum analyzer or a short lab session finds a harmonic or a power problem while there is still time to change a matching network.

The Directive makes the change rule explicit for the EU: manufacturers must ensure that “changes in radio equipment design or characteristics” are “adequately taken into account” in keeping series production in conformity (Article 10(5)). Treat the frozen radio configuration like a signed release: tagged, recorded and changed only through a written change request.

In the US, the lab’s report goes to whoever grants the certification. TI’s note explains that certification comes “directly from the FCC or through a Telecommunications Certification Body (TCB)”, and that the FCC is the only option when “the equipment uses new technology or when the test methods are undefined or unclear.” Ask your lab which route your product takes when you book; it changes the paperwork, not the firmware.

Bluetooth products carry one more program alongside the regulatory tests. TI’s note states that “All Bluetooth Low Energy products must go through the Bluetooth SIG qualification and declaration process,” and its RF-PHY tests use the same DTM interface. Matter, Thread and Wi-Fi Alliance programs sit beside the legal approvals in the same way; the Matter over Thread or Wi-Fi article covers those certification paths. None of them replaces a radio approval, and a radio approval replaces none of them.

Example (fictional). Loamwick’s LW-1 soil probe talks Bluetooth LE to a gateway and runs on two AA lithium cells. During DVT, the team adds an RF test build to the same Zephyr code base. It shares the radio configuration file with the shipping build, exposes DTM on two test pads, and adds a four-command console for carrier, modulated transmit, receive and version. For the chamber, holding the probe’s button at power-up starts continuous transmit, and each later power-up moves on to the next of bottom, middle and top channel, with the LED showing which.

The first bench pre-scan finds the low-battery cutoff above 85% of the nominal supply, so the test build moves it below the lab’s range and the lab notes say why. Test builds are signed with the development key, which Rev D probes refuse, and the release CI fails if any test command string appears in the shipping image. The team books the lab with a one-page command reference, three conducted samples with a cable at the antenna feed, three unmodified radiated samples, and new cells.

When a test fails: what firmware can and cannot fix

Some failures can be fixed in firmware and some need the board to change, and the test build helps tell them apart. Power slightly over a limit, or an edge channel that fails while the middle passes, points first at the power table. Harmonics, poor receiver blocking and antenna problems point first at the hardware.

Per-channel power is the firmware fix to understand first. Espressif’s power-limit tool shows the mechanism: the power table is set per country and per channel, the certification result supplies an attenuation value, and the new setting is the target power minus that attenuation, in the tool’s quarter-decibel units. Its guide recommends that, “Except for low and high channels, the power values of other channels should be set to the same as the middle channel.” In other words, back off the channels that need it and leave the rest alone.

Symptom in the lab Where to look first What firmware can do
Output power over the limit on some channels Firmware Lower those channels in the power table
A failure only at the bottom or top channel Firmware, sometimes hardware Back off power on the edge channels
Frequency error Firmware or crystal choice Adjust crystal trim if the chip supports it
Harmonics or spurious emissions far from the carrier Hardware Little; filtering and matching are board changes
Emissions from clocks and buses Hardware, sometimes firmware Clock and drive-strength settings, where the chip offers them and the results point there
Mode will not hold for a long measurement Firmware Fix sleep, watchdog and supervision handling in the test build

Every fix to the power table changes the tested product, so record it the same way as any release: a new radio configuration version, a note of which measurements it affects, and the lab’s view on which tests to repeat. Then carry the same table into the shipping build, because the lab’s numbers only describe your product if both builds use it.

What to prepare before booking the lab

Ask the lab for its sample and firmware requirements in writing when you book, then work through this list. Most of it is firmware; the rest is paperwork a firmware engineer is well placed to write.

Item What to bring Why
Test build Signed, versioned, with a one-page command reference The lab drives every mode from it
Cable-free mode Button or stored setting that boots into each fixed mode Radiated tests in the chamber
Shipping build The exact image you will ship Normal-operation and emissions tests
Signaling build, Wi-Fi in the EU Associates with the lab’s access point and streams traffic Adaptivity tests
Conducted samples Units with a cable at the antenna feed Power and spectrum measurements by cable
Radiated samples Unmodified units in the final enclosure Chamber measurements
Power New batteries, and a supply connection if possible 15.31(e) testing with a new battery and a varied supply
Radio data Channels, power per channel and rate, antenna type and gain The lab sets up and reports against it
Module records Module FCC ID, grant conditions, integration instructions Modular approval conditions
A person A firmware engineer reachable during the session Questions and a fix if a mode misbehaves

Then check the firmware against the lab’s list one last time:

  1. Every requested mode works on bottom, middle and top channels, at every power level and data rate you will ship.
  2. The supply range holds. The unit transmits steadily from 85% to 115% of nominal, with no cutoff inside the range.
  3. Boot output prints firmware version, radio configuration version and hardware revision.
  4. Long runs survive. Continuous transmit runs for an hour without a reset, a sleep or a drop in power.
  5. The shipping image is clean. No test command, DTM entry or power override is reachable, and CI proves it.

If your first wireless product is heading for the lab, our connectivity work includes the test firmware a lab needs, built from the same code and radio configuration as your product. Fumbar is a Denver firm that writes embedded firmware and over-the-air update paths for hardware startups; it prepares firmware for labs and does not certify products. For the wider order of work, from keeping or rewriting prototype code to the four builds a product needs, see what to change before the first production run, and if you want a second reader before you book, a firmware review can look at the radio plan along with the code.

Facing this decision on a real device?

If this article describes a choice in front of you, a firmware review puts it in writing for your board, your code and your update plan.

Request a firmware review

Or email sales@fumbar.com, or call (720) 675-4202.