A 65 W desktop adapter passed its emissions test at 10 m with about 5 dB of margin, and the report was issued against Class A limits. Eight months later the same adapter was quoted to a US retail chain. Their compliance team asked for Class B evidence and sent the file back the same week.
Nothing in the report was wrong. The laboratory measured against the limit set it was given. The decision that caused the problem was made much earlier, when the class was chosen to fit the test budget rather than the channel the product would eventually be sold through.
Class A and Class B are the two emission limit sets used by most EMC emission standards. The choice between them is not a formality: it drives filter cost, test scope and the wording of the manual, and if it is made after tooling, it drives the enclosure as well.
What Class A and Class B actually describe
In CISPR 32, published in Europe as EN 55032, equipment is divided by intended environment, not by quality, price or power. Class B covers equipment used in residential environments: homes, apartments, hotel rooms, anywhere a television or radio receiver sits within a few metres. Class A covers everything else in scope - commercial, industrial and light-industrial premises - on the assumption that it will not be used where people live.
That is the whole distinction. A Class B product is not a better product. It is a product that emits roughly 10 dB less, because it has to coexist with broadcast reception in a living room.
The class is not a certificate, a mark, or an approval issued by an authority. It is a limit set inside a standard, recorded on the test report, and in the United States mirrored by a statement that must appear in the user manual. Two products can be identical in hardware and still be assessed to different classes; one of them will simply fail the tighter limit.
One caution before going further: not every product standard uses the A/B split. CISPR 32 uses it for multimedia equipment, CISPR 11 uses groups and classes for industrial, scientific and medical equipment, and some product standards apply a single set of limits. The question only has an answer once you know which standard applies to your product - which is normally the first thing to settle, and the subject of how a laboratory picks the standard.
How far apart the limits really are
For multimedia equipment under CISPR 32 / EN 55032, the gap looks like this:
| Measurement | Class A | Class B |
|---|---|---|
| AC mains conducted, 150 kHz-30 MHz, quasi-peak | 79 dBµV (150-500 kHz), 73 dBµV (0.5-30 MHz) | 66→56 dBµV (150-500 kHz), 56 dBµV (0.5-5 MHz), 60 dBµV (5-30 MHz) |
| Radiated, 30-230 MHz at 10 m | 40 dBµV/m | 30 dBµV/m |
| Radiated, 230-1000 MHz at 10 m | 47 dBµV/m | 37 dBµV/m |
| Radiated, 1-3 GHz at 3 m | 76 dBµV/m peak / 56 average | 70 dBµV/m peak / 50 average |
| Radiated, 3-6 GHz at 3 m | 80 dBµV/m peak / 60 average | 74 dBµV/m peak / 54 average |
In the United States, 47 CFR 15.107 and 15.109 use the same structure at a different measurement distance. Class B radiated limits are 100, 150, 200 and 500 µV/m at 3 m across the four bands between 30 MHz and 1 GHz; Class A limits are 90, 150, 210 and 300 µV/m at 10 m. The conducted limits in 15.107 for Class B are numerically the same as the CISPR 32 Class B figures above.
Ten decibels is a factor of about three in voltage and ten in power. If a product passes Class A with 4 dB of margin, it sits roughly 6 dB above the Class B limit at the same frequency - and 6 dB at 40 MHz is not something a different measurement session fixes. It is a filter, a shield, a grounding change or a different adapter.
That is the number that decides whether the class is a design input or a redesign: 10 dB in radiated emissions, 10 to 13 dB in conducted emissions.
Why running it again as Class B is not a paperwork step
The limit line moves down. The spectrum of the product does not. Whatever sat 5 dB under Class A now sits 5 dB over Class B, at the same frequencies, with the same cable layout and the same adapter.
Where the fix lands depends entirely on what the peak is attached to:
- If the peak follows the DC cable - common on adapters and small appliances - the emission is common-mode current on the lead, and the fix is a choke, a ferrite, better Y-caps or a different adapter. That adapter is often a bought-in item with its own report, so the change is a sourcing change rather than a PCB change.
- If the peak follows the board - switching node, clock harmonics, display ribbon - the fix sits at the source: slower edges, a snubber, spread spectrum, local shielding, or a ground stitch that was missing.
- If the peak only appears with the enclosure closed, the enclosure is resonating or the seams are radiating. That is a mechanical fix and the slowest of the three.
None of those are decisions a laboratory can make for you. What a laboratory can do is measure the new build against the tighter limit and say whether it holds - which is why the usual causes of EMC failure are worth reading before a formal booking rather than after it.
Four ways the class gets decided wrongly
1. The cheaper report. A product is quoted for both classes during development, and Class A is chosen because the run is shorter and the margin easier. The product then goes to retail, where the buyer's compliance team requires Class B. The adapter gets re-engineered, the emissions test is repeated, and if the fix touches the mains input, part of the safety assessment repeats with it. The saving at quotation stage is a fraction of that.
2. Two channels, one manual. A driver or controller is sold to installers for years, where Class A is defensible. Then the same model is listed on a marketplace for consumers. The hardware did not change and the report did not change, but the manual still carries the Class A statement telling the user that operation in a residential area is likely to cause interference and that the user must correct it at their own expense. The sentence is honest, and it is also a return reason.
3. The statement does not match the tested class. A product is tested to Class B, but the manual was copied from an older model and still contains the Class A text - or the reverse. Buyers' reviewers read the manual before they read the report, because the manual is what their customers will see. A mismatch between the statement and the assessed class is a documentation nonconformity that takes one email to find and about a week to correct.
4. Family coverage that covers the wrong build. The variant with the 12 V/2 A adapter is tested to Class B. The retail SKU ships with a different adapter that only appears in an older Class A report, or in no report at all. The paperwork is complete and describes a configuration nobody buys. The same gap opens when a second-source adapter is introduced after certification and nobody asks whether the emissions data still represents the shipped product.
Who decides the class, and when
Formally, the manufacturer declares the intended environment, and the class follows from it. In the EU the class is part of the assessment recorded in the technical file and reflected in the test report; the Declaration of Conformity cites the standard, not the class. In the US the class decides which limits under 15.107 and 15.109 apply and which user statement the manual has to contain under 47 CFR 15.105(a) or (b).
Commercially, three other parties decide it for you:
- The importer or retailer, who can require Class B for anything a consumer can buy, whatever the standard would technically permit.
- The marketplace, which asks for the report and the manual, and rejects a listing when the two disagree.
- The market surveillance authority, which reads the manual statement as a declaration of intended environment, then asks why a Class A product is being sold in a consumer channel.
The decision point is before layout, not before testing. A class is a filter and BOM decision; after tooling it becomes a re-spin. The rule we apply: if there is any realistic path to residential use - through any channel, in any market - test to Class B.
When Class A is the honest answer
Class A exists for a reason, and there are products where paying for Class B is waste. A machine controller installed in a factory, wired into a cabinet by qualified personnel, sold to businesses and never offered to consumers is a legitimate Class A product. So is equipment permanently installed in commercial premises that cannot reasonably end up in a home.
Two things still have to be true. The manual has to carry the Class A statement, and the technical file has to record the intended use, because that is the only thing supporting the choice if a buyer or an authority asks. And the choice has to be made because the environment is non-residential - not because the first Class B attempt failed. Using the weaker class to rescue a failing design is the one version of this decision that reliably comes back.
If you already have a Class A report and need Class B
A Class A report cannot be re-issued as a Class B report. The measurement result changes, so the evidence changes. A laboratory that offers to change the class line on an existing report is offering something that will not survive a buyer's technical review.
What does work is a delta programme:
- Pre-scan the current build against the Class B limits to see where the gap actually is, frequency by frequency.
- Fix the hardware, then re-scan to confirm the margin - with the production adapter, the production cable and the final enclosure.
- Book one formal emissions run on the modified build, with the test configuration agreed up front so it matches the build you ship (see what a test plan should contain).
- Check whether the fix touched the mains input. A new Y-capacitor or varistor is an emissions fix and a safety change at the same time.
RF testing usually survives this, because the radio path did not move. Safety usually survives it too, unless the change sits on the primary side. Emissions is the part that has to be repeated - and the part that a pre-scan makes cheap.
We run EMC, RF and SAR testing to both classes in-house, and the first question we ask on a Class A enquiry is which channel the product is going into. If the answer is retail or a marketplace, the Class A report will not be enough, and it is better to know that before a formal slot is booked. If the answer is genuinely industrial, we test to Class A and put that in writing.
The answer in one paragraph
Class A and Class B are two emission limit sets inside the same standard, split by intended environment rather than by product quality. Under CISPR 32 / EN 55032, Class B radiated limits at 10 m are 30 dBµV/m from 30 to 230 MHz and 37 dBµV/m from 230 MHz to 1 GHz, against 40 and 47 dBµV/m for Class A - a 10 dB difference - while the conducted limits differ by 10 to 13 dB. In the United States the class determines which limits under 47 CFR 15.107 and 15.109 apply and which user statement the manual must carry under 15.105(a) or (b). Because the difference is 10 dB of actual emission, a Class A report cannot be turned into Class B evidence by paperwork: the measurement has to be repeated on hardware that meets the tighter limit, and the cheapest moment to build that hardware is before the tooling is cut.
Tell us the product, the target market and where it will actually be sold - retail, installer or industrial - and we will confirm which class the applicable standard expects and whether your current report can carry it.