When should a certification test plan be written?

As soon as the product architecture is frozen and before any campaign is booked at a laboratory. The plan locks the product configuration, the regulatory scope and the applicable standards. Without that anchor, the lab is asked to test a moving target, samples are wasted and the campaign drifts. A typical milestone is engineering sample (ES) maturity: the plan is issued, reviewed and signed before the first formal slot is booked.

What is the difference between a project test plan and a campaign test schedule?

The project test plan is the framework document covering all regulatory scopes and the full product lifecycle, including re-certification rules. The campaign schedule is the operational slot booked at a given laboratory for a given subset of tests, with dates, samples and witnesses. One project plan spawns several campaign schedules, for example an EMC campaign, a radio campaign and a safety campaign run in parallel or in sequence.

How are golden samples handled in the plan?

The plan defines what a golden sample is for the project, lists its hardware revision, firmware version and accessory configuration, and ties each sample to a serial number, a photograph and a status tag (engineering, pre-production, production). Any drift between a golden sample and a tested unit during the campaign is grounds for pause and re-validation. The same logic applies to accessories and cables, since they influence EMC and radiated test results.

How is the sample count fixed?

By cross-referencing the test sequence, the destructive tests and the parallel campaigns. A typical EMC plus radio plus safety campaign needs several units, because some tests are destructive (transient immunity, abnormal operation) and some tests cannot share a sample with others (SAR head and body, conducted versus radiated emissions). The plan lists the minimum count per phase, plus a contingency margin for failed units or re-runs.

What acceptance criteria should the plan reference?

Each immunity test family in EMC has performance criteria, typically A (no degradation), B (temporary degradation, auto-recovery) and C (degradation requiring user action). The plan declares which criterion applies to which test, drawing on the generic standard that covers the product (EN 61000-6-1 and 6-2 for residential and industrial, dedicated product standards otherwise). For radio, acceptance is binary against published limits. For safety, the criterion is no hazardous fault under the declared conditions.

What triggers a partial or full retest after certification?

Any change that may affect the conformity envelope: a BOM substitution on EMC-critical components, a layout change near a transmitter, a firmware update that alters duty cycle or output power, a new accessory in the kit. The plan should reference the change control rules and the FCC Class II Permissive Change logic for radio, plus the harmonised standards re-baselining when an edition is replaced.

How does the plan handle pre-compliance versus formal compliance?

Pre-compliance is the screening phase, run in-house or in a pre-compliance lab, before any formal slot. The plan declares which tests are screened (emissions sweep, ESD spot checks, climatic pre-conditioning) and which criteria gate the entry into a formal campaign. A red flag on pre-compliance pauses the booking and returns the product to engineering. Pre-compliance does not substitute for formal accredited tests but de-risks the campaign.

Certification test plan: template and checklist

Q: Who signs a certification test plan?

At a minimum the project manager, the responsible electronics or radio engineer, and the quality or compliance lead. For products with a notified body involvement (RED Module B, machinery, medical), the notified body is informed and the plan may be reviewed during the technical assessment. The plan is a controlled document of the technical file, versioned and traceable to the engineering BOM.

Author Hugues Orgitello Last updated 27 May 2026 ~12 min read

Guide, certification test plan

A certification test plan is the controlled document that turns a regulatory ambition (CE, FCC, RED, PTCRB and so on) into a verifiable campaign. It names the product variants and configurations, the regulatory scope, the standards and their editions, the sample plan, the test sequence, the laboratories, the acceptance criteria, the risk register, the change control hooks and the deliverables. Without that document, the laboratory tests a moving target, samples are wasted and campaigns slip. This guide presents a structured template, the difference between a project-level plan and a per-campaign schedule, the rules of sample management, the sequence logic for EMC, radio, safety and SAR, the lab selection criteria, the acceptance criteria for immunity (A, B, C), the change control triggers and the deliverables that close the loop into the technical file.

Purpose and scope of a certification test plan

The test plan converts engineering intent into a controlled compliance campaign. It exists to remove ambiguity at three levels: what is the product, what regulations apply, and how is conformity demonstrated. Three readers consume it: the project manager who tracks dates and slots, the electronics or radio engineer who owns the design and the lab interface, and the quality or compliance lead who signs the declaration of conformity at the end.

A useful test plan answers seven questions in writing:

What product variants are in scope, with hardware revision, firmware version and accessories?
Which markets and certificates are targeted (CE, FCC, ISED, UKCA, RED, PTCRB and so on)?
Which standards apply, at which edition and with which national deviations?
How many samples are needed, in which configurations, and how are they tagged?
In what order are tests run, and which tests are destructive or sequence-dependent?
Which laboratories are used, with which ISO/IEC 17025 scope and which witness mode?
What are the acceptance criteria, the risk register and the change-control triggers?

The plan is a controlled document of the technical file. It is versioned, signed by the project manager, the responsible engineer and the quality or compliance lead. For a notified body procedure (RED Module B, machinery, medical), the notified body is informed and may comment during the technical assessment.

Project test plan versus per-campaign test schedule

The two documents serve different purposes and live at different cadences.

Aspect	Project test plan	Campaign test schedule
Scope	Full regulatory perimeter, all markets, full lifecycle	A subset of tests at a given laboratory
Cadence	Issued once, updated on major changes	Re-issued per booked campaign
Owner	Project manager and compliance lead	Lab interface engineer
Content	Variants, standards, sample plan, sequence, risks, change control	Dates, samples shipped, witnesses, lab contact, expected report date
Lifetime	Lives as long as the product is on the market	Lives for one campaign

A single project plan typically spawns three to five campaign schedules: EMC, radio, safety, SAR, sometimes a separate field-trial schedule for cellular network acceptance. Treating them as one document creates two failure modes: an oversized campaign schedule that no one updates, or a project plan that drifts because every lab booking rewrites it.

Product configuration: variants, options, worst case

The plan locks what is tested. Variants and options multiply the test matrix, so the team must declare which configurations represent the worst case for which test family.

Three families of decisions:

Hardware variants, for example two antennas (internal, external), two PSU options (USB-C, 12 V DC jack), two enclosures (plastic, metal). The plan declares the worst case per test: highest EIRP antenna for radio, lowest impedance PSU for conducted emissions, smallest enclosure for thermal.
Firmware configurations, for example test mode, certification firmware, production firmware. The plan declares which build is loaded for which test, with version and checksum. The certification firmware is usually a special build that exposes test commands (continuous TX, channel hopping disabled) and is not shipped to end users.
Accessory and cable set, for example bundled USB cable, antenna pigtail, power supply. The plan declares which accessories are in the typical kit and which represent the worst case for emissions or immunity.

For EMC, the worst-case logic typically combines the highest emitter (active radio, fastest processor mode), the most exposed accessory (longest cable) and the most resonant enclosure (smallest, with apertures). For SAR, the worst-case combines the highest output power, the closest user contact and the most exposed antenna. The plan documents the rationale so a notified body or a TCB can re-trace the choice.

Standards selection table by market

The heart of the plan is the standards selection table. It cross-references markets, certificates and standards with their edition and amendment status.

Market	Certificate	Standard family	Example
EU	CE under RED	Radio	EN 300 328, EN 301 893
EU	CE under RED	EMC for radio	EN 301 489-1, EN 301 489-17
EU	CE under EMC Directive	Generic EMC	EN 61000-6-1, EN 61000-6-3
EU	CE under LVD	Safety	EN 62368-1
EU	CE under RED Article 3.3	Cybersecurity	EN 18031-1
US	FCC ID	Radio Part 15	47 CFR 15.247, 15.407
US	FCC SDoC	Unintentional radiator	47 CFR 15.107, 15.109
US	PTCRB	Cellular	3GPP TS 36 and 38 plus PRDs
Canada	ISED RSS	Radio	RSS-247, RSS-Gen
UK	UKCA	Radio	EN 300 328 designated by OPSS

For each entry, the plan records the edition (year), the amendments, the date of cessation of presumption of conformity for the previous edition and the test methods that change. A common mistake is to start a campaign against a superseded edition and discover at report stage that the EU Official Journal no longer cites it.

The plan should declare the methodology for keeping this table current: a monthly check of the EU Official Journal of harmonised standards, the FCC Rules and KDB updates, the PTCRB PRD release notes and the ETSI TC ERM publications. See certification timeline for cross-cutting orders of magnitude per phase.

Sample plan: golden samples, pilot run, production-representative

Samples are the second-most expensive item after laboratory fees, and the most common cause of campaign slippage. The plan defines three sample types:

Golden samples: validated engineering units, tagged with a serial number, photographed, with documented hardware revision and firmware checksum. Used for pre-compliance and the first formal slots.
Pilot run samples: built on production tooling but not yet on the production line. Used to validate that the production process does not degrade conformity (assembly, soldering, shielding).
Production-representative samples: pulled from the production line, with the production firmware. Used for the final formal slots and reserved as reference samples for surveillance.

The plan locks for each sample: serial number, hardware revision, firmware version, accessory kit, photographs of internal and external markings, and a tag tracked across the campaign. A drift between two samples (different PCB revision, different antenna lot) during a campaign is grounds for pause and re-validation.

Sample count is fixed by cross-referencing destructive tests, parallel campaigns and contingency. A typical CE plus FCC plus PTCRB campaign for a cellular IoT product needs eight to twelve units split across EMC (three), radio and SAR (three), safety (one destructive), PTCRB (two) and contingency (one to three). The exact figure is set in the plan, not improvised at lab arrival.

Test sequence and dependencies

The sequence is not arbitrary. Some tests pre-condition the sample, some are destructive, and some cannot share a sample with others.

A reference sequence for a connected consumer product:

Climatic pre-conditioning: temperature and humidity cycling per the applicable safety standard, to season the unit before electrical tests.
Pre-compliance scans: radiated and conducted emissions sweep in-house or at a pre-compliance lab. See conducted emissions LISN test.
Formal EMC emissions: at an ISO/IEC 17025 accredited lab, against the generic or product-specific standard.
Formal EMC immunity: ESD, radiated RF immunity, conducted RF immunity, EFT, surge, dips and interruptions, magnetic field. See ESD IEC 61000-4-2, radiated RF immunity, conducted RF immunity.
Radio tests: TX power, spectrum mask, adjacent channel, spurious, duty cycle, DFS where applicable. Done after EMC because radio tests put the unit in continuous TX, which can damage near-by EMC reference loads.
SAR: dedicated rig, dedicated sample because of phantom contact. Body, head and limb positions per the applicable standard.
Safety: electrical, mechanical, thermal, abnormal operation per EN 62368-1 or the relevant product standard. Includes destructive tests (component faults, glow wire, ball pressure).
Functional safety where applicable: SIL or PL evaluation for safety-related functions, see IEC 61508 functional safety.
Laboratory measurement safety where applicable: for laboratory and measurement equipment, see IEC 61010 lab measurement safety.
Cellular acceptance: PTCRB at a recognised PTCRB lab, followed by carrier-specific acceptance (Verizon ODI, AT and T NAF and so on).

The plan declares which step uses which sample, the slot date, the witness mode and the expected report date. Parallel execution is encouraged when samples and labs allow it (EMC at lab A while safety prep at lab B), but the plan must explicitly list the dependencies.

Laboratory selection: scope, witness, MRA recognition

Lab selection is governed by four criteria: accreditation scope, recognition, witness mode and cost. The plan documents the choice for each campaign.

Criterion	What to verify	Where to find it
ISO/IEC 17025 scope	The exact standards and editions the lab is accredited for, not a generic claim	Accreditation body register (UKAS, COFRAC, A2LA, DAkkS, SAC)
Recognition	TCB recognition for FCC, CAB recognition for MRA partner countries, designation by a national regulator for radio	FCC TCB list, IC REL listing, national regulator pages
Witness vs full-service	Whether the team brings the EUT and runs the tests under accreditation, or hands over to the lab	Lab quote and the project plan
MRA recognition	Whether the lab can issue reports recognised under the EU-US, EU-Canada, EU-Japan, EU-Switzerland MRAs	MRA designation lists at the EU Commission and the regulator

Witness testing is faster when the engineering team holds deep test know-how, and reduces lab-side debug iterations. Full-service is faster when the team is small or the test family is unfamiliar. The plan records the choice and the contractual scope.

For radio, the lab must be recognised by the destination country regulator: a TCB-recognised lab in the US, a CAB-recognised lab for the Canada-EU MRA, a designated lab in Japan (MIC), and so on. A test report from an accredited but non-recognised lab is technically valid but not usable for type approval.

Acceptance criteria: performance criteria A, B, C

For EMC immunity, the generic standards define three performance criteria that the plan must declare per test family.

Criterion	Behaviour during the test
Criterion A	No degradation of performance, no loss of function. The EUT operates as specified throughout.
Criterion B	Temporary degradation or loss of function during the disturbance, with self-recovery without operator action.
Criterion C	Degradation or loss of function requiring operator action to restore normal operation (reset, power cycle).

The applicable criterion depends on the test family and the product family. For a residential product against EN 61000-6-1, the typical mapping is:

ESD contact discharge: B
Radiated RF immunity: A for continuous functions, B for non-continuous
Conducted RF immunity: A
EFT (electrical fast transient): B
Surge: B
Dips and short interruptions: B or C depending on duration

For an industrial product against EN 61000-6-2, criteria tighten (A or B where the residential plan would tolerate B or C). The plan declares the criterion per test and the means of observation (visual, log file, automated check).

For radio, acceptance is binary against the published limit. For safety, acceptance is the absence of a hazardous fault under the declared conditions.

Risk register and pre-compliance

The plan carries a risk register listing the known threats to the campaign and the mitigation. A typical register includes:

Emissions over-limit on USB cables: mitigated by common-mode chokes and shielded cables, screened in pre-compliance.
Radiated RF immunity failure on a high-impedance input: mitigated by RC filtering and ground plane review, screened with a near-field probe.
SAR margin tight: mitigated by antenna re-tuning or by declaring a usage condition.
DFS failure on Wi-Fi 5 GHz: mitigated by firmware tuning, screened with a chamber and a radar pulse generator.
Edition change of a harmonised standard mid-campaign: mitigated by issuing the test plan against the latest edition cited in the OJEU and tracking the cessation date.

Pre-compliance is the screening of these risks before any formal slot. The plan declares which tests are screened, what the gating criterion is and who owns the go or no-go decision. Pre-compliance does not substitute for accredited tests but reduces the probability of a formal failure that wastes a lab slot. See certification costs for orders of magnitude per phase.

Change control: what triggers retest

A certification is tied to a configuration. The plan defines what changes break the conformity envelope and what tests are repeated.

Change type	Typical retest	Reference
EMC-critical component substitution	Targeted retest on affected emissions or immunity family	component substitution rules
Antenna change	Full radio retest, SAR retest if applicable	FCC Class II Permissive Change, see FCC Class II PC
Firmware change altering radio behaviour	Radio retest at minimum, full report update for FCC Class II PC	KDB 178919
Layout change near a transmitter	Full radio retest plus SAR	FCC and ETSI rules
Enclosure change affecting EMC	Full EMC retest	Generic standards
Safety-critical component change	Targeted safety retest, ISO 26262 or IEC 61508 impact analysis if applicable	Safety standard chosen
Edition change of an applicable standard	Re-baselining against the new edition	OJEU citation, FCC Part rule update

The plan declares the change-control workflow: who logs the change, who assesses the impact, who triggers the retest. This workflow is usually integrated with the engineering change order (ECO) system.

Deliverables and traceability

The plan ends with the deliverables it produces, in the form they take in the technical file.

Test reports, one per campaign, signed by the lab, including the standard edition tested, the sample serial numbers, the test results, the photographs and the date.
Declaration of conformity drafts, one per market, referencing the test reports and the directives or rules met.
Technical file outline: product description, risk analysis, list of standards applied, test reports, DoC, user information.
Traceability matrix: each test result is tied to a standard clause, a sample serial number, a date and a report number.

The traceability matrix is the closing artefact. It allows a surveillance authority, a notified body or a customs official to walk back from any certificate to the underlying test result without ambiguity. It also feeds change control: when a standard edition changes, the matrix shows which results need re-baselining.

One-page test-plan skeleton

A template the team can copy into the project repository as a starting point.

Project: <name>
Product: <reference>, hardware rev <X.Y>, firmware <Z.W>
Variants: <list with worst-case rationale>
Accessories in kit: <list>

Regulatory scope:
  - EU: CE under RED, EMC Directive, LVD, RoHS
  - US: FCC ID, FCC SDoC
  - Canada: ISED RSS
  - UK: UKCA
  - PTCRB (cellular)

Standards table:
  - <market>, <directive>, <standard>, <edition>, <amendments>

Sample plan:
  - Golden: <count>, S/N <list>, HW rev, FW ver
  - Pilot: <count>, S/N <list>
  - Production-representative: <count>, S/N <list>
  - Contingency: <count>

Sequence:
  1. Climatic pre-conditioning
  2. Pre-compliance scans (in-house)
  3. EMC emissions (lab A, witness)
  4. EMC immunity (lab A, full-service)
  5. Radio (lab B, TCB-recognised, witness)
  6. SAR (lab B, full-service)
  7. Safety (lab C, full-service)
  8. PTCRB (lab D, PTCRB-recognised, full-service)

Acceptance criteria:
  - EMC immunity: <criterion A/B/C per test family>
  - Radio: <limit values per band>
  - Safety: no hazardous fault

Risk register:
  - <risk>, <mitigation>, <screened how>

Change control:
  - Triggers: BOM change on EMC-critical part, antenna change, FW change altering radio
  - Workflow: ECO log, compliance impact assessment, retest

Deliverables:
  - Test reports per campaign
  - DoC per market
  - Technical file
  - Traceability matrix

Signatures:
  - Project manager
  - Responsible engineer
  - Compliance lead
  - Date and version

This skeleton fits on one printed page. The detail goes into linked annexes (standards table, sample register, lab quotes, change-control log).

Cross-cutting references

CE procedure: EU regulatory framework and the directives the plan addresses
FCC procedure: US regulatory framework and Part 15 versus Part 22 to 27 split
RED procedure: radio-specific EU framework, Articles 3.1, 3.2 and 3.3
Certification timeline: cross-cutting orders of magnitude per phase
Certification costs: structure of lab fees, notified body fees, agent fees
Component substitution rules: triggers for retest after BOM change
IEC 61010 laboratory measurement safety: test plan considerations for laboratory equipment
Glossary: definitions of golden sample, witness testing, TCB, MRA, performance criterion

Frequent pitfalls

Pitfall	Consequence
Issuing the plan after the first lab slot is booked	Plan rewritten under deadline pressure, scope creep, missed standards
Skipping the standards edition in the table	Campaign run against a superseded edition, OJEU cessation discovered at report stage
Mixing project plan and campaign schedule into one document	Document becomes too large to maintain, version control drifts
Underestimating sample count	Campaign paused mid-way to build more units, slot lost
Treating accessories as out of scope	Failure on an accessory cable retroactively widens the test scope
No worst-case rationale for variants	Notified body or TCB challenges the choice, full retest demanded
No change-control linkage	A BOM substitution slips through, surveillance audit triggers a recall
No traceability matrix	Technical file fails an audit even though all tests passed

Sources & references

IEC 61000-6-1, EMC generic immunity for residential environments , IEC webstore.iec.ch/publication/59678
IEC 61000-6-2, EMC generic immunity for industrial environments , IEC webstore.iec.ch/publication/59679
ISO/IEC 17025, General requirements for the competence of testing laboratories , ISO www.iso.org/standard/66912.html
FCC KDB 178919, Permissive change policy , FCC apps.fcc.gov/oetcf/kdb/index.cfm
PTCRB Permanent Reference Documents (PRD) , PTCRB www.ptcrb.com/documents/
ETSI EN 301 489 series, EMC for radio equipment , ETSI www.etsi.org/standards-search