Your samples left the factory last week. The deposit is paid, the purchase order is confirmed, and since then you have received one email saying the samples arrived and then silence. For most clients this is the least comfortable stretch of the whole project. You cannot see the bench, you do not know what is being done to your product, and a week without news feels like a week of nothing.
It is not nothing. A testing project runs through a fairly fixed sequence, and most of it happens quietly. Here is what actually happens after your samples arrive, and why the questions that suddenly appear in the middle of the project are a sign that things are going normally.
First: sample intake and the configuration check
The first task is not testing. It is comparing what arrived against what the documents say should have arrived. Samples are photographed, serial numbers and firmware versions are logged, and the physical units are checked against the bill of materials, the manual draft and the ratings that will appear on the label.
This is where small mismatches surface. A BOM lists a 12 V / 1 A adapter, but the sample arrived with a 12 V / 2 A unit because the factory substituted what was in stock. A manual describes a settings menu the current firmware does not have. The model name on the carton does not quite match the model list in the quote. None of these are disasters on day one. All of them are expensive to discover after the report is issued, because a report describes the configuration it actually tested - and if that is not the configuration you ship, the paper does not cover your product.
Engineers spend real time on this stage for exactly that reason. The comparison is boring, and it prevents retests.
Next: the test plan and worst-case selection
From the scope in the quote, the laboratory turns the standards into concrete test items and decides which configuration of your product gets exercised in each one. Not every variant is tested - the plan picks the worst case, the mode most likely to produce the highest emissions or the weakest immunity.
Typical worst-case decisions: transmit mode at maximum power rather than idle, motor at full load, screen at maximum brightness while the battery charges, longest cable attached. For a device with more than one radio - say Bluetooth plus a proprietary 2.4 GHz protocol - the plan also decides whether one mode or both go on the bench. This is exactly why the laboratory asks about every radio mode before testing starts. It is not to inflate the quote; an un-declared mode surfaces at this stage anyway, by which point the schedule is already built around it.
Then: pre-scans, followed by the formal runs
For emissions work, testing usually starts with pre-scans. A pre-scan is a quick, exploratory measurement: the data does not go into the report, and its only purpose is to show where the peaks are while problems are still cheap to fix. Formal runs come after, on a calibrated setup with the cable routing documented and the arrangement photographed, so the measurement could be repeated by another laboratory and reach the same result.
Immunity testing runs the other direction. Instead of measuring what your product sends out, the bench deliberately throws interference at it - electrostatic discharge, fast transients on the power line, radio fields at defined frequencies - and watches how it behaves. Whether a brief display flicker during a transient burst counts as acceptable or a failure is part of the acceptance criteria agreed in the plan, which is another reason the questions at the start matter.
The bench measurement itself is often the shortest part. Setup, documentation and data reduction take longer than the measurement does.
The questions in the middle of the project
Sooner or later, an engineer writes with a question. This is normal, and it is usually a good sign - it means someone has looked at your product closely. The common ones:
- Which firmware version is on this sample, and is it final? The report will name the version it tested. If production ships with different firmware, the report no longer describes the product you sell.
- What is the longest cable a customer will use? Radiated results at the lower frequencies are dominated by currents on the cables, so a 3 m cable and a 1.2 m cable can produce different plots.
- Can the enclosure be opened without tools? This decides how the safety testing approaches the internals.
- Does it matter if the display resets during a transient test? This sets the acceptance criteria for the immunity runs.
The firmware question deserves its own warning. Projects do get halfway through the bench when a firmware update arrives from the R&D side - a bug fix, or a feature the sales team promised a buyer. The right move is to stop and tell the laboratory, not to wait and mention it at report time. A mid-project firmware change may require part of the testing to be repeated, and the earlier that decision is made, the less bench time is wasted. Freezing firmware before samples ship is one of the cheapest scheduling decisions available to you.
When something fails along the way
If a pre-scan shows a problem, many laboratories pause and tell you before spending formal bench time on a configuration that cannot pass. You then decide whether to fix and re-test. If a formal run fails, the result is documented and discussed either way - a failure is not hidden, it is data. What happens after a failure, from root-cause discussion to retest scope, is covered in detail in what happens if a product fails certification testing.
After the bench: the report
When the final runs are done, the raw data is reduced into plots and tables, the results are compiled into a draft, the draft is reviewed, and the report is issued. It records the tested sample, its configuration and firmware version, the standards and editions used, the test dates and the results. The drafting and review stage takes days rather than hours - the step-by-step durations are laid out in how long certification testing takes.
Can you watch the testing?
Most laboratories allow witness testing if it is arranged in advance, and for a complex or safety-relevant product it can be worth the trip. In practice, the usual arrangement is lighter: setup photographs, progress updates and partial data at agreed milestones. If you want to witness, say so when the project starts, not when the bench is already running.
What to have ready while testing runs
The customer side controls more of the schedule than most clients expect. Four things help:
- One contact who can answer engineering questions within a day. A bench can sit idle for a week waiting for an answer about cable length.
- Frozen firmware. The version declared and the version shipped must match.
- Spare samples and accessories. Some tests are destructive, and a missing adapter can pause the project.
- Honesty about what might still change. If a redesign is planned, the laboratory can sequence the tests so the stable parts go first.
It also helps if the samples you send are the ones you intend to certify. The difference between engineering samples and production samples decides how much of the report survives into your shipping product.
Where we come in
On our side, the sequence above is planned before the samples arrive: the test plan, the worst-case configurations and the acceptance criteria are agreed in writing at the start, so the questions that normally appear mid-project are asked while the plan is being built instead. During the project you get photographs and status updates at fixed points, and when a question does come up, it comes with the reason and the options - not just a request for information. The full scope of our testing and certification work covers the EMC, radio, safety and battery testing behind most consumer and industrial electronics.
If your samples are about to ship, send over the firmware version, the accessory and cable list, and every operating or radio mode you plan to declare. We will confirm the test plan and worst-case configurations with you before the first measurement - the problems that cost the least are the ones caught before the bench starts.