EMC Pre-Compliance Testing for Maker Electronics: Near-Field Probes and Passing FCC/CE Before You Ship

This site's guide to EMI/EMC noise suppression covers the design side — ferrite beads, decoupling, grounding, the things you do while building a circuit to keep it quiet. This is the other half: how you actually find out, before you ship or sell something, whether it's radiating more than it should. The moment a project stops being a one-off for your own bench and starts being something you sell, give away at scale, or connect to someone else's mains wiring, "it seems to work fine" stops being good enough — and the gap between a clean-looking schematic and a part that actually passes FCC Part 15 or CE EN 55032 is bigger than most makers expect.

What You're Actually Being Tested Against

Two separate things get measured, and they fail for different reasons:

• Radiated emissions: RF energy your device broadcasts through the air, measured by an antenna some distance away in an anechoic chamber or open-area test site during formal certification. This is what a switching regulator's harmonics, a microcontroller's clock harmonics, or a poorly-filtered motor drive leak out.
• Conducted emissions: noise that travels back out through the power cord itself, measured with a Line Impedance Stabilization Network (LISN) between your device and the mains. This mostly matters for anything that plugs directly into AC — a battery-powered ESP32 gadget largely sidesteps this category.

In the US, unintentional radiators (basically everything that isn't a deliberate transmitter) fall under FCC Part 15 Subpart B. In the EU, the relevant harmonized standards are EN 55032 (emissions) and EN 61000-4-x (immunity), rolled up under the EMC Directive. A one-off hobby build for yourself needs none of this. A product you sell, even in small numbers, generally does.

Pre-Compliance: Finding Problems Before the Lab Does

A full certification test at an accredited lab runs anywhere from several hundred to several thousand dollars per test cycle, and failing it means paying again after a fix. Pre-compliance testing on the bench won't give you a certificate, but it will catch the obvious failures — the ones that would otherwise cost you a wasted lab booking — for the price of equipment you may already own.

ToolWhat It Tells YouLimitations Near-field H-field/E-field probe setWhere on the board the noise is coming from — wave the loop probe over traces and ICs while watching a spectrum analyzer to localize hot spotsNear-field strength doesn't translate directly to far-field radiated emission level; it's a localization tool, not a pass/fail measurement Spectrum analyzer (tinySA Ultra, or a benchtop Rigol/Siglent)What frequencies are present and roughly how strong, paired with a probe or a basic antennaA tinySA's dynamic range and amplitude accuracy are well below lab-grade; treat its readings as relative, not absolute Current probe / clamp-on RF choke testWhether conducted noise is riding on a cable, by clamping a snap-on ferrite around it and watching for a changeQualitative only — confirms a cable is a culprit, doesn't quantify the conducted emission level A simple biconical/log-periodic antennaA rough stand-in for a real radiated-emissions antenna if you don't have a near-field probe kitNo calibration factor applied, and your test environment (a garage, not a chamber) adds its own reflections and ambient noise

A Practical Pre-Compliance Pass

1. Power the device from a representative supply and let it reach normal operating state — emissions during a boot transient are not representative of steady-state behavior
2. Sweep broadly first with a spectrum analyzer and an antenna or probe from roughly 30 MHz (the usual low end of radiated emissions limits) up through at least a few GHz if the device has any wireless radio in it, noting any peaks well above the general noise floor
3. Use a near-field probe to walk the board and localize each peak to a specific trace, connector, or component — crystal oscillators, switching regulator inductors, and any unshielded high-speed digital bus are the usual suspects
4. Try a fix (a ferrite, a grounded shield can, a snubber across a switching node, a slower slew rate on a gate driver) and re-sweep to confirm the peak actually dropped rather than just moved frequency
5. Pay particular attention to cable exits — a clean PCB with a noisy, unshielded cable acting as an antenna is one of the single most common real-world failures

When to Actually Get Certified

If you're selling a finished product in the US or EU, formal testing and a Declaration of Conformity (self-certified for most Part 15 Subpart B products, or via a Notified Body for certain EU categories) are generally required regardless of how clean your pre-compliance sweeps look — pre-compliance work reduces the number of expensive re-test cycles, it doesn't replace the test. Look for a local EMC test house that offers a "pre-scan" or "pre-compliance" service slot at a lower rate than full certification; many will let you bring a prototype in for a few hours on their real chamber hardware, which is worth doing once you've exhausted what a bench probe and a tinySA can tell you.

None of this requires exotic equipment to get started — a near-field probe set runs well under $50, and the spectrum analyzer half of the bench overlaps entirely with the one covered in this site's tinySA buying guide. What it requires is building the habit of sweeping a board before you ship it, the same way you'd bench-test a circuit's function before assuming the schematic was right.