Post-Market · Component Changes

Does Changing a Component Require Retesting?

A second source, an end-of-life notice, a cheaper capacitor: which changes the report survives, which need re-measurement, and how to prove it either way.

← All guides EMC test bench where a product configuration is re-measured after a component change

A report does not describe a product family. It describes the product that was in front of the engineer on the day it was tested - that board revision, that firmware version, that adapter, and the component list recorded in the paperwork. So when a part changes a year later, the honest question is not "is the report still valid" but "did the change touch anything the report actually relied on".

That question has three possible answers, and all three are common. Nothing was touched, and a written assessment closes it. Some clauses were touched, and those clauses get re-measured. Or the change went through the middle of the tested configuration, and the product has to be treated as a new one for the affected requirements.

What decides which of those three you land in is not how expensive the part was. It is whether the part appears in the report's component list, and whether it behaves the same in the way the test measured.

What the report actually records about your components

Before deciding anything, it is worth opening the report and seeing what it committed to. Most certification reports carry the same five blocks:

Everything in those blocks is part of the claim being made. Change a part that appears in the component list, or an item in the product description, and you have changed the basis of a document you may already be relying on. Change a part that appears nowhere in the file, and the report is normally untouched.

Three questions that decide whether retesting is needed

  1. Does the change touch something that was measured? Four areas matter: the radio path (module, antenna, matching network, output power, channel plan), the emission path (switching controller, oscillator, filters, cable routing, shielding, grounding), a safeguard (insulation, fuse, capacitor, transformer, enclosure material), or a substance-restricted material (solder alloy, plating, cable insulation, plasticisers). If the changed part sits in one of those areas, that part of the report is in play.
  2. Is equivalence demonstrated, or only asserted? "Same specification" is a starting point, not evidence. The test measured behaviour: emissions at particular frequencies, temperature rise under a fault, output power in a specific mode. Equivalence has to be argued on the parameters the test depended on, not on the two lines of the datasheet that happen to match.
  3. Does the market have its own rule for changes? Some do, formally. The FCC sorts post-grant changes into classes under its permissive change rules in 47 CFR 2.1043 - changes that do not affect the radio frequency characteristics, changes that do affect them but stay within the same band and emission type, and changes that go beyond that and need a new certification. Where a notified body issued a certificate, that certificate covers the design the body examined, so a change to the certified design has to go back to that body. And a buyer, importer or marketplace may simply require that they are told. Scheme rules are not negotiable by argument.

Component changes by category

This table is the working version of those three questions. It is a starting point for a decision, not a replacement for one: two products can change the same kind of part and land in different places, because the margin each one showed when it was tested is different.

What changedWhat is affectedUsual route
Same part number from the same manufacturer, new production batch or date codeNothingNo assessment. Record the batch for traceability
Second source for a passive part of comparable construction (resistor, non-critical capacitor, connector)Usually nothing, occasionally EMC if it sits in the filter or cable pathDocumented equivalence check against the component list, supplier evidence filed
Filter parts: X and Y capacitors, common mode chokes, ferritesConducted and radiated emissions; safety, for X and Y capacitorsRe-measure the affected EMC clauses; update the safety component list with an equally approved part
Switching controller, oscillator, or a change to switching frequencyEMC emissions; possibly temperature under safety testingRe-measure emissions; re-check temperature rise if the operating point moved
Transformer, fuse, varistor, thermal cut-off, optocoupler, insulation or internal wiringSafetySafety re-evaluation. The replacement must carry an equivalent component approval, not just equal ratings
Enclosure material (for example PC+ABS to ABS)Safety - flammability and mechanical strength; EMC, if the enclosure was doing any shieldingSafety re-evaluation against the material requirements; EMC check only if shielding or the internal layout changed
Radio module, antenna, RF matching, output power, channel planRadio testing and RF exposureRe-measure the radio tests for the affected modes, then follow the market's change route
Battery cell, cell supplier, or the protection circuitBattery safety and transportRe-qualify the cell or pack and the transport test route
Solder alloy, plating, cable insulation, plasticisersRoHS / REACHSubstance reassessment; XRF screening, wet chemistry where screening is inconclusive
Firmware that changes transmit power, frequency use or enabled modesRadio; EMC, if switching behaviour changesRe-measure the affected tests. Firmware touching the radio is a configuration change, not a version number change
Label artwork, packaging, cosmetic moulding changeNothing, provided the mandatory marking and ratings are unchangedCheck the marking against the file. No retest

Why "same specification" is not "same part"

This is where most second-source plans go wrong, because the paperwork looks clean right up to the point where something is measured.

None of these are "different specifications" in the sense a purchasing system understands. All of them can change what the bench saw. The parameters that decide the question are rarely the parameters listed first on a datasheet.

What counts as evidence that a change does not need retesting

If the change is genuinely outside the report, that conclusion should exist as a document. Five things make up a defensible change assessment:

  1. A comparison table covering the parameters the test depended on, old part against new part, with the datasheet lines cited.
  2. Component approvals that carry across. For safety parts, an equivalent approval in the same recognised component category. A certificate on the replacement part is transferable evidence; an unapproved equivalent of the same rating is not.
  3. Supplier declarations in writing - material and substance declarations, an equivalence statement, a change notification. Collected when the change is made, not three years later when a buyer asks.
  4. A written conclusion filed in the technical file, signed by the manufacturer - or by the importer, if the manufacturer sits outside the EU. The manufacturer declares conformity; the laboratory only issues reports.
  5. The filing or notification the scheme requires, where a notified body, a certification body or a marketplace is involved.

One clarification that avoids a common mix-up: the CE marking is not re-issued after a change. It is the manufacturer's claim, and it stays valid only for as long as the claim holds. What gets updated is the technical file and, where the standards or the scope changed, the Declaration of Conformity, which lists the standards applied.

Four changes that come up most often

An end-of-life notice on a controller. A supplier sends a product change notification, the replacement is pin compatible, and the schematic is untouched. Pin compatibility says nothing about the oscillator, the internal regulator or the switching edge rates - and those are exactly what emissions were measured against. This is the change most often treated as administrative, and the one most often responsible for a re-measurement.

A cost-driven second source on the mains filter. The choke is bought from a second supplier at a lower price, with the same inductance and current rating. On the bench, conducted emissions in the lower part of the band creep up. Nothing failed the first time, which is the point: the original test had margin, the second source ate the margin, and the product now sits close to a limit it used to clear comfortably. This is why knowing the margin in your report is worth the effort before you change anything.

An enclosure material change. PC+ABS moves to ABS for cost or supply reasons. The safety report referenced the original material for flammability and mechanical strength, so the material is part of the safety case rather than a cosmetic detail. If the enclosure was also contributing to shielding or to the creepage distances around the board, the EMC file and the insulation review come into scope as well.

An "equivalent" radio module. The new module carries its own grant or certificate, which feels like it settles the matter. It does not. That document covers the module as tested, in the reference configuration, and not your host board, your antenna or your enclosure - the same limitation that applies when you are working out whether a product needs its own FCC ID or can rely on SDoC. Where a module's approval can be used, it is used through the host's own assessment, not instead of it.

Handling this without re-running the whole project

A component change register built from the report's own component list, rather than from your full BOM, does most of the work. Only the parts the file relies on can put the file at risk, and that list is short.

  1. Know which parts are in the report. Everything else can be sourced, substituted and re-priced freely from a certification point of view.
  2. Screen every engineering change before it is released to production. Two questions are enough at that stage: does this part appear in the report's component list or product description, and does the change sit in the radio path, the emission path, a safeguard or a restricted material.
  3. Collect evidence at the moment of the change. Suppliers answer equivalence questions quickly while they are still trying to win the business.
  4. Write the conclusion down. An assessment that was made and not filed does not exist when a buyer, an importer or a market surveillance authority asks for it.
  5. Tell the parties who need to know. Which one that is depends on the route - a buyer, a marketplace, a notified body or a certification body.
  6. Keep the record. In the EU, the technical documentation and the Declaration of Conformity must be kept for 10 years after the product is placed on the market, under Article 7(3) of the EMC Directive. Keep the change assessments in the same file and they answer most questions for the life of the product.

If you are still freezing a configuration before the first test round, the same discipline applies earlier: engineering samples versus production samples is the pre-test version of this problem, and it is much cheaper to get right before the bench slot than after. The wider question of when a report stops supporting the claim you are making with it - standard editions, configuration, and what the acceptor will accept - is set out in how long a test report is valid.

Where we come in

Change-impact reviews are routine work here, and they are usually small. We take the report you already hold and the list of parts that changed, then map each changed part against what the report recorded and what it relied on: the component list, the tested configuration, the modes and power levels, the material declarations.

You get back a written answer rather than a quotation for a full project - which clauses are affected, which can be closed with a documented change assessment, what evidence we need from your supplier to close them, and where measurement is unavoidable, which tests specifically. Where a scheme filing or a notified body notification is required because of how the product was certified, that is shown as its own step so it does not surprise you at the shipping date. If the change turns out to be covered by documentation, we will say so rather than sell you bench time. Our testing and certification scope lists what we run in-house and what we handle through partner laboratories, and our project approach explains how these reviews are scoped and priced.

If a component has been substituted, a second source added, or an end-of-life notice landed on your desk, send us three things: the test report you are relying on, the list of parts that changed, and the markets the product is sold into.

We will come back with which parts of the report are affected, which changes can be closed with a documented assessment, and which need measurement - clause by clause, rather than as a whole new project.

Send a component change for a retest assessment

Frequently asked questions

Does changing a component after certification require retesting?

Not automatically. It depends on whether the changed part appears in the report's component list or tested configuration, and whether it affects something the test measured - the radio path, the emission path, a safety safeguard, or a substance-restricted material. Changes outside those areas are normally closed with a documented change assessment. Changes inside them usually mean re-measuring the affected clauses.

Can I use a second-source component without retesting?

Only if you can show equivalence on the parameters the test depended on, not just equal ratings on the datasheet. For safety parts the replacement normally needs an equivalent component approval in the same recognised category. File the comparison and the supplier's declaration with the technical file, because the manufacturer is the party that has to defend the assessment.

My supplier sent an end-of-life notice. Do I have to retest?

Treat the replacement as a new part rather than a continuation of the old one, then apply the same three questions. Pin-compatible replacements frequently differ in oscillator behaviour, switching edges or internal regulation, and those are the parameters that were measured. Where the replacement is outside the tested areas, a documented assessment is normally enough.

The new component has its own test report. Does that cover my product?

No. A report or approval on a component or subassembly covers that item as tested, in the reference configuration. It is useful evidence inside your own assessment, but it does not extend to your host board, antenna, enclosure or production process. That is the same reason a module's radio approval is used through the host's own assessment rather than instead of it.

Do I need a new Declaration of Conformity after a component change?

Update it when the change affects the standards or the scope the declaration states, or when the product description in it no longer matches what you build. For a change fully covered by a documented assessment, the declaration can stay as it is, with the assessment itself added to the technical file. In the EU the declaration and the technical documentation must be kept for 10 years after the product is placed on the market.

How do I know if a change is serious before I pay for testing?

Look at where the part sits, not what it costs. A cheap part in the mains filter, in the radio path or acting as a safeguard carries far more risk than an expensive mechanical part that appears nowhere in the report. If you are unsure, a change-impact review against the report is usually a couple of days of work rather than a full test campaign.

WhatsApp
WeChat ID copied!