A pre-compliance test is an engineering measurement taken before the formal one: the same kinds of emissions and immunity checks, on a bench or in a shielded room, with no accredited report at the end. Clients ask whether it is worth the money because nobody enjoys paying for the same measurement twice.
It is worth it in one situation - when failing the formal test would cost you more than the pre-test. That depends less on the product than on your deadline, how mature the design is, and how many changes are still open. Below is how we decide, including the cases where a pre-test is simply a wasted bench round.
What pre-compliance testing actually is
Pre-compliance testing is the informal round: measurements taken to find out where a product stands, not to prove where it stands. It runs on engineering samples or early production units, it can be done in-house or as a laboratory's engineering service, and the deliverable is plots and engineering data rather than a test report.
The terminology matters here, because it is where expectations go wrong. A test report is a laboratory's record of an accredited test against named standards; a Declaration of Conformity is the manufacturer's own declaration of conformity; a certificate is issued by a certification body under a particular scheme. A pre-test produces none of those three things, and no amount of good data changes that.
What a typical pre-test covers
- Conducted emissions, 150 kHz to 30 MHz. This is where a pre-scan is cheapest and finds the most, because a spectrum analyser with a LISN on a bench is enough to see the shape of the problem.
- Radiated emissions, 30 MHz upward. A near-field probe sweep locates the frequencies that radiate, but absolute levels need a proper chamber, a calibrated antenna and the standard's own layout. Use probes for relative comparisons, the chamber for verdicts.
- Immunity spot checks. ESD, EFT/burst and radiated immunity at the levels your market's standards prescribe. A reset or a dropped packet during immunity is a design problem, not a filtering problem, and it is much cheaper to find it before the design freezes.
- Radio parameters for wireless products: output power, occupied bandwidth, spurious and unwanted emissions, and band-edge behaviour for each mode you intend to declare.
What a pre-test can and cannot tell you
| A pre-test can | A pre-test cannot |
|---|---|
| Show which frequencies are likely to fail, and by how much margin | Certify that a product complies with any standard |
| Compare a change reliably when the before and after runs use the same bench and setup | Guarantee the same result in another laboratory's chamber, or on another day |
| Expose design-level causes early - grounding, return paths, poor shielding, a noisy switching supply | Replace the formal test or be submitted to a buyer, marketplace or authority as evidence |
| Reduce the number of unknowns in the formal session | Compensate for a sample that does not represent production |
There is also a measurement caveat worth stating plainly. An accredited formal test follows the standard's specified setup: the right LISN, the chamber, the table height, the cable dress, calibrated instrumentation. A quick bench scan with a different LISN, a different cable layout and an uncalibrated antenna can differ from that setup by several dB. That difference is fine when you are hunting for problems. It is not fine when you are deciding that you are safe because you passed by 1 dB.
The real cost of failing the formal test
When clients compare a pre-test fee with a formal test fee, they are usually comparing the wrong two numbers. What a fail actually costs:
- Bench time again. A re-test costs roughly the same as that item's original test. The bench slot is the expensive part, not the paperwork.
- The rebooking delay. Because chamber schedules are booked out, a failed item usually waits for the next available slot rather than for tomorrow. This is frequently the larger cost, and it lands on your shipping date rather than on the invoice.
- The fixes themselves. Filters, ferrites, a shielding gasket, a re-spun board - components and engineering time that a pre-scan would have surfaced while the design was still open and changes were cheap. The causes behind those failures, and which fixes actually work, are covered in why products fail EMC testing.
- Schedule damage downstream. A two-week slip near a shipment window can turn sea freight into air freight. That single decision often decides whether the whole test budget mattered.
The sequence after a fail - what the laboratory sends, what you ask for, and what has to be retested - is a process question rather than a cost question, and it is worth reading what happens after a failed certification test before you are in the middle of one.
When a pre-test pays for itself
- A new platform that has never been certified. First-time designs have no track record, so the first formal test effectively becomes a debugging session at formal-test prices.
- A radio you have not certified before. A new module, a new band, a new mode, or a variant your laboratory has not seen.
- Changes made after the last test. A different power supply, a re-routed harness, a new antenna, a different switching regulator, or firmware that alters output power.
- A hard date. A buyer delivery window, a peak retail season, a container booking that cannot move.
- A history of marginal results. If your last product passed with almost no margin, assume the same design language will do it again.
- Expensive failure modes. Where rework is costly, or where tooling is already cut and a board change is not casual.
When it is a waste of money
- A proven design with nothing changed. If the platform has been certified repeatedly with the same architecture, the same supply and the same layout family, a pre-test adds little beyond the formal test you have to run anyway.
- Cosmetic changes. A new colour, a new label, a new enclosure in the same material and thickness. That is a documentation question, not a bench question.
- A sample that is not representative. Pre-testing a hand-built prototype and then producing with a different board revision or a longer internal harness simply relocates the surprise. If your samples are still engineering builds, say so and list every difference - the distinction between an engineering build and a production unit is set out in using engineering samples for certification.
- Chasing margin with no stopping rule. "One more filter" turns into an endless loop, and the loop eventually costs more than the re-test it was meant to avoid.
Four situations that come up every month
The pre-scan used a prototype power supply. Conducted emissions passed with comfortable margin - measured with an adapter that was not the production one. When the production-intent unit arrived with a different supply, the low-frequency end of the plot moved well past the limit. The measurement had been correct; the sample had not been. The working rule since then: the pre-tested unit and the production BOM have to be the same list, or the differences have to be written down before the scan.
Only one radio mode was measured. A product with two wireless modes was pre-scanned in the mode that happened to be working that week. The formal scope declared both, and the worst case turned out to be the mode nobody had measured. Multi-mode products need the worst-case decision made before the scan, not after.
A component swapped after a pass. A client replaced a DC-DC converter with a cheaper pin-compatible part on the reasonable assumption that pin-compatible means electrically equivalent. Switching frequency and edge rate are not part of the pinout. The radiated results moved enough that the earlier scan no longer described the product, and the change was not in the paperwork either.
No pre-test, to save the fee. A client booked the formal test directly and failed conducted emissions by a narrow margin on a single item. The retest itself cost what a retest always costs; the rebooking delay was the part that hurt, because the chamber was full and the shipment date was not flexible. The saving was real, and so was the delay.
How to scope a pre-test so it is actually useful
- Write the pre-test scope from the formal scope. Same standards, same modes, same supply conditions. A pre-test that measures something other than what the formal test will measure buys nothing.
- Test the worst case deliberately. Highest power mode, longest cable, the supply voltages that matter, and temperature extremes where the standards call for them.
- Use the production-intent configuration and document the differences if you cannot. That single page decides whether the results mean anything.
- Ask for plots and raw data, not a pass letter. The value is in the spectrum: where the margin is thin, which frequency is closest to the limit, and how it moves when you change one thing.
- Set a margin target and stop there. Many laboratories and manufacturers work to an internal margin of a few dB below the limit as a buffer against sample variation and different benches. That is an engineering convention, not a requirement written into the standards - but whatever number you choose, decide it before the first scan rather than after.
- Freeze the configuration after the pass. Treat the pre-tested build as a baseline, and assess every change against it instead of assuming.
How we run pre-testing projects
Pre-testing only saves money when it is planned against the formal scope, so that is where we start. We run pre-scans as an engineering service, hand back the plots and a short list of items that need attention with the margin at each one, and state which configuration was measured and what would change the outcome. When a design is mature and unchanged, we say the pre-test is unnecessary and quote the formal scope directly - a wasted bench round is not a service to anyone. When it is a first-article radio or a platform with a fixed delivery date, the pre-scan and the formal session are planned together, so the formal booking is not the first time the product sees a chamber. The testing and certification services we provide cover EMC, radio, safety and battery work, and the pre-test plan is written by the same engineers who will interpret the formal results.
If you are weighing up whether to pre-test, send the product description, the standards your buyer or market expects, what has changed since the last test, and your delivery deadline. We will tell you whether a pre-scan is worth booking - and if it is, exactly what it should cover. If it is not, we will say so and quote the formal scope instead.