Measuring electromagnetic interference (EMI) means capturing the unwanted electrical noise a device produces and comparing it against regulatory limits. There are two fundamental types to measure: conducted emissions, which travel along power cables, and radiated emissions, which broadcast through the air. Each requires different equipment, different setups, and different environments.
Conducted vs. Radiated Emissions
Conducted emissions are noise signals that a device pushes back onto its power cord and into the electrical grid. These are measured along the wire, in the frequency range of 150 kHz to 30 MHz. Radiated emissions are signals that escape from the device or its cables into the surrounding air, typically measured from 30 MHz and up. Both are expressed in decibels relative to one microvolt (dBμV) for conducted measurements, or in volts per meter (V/m) for radiated measurements.
You need to measure both types separately because they travel through different paths and require completely different test setups. A device can pass one test easily and fail the other.
Equipment You Need
The core instrument for any EMI measurement is a spectrum analyzer or an EMI receiver. This displays signal strength across a range of frequencies, letting you see exactly where your device is producing noise and how strong it is. A standard oscilloscope with an FFT (fast Fourier transform) function can substitute for early-stage testing, though it lacks the precision of a dedicated receiver.
For conducted emissions, you also need a Line Impedance Stabilization Network, or LISN. This is a box that sits between your device and its power source. It does two things: it presents a standardized 50-ohm impedance to your device (so measurements are repeatable regardless of what power outlet you use), and it isolates the device from noise already present on the power line. A typical LISN contains a 50 μH air-core inductor in series with each power conductor, plus capacitor and resistor networks that create that defined impedance.
For radiated emissions, you need antennas suited to the frequency range you’re testing. Biconical antennas cover lower frequencies, while log-periodic or horn antennas handle higher bands. You’ll also need a turntable for your device and a mast that raises and lowers the antenna to find the angle of maximum emission.
How to Measure Conducted Emissions
Connect the LISN between your device and its power supply. Run a coaxial cable from the LISN’s measurement output to your spectrum analyzer or oscilloscope, making sure the input is set to 50-ohm impedance for proper signal matching.
Before powering up your device, configure the analyzer. Set the frequency range to 150 kHz through 30 MHz. Choose your resolution bandwidth and detector type (more on detector types below). Then take a baseline measurement with the device powered off. This captures the noise floor of your test setup, so you can distinguish your device’s actual emissions from background noise. Power on the device and take your measurement. You need to measure between each power line conductor and ground separately, checking both the line and neutral connections.
Compare your results to the applicable limits. Under FCC Part 15, a Class B device (intended for residential use) must stay below 66 dBμV quasi-peak at 150 kHz, decreasing to 56 dBμV at 500 kHz, then holding at 56 dBμV up to 5 MHz, and 60 dBμV from 5 to 30 MHz. Average limits run about 10 dB lower across each band. Class A devices (commercial environments only) get more generous limits, with quasi-peak values of 79 dBμV below 500 kHz and 73 dBμV from 500 kHz to 30 MHz.
How to Measure Radiated Emissions
Radiated emission testing is more complex because you need to control the environment. The standard setup places an antenna at a fixed distance from the device, typically 3 meters for many product categories under FCC Part 15 and CISPR 32. Some standards specify 10-meter distances. The device sits on a turntable that rotates through 360 degrees, while the receiving antenna scans vertically to find the orientation that produces the strongest signal. You’re looking for the worst-case emission, so you systematically sweep through positions and polarizations.
The environment matters enormously. Any reflective surface, nearby electronics, or radio signals in the area will corrupt your measurement.
Where to Measure: Test Environments
The gold standard for radiated testing is an Open Area Test Site (OATS), an outdoor location with a flat, conductive ground plane and no reflective objects nearby. OATS works because the only reflection comes from the ground plane, which is predictable and accounted for in the measurement method (the antenna is height-scanned to find the maximum signal, capturing both direct and reflected waves). The downside is vulnerability to ambient radio signals from cell towers, broadcast stations, and other electronics, plus weather exposure. Ground plane reflections can cause measurement variations of about 5 dB.
A semi-anechoic chamber solves the ambient noise problem by enclosing the test area in a shielded room lined with radio-absorbing material on the walls and ceiling. The floor remains reflective to replicate OATS ground-plane conditions. These chambers produce much smoother, more reliable results than unlined shielded rooms, where wall reflections can boost signals by as much as 15 dB and distort measurements badly. The tradeoff is cost: a modest OATS might run tens of thousands of dollars, while a large anechoic chamber can cost millions.
Understanding Detector Modes
Your spectrum analyzer offers three detector modes, and regulatory limits are defined against specific ones, so choosing correctly is essential.
- Peak detection captures the highest amplitude at each frequency. It represents the absolute worst case and is the fastest way to scan. If your device passes peak limits, it will pass the other detector modes too, making peak a useful quick screening tool.
- Quasi-peak detection weights signals based on how often they repeat. A noise spike that occurs frequently gets a higher reading than one that fires rarely. This is the primary detector mode referenced in most FCC and CISPR conducted emission limits.
- Average detection reports the mean amplitude of each signal component over time. Average limits are always lower than quasi-peak limits (typically 10 dB lower), so a signal that passes quasi-peak can still fail on average.
A practical approach is to scan first with the peak detector, which is fast. If peak readings fall below the quasi-peak limits, you’re compliant and don’t need to run the slower quasi-peak or average scans. If peak readings exceed the limits, switch to quasi-peak and average to see if the actual weighted measurements still pass.
Using Near-Field Probes to Find EMI Sources
When your device fails a conducted or radiated test, you need to find which component or trace is generating the noise. This is where near-field probes come in. These are small, handheld probes that you move close to your circuit board to map where electromagnetic fields are strongest.
Magnetic field probes (small wire loops) detect current flow and are useful for finding noisy traces, switching regulator inductors, and current loops on a board. Electric field probes (open-ended conductors without a loop) detect voltage changes and help identify high-impedance noise sources. By slowly scanning across a board while watching the spectrum analyzer, you can pinpoint the exact component or area responsible for a failing frequency. This technique doesn’t give you calibrated emission levels, but it shows you where to focus your filtering and shielding efforts.
Pre-Compliance vs. Formal Testing
Full compliance testing in an accredited lab is expensive and time-consuming. Pre-compliance testing at your own bench lets you catch problems early, before committing to a formal test. For conducted emissions, a LISN and a spectrum analyzer or FFT-capable oscilloscope give you a reasonable approximation of what you’ll see in a real lab. For radiated emissions, near-field probing and simple antenna measurements in a quiet area can reveal obvious problems, though they won’t replicate the precision of a chamber measurement.
The key to useful pre-compliance work is establishing your noise floor. Always measure with the device off first, then compare that baseline to your powered measurement. Any signal that doesn’t rise meaningfully above the noise floor isn’t coming from your device. Signals that sit close to regulatory limits in a pre-compliance setup will likely be borderline in formal testing too, so treat them as problems worth fixing before you book lab time.