← Journal Testing · · 7 min

EMC pre-compliance testing before you book a lab

EMC pre-compliance testing on the bench finds most emissions and ESD problems early, so your first certification lab visit is not spent debugging.

Milo Perović
Founder & lead engineer

  • EMC
  • Pre-cert
  • QA
  • RF
EMC pre-compliance testing before you book a lab

Most connected products do not fail certification on the radio. They fail on a switching regulator nobody looked at, a USB cable that behaves like an antenna, or a button that resets the board when someone touches it on a dry winter day. EMC pre-compliance testing is how we find those problems on our own bench, weeks before a lab slot is booked and paid for.

This post is for founders and engineers who are a few months from certification and want to know what can realistically be checked in-house, with modest equipment, and what still needs an accredited chamber.

Why a first lab visit so often fails

A full emissions and immunity run at an accredited lab is expensive, and the calendar is often the bigger cost. If a product fails, you usually leave with a plot showing a peak over the limit and little else. The lab is not there to debug your board, and every retest means another booking.

The failures themselves are rarely exotic. In our experience the same handful of causes come up again and again:

  • Switch-mode power supplies, whose fast switching edges produce harmonics that reach well into the radiated band.
  • Clocks and fast digital edges, such as crystal harmonics, SPI or SDIO buses and USB, especially when routed over a gap in the ground plane.
  • Cables, which turn small common-mode currents into efficient radiators. A clean board can fail as soon as a one-metre lead is attached.
  • Poor return paths: split planes, missing stitching vias and connectors with a single ground pin.
  • Unprotected user-facing interfaces, such as buttons, exposed metal, USB ports and antenna feeds, that let ESD into sensitive circuitry.

None of these needs a chamber to find. They need a methodical look with the right probes, done early enough that a layout change is still cheap.

What EMC pre-compliance can and cannot tell you

An office or workshop is full of ambient signals: FM broadcast between 88 and 108 MHz, cellular bands, Wi-Fi and every other device in the room. Without a calibrated antenna, a known distance and a reflection-controlled environment, a radiated measurement on the bench is not an absolute number you can compare directly against a limit line.

What it does give you is relative data, and that is most of what you need. You can find which part of the board a peak comes from, see whether a ferrite or a series resistor reduced it by 10 dB, and check that a firmware change did not create a new problem. Conducted emissions on a mains port and ESD immunity can be measured much closer to the real test, because their set-ups are simpler to reproduce.

We treat pre-compliance as a way to arrive at the lab with known margins and no surprises, not as a replacement for the lab.

A modest bench set-up

You do not need a six-figure budget. A useful in-house kit for a small connected product looks like this:

  1. A spectrum analyser covering at least 9 kHz to 1 GHz, ideally higher if the product contains clocks or radios above that. Entry-level instruments with an EMI filter and quasi-peak detector option are good enough for relative work.
  2. A near-field probe set, with both magnetic loop (H-field) and electric (E-field) probes in a few sizes.
  3. A current clamp for measuring common-mode current on cables. This is often the best predictor of radiated failures in products with leads attached.
  4. A LISN (line impedance stabilisation network) for any product with a mains or mains-adapter port, so conducted emissions from 150 kHz to 30 MHz can be measured against a known source impedance.
  5. An ESD simulator to IEC 61000-4-2, which can be rented if buying one is hard to justify, plus a ground reference plane and the standard bleed resistor cables.

Running emissions checks that are worth the time

Start with an ambient scan, with the product switched off, and save it. Every later measurement is compared against it, otherwise you will spend an afternoon chasing a local radio station.

Then power the product in its worst realistic mode: radio transmitting, display on, motors running, charging if it charges. Sweep the board with a large H-field probe to find the hot areas, then switch to a small probe to pin down the source. Note the frequency of each peak and work out what produces it: a regulator switching frequency and its harmonics, a crystal multiple or a bus clock are usually easy to identify.

Next, attach the cables the product will ship with and measure common-mode current with the clamp. If a peak appears on the cable that is not obvious on the board, the fix is usually filtering or better grounding at the connector, not a change in the circuit producing it.

Fixes we reach for first

  • Series resistors or slower edge-rate settings on fast digital lines, where timing allows.
  • Spread-spectrum clocking on regulators and clock generators that support it.
  • Common-mode chokes or ferrites at cable entry points.
  • Stitching vias around board edges and alongside signals that change reference layer.
  • A shielding can over the noisiest section, with footprints added to the layout early so the can is optional rather than a respin.

Measure before and after every change, and record both. That record becomes part of the evidence in your bench report, and it saves time when the lab asks what has changed since the last build.

Immunity: ESD first, then the rest

For most small connected products, ESD is the immunity test most likely to cause a failure. The levels commonly applied to multimedia and radio equipment are ±4 kV contact discharge and ±8 kV air discharge, though the exact levels and pass criteria depend on the product standard you are tested against. Discharge to every user-accessible point: screws, seams, buttons, connectors and the gaps around a display.

Watch for three outcomes, not two. The product may carry on, it may reset and recover, or it may lock up or lose data. Product standards usually tolerate a temporary loss of function for ESD that recovers on its own, but a lock-up needing a power cycle is typically a failure. Firmware matters here: a watchdog, sensible brown-out handling and state that survives a reset can turn a failure into a pass.

Radiated immunity (IEC 61000-4-3) and fast transients (IEC 61000-4-4) are harder to reproduce properly in-house. We usually do rough checks only, such as a nearby handheld transmitter or a relay switching inductive loads, and leave the formal tests to the lab.

Knowing which standards apply

Pre-compliance is only useful against the right limits. For a radio product sold in the EU under CE RED, EMC is typically assessed against the EN 301 489 series, which refers to EN 55032 for emissions and sets its own immunity requirements. In the US, the digital parts of a device fall under FCC Part 15 Subpart B, and the intentional radiator under Subpart C. Using a pre-certified radio module simplifies the radio side considerably, but it does not cover the emissions of the rest of your board.

Confirm the list of applicable standards with your test lab before you start measuring. It is a short conversation, and it decides which frequency ranges, limits and test levels you should be aiming for.

What we would do before booking a lab

  • Agree the applicable standards and test modes with the lab in writing.
  • Save an ambient scan and keep it with every measurement.
  • Test the worst-case operating mode with production cables attached.
  • Aim for a comfortable margin below the limits, commonly around 6 dB, to cover the gap between bench and chamber.
  • Run ESD on every accessible point and confirm recovery without a power cycle.
  • Keep footprints for ferrites, filter capacitors and a shield can on the board, even if they are not populated.
  • Bring spare parts and a tested fix list to the lab, so a marginal result can be tried on the spot.

EMC is far cheaper to design in than to test out. Pairing these checks with environmental testing on a small-batch budget means the prototype batch arrives at the lab having already been stressed in the ways most likely to break it.

Working on something similar?Free scoping call →