What to Check When Radiated Emissions Exceed the Limit

When a radiated-emission test goes over the limit, the urge is to swap parts right away, but following a sequence cuts the number of fix iterations a lot. Here is the flow used in practice.

Step 1. Verify the measurement first

• Check the conversion: E = receiver reading + antenna factor + cable loss + attenuator − preamp gain. A single wrong sign shifts the result by several dB. Confirm that the converted value is the same with the preamp on and off. (Calculator: Measured voltage → field strength)
• Check ambient noise: scan the same setup with the EUT off to separate broadcast and communication signals.
• Polarisation and antenna height: confirm that the maximum over both polarisations and a 1–4 m height scan was used.
• EUT mode: confirm it was tested in the worst-case mode (maximum load, highest clock, all ports active).

Step 2. Infer the cause from the frequency pattern

• Evenly spaced peaks are clock harmonics (f = n × f_clk). A 25 MHz clock gives peaks every 25 MHz.
• If harmonics of a 100 kHz–few MHz switching frequency extend to tens of MHz, suspect the switching supply. A broad hump across tens to hundreds of MHz is often ringing on switching edges.
• A large gap between peak and average readings indicates intermittent (burst) signals; a small gap indicates a continuous narrowband signal.
• If a frequency matches a structural length, suspect resonance. Compute λ/2 and λ/4 for cables and slots. (Calculator: Frequency ↔ wavelength)

Step 3. Check whether the cable acts as an antenna

A large share of radiated failures come from common-mode current on cables. If moving the cable or clamping a ferrite on it changes the reading noticeably, the cable is the radiator. Then look for where common-mode current is generated inside the EUT (ground discontinuity, missing filter).

Step 4. Trace the source with a near-field probe

Sweeping a near-field probe tuned to the problem frequency over the board and enclosure finds the strongest location. Whether it is near the oscillator, the power stage or a connector decides the fix.

Step 5. Countermeasures: source, then path, then last resort

• Source suppression: series resistors to slow rise time, spread-spectrum clocking, snubbers, smaller switching-loop area. Cheapest and most durable. (Calculator: Rise time → bandwidth)
• Path blocking: I/O filters, common-mode chokes, decoupling, continuous ground plane, 360° cable-shield termination.
• Coupling suppression and shielding: reduce aperture leakage (keep slots below λ/20), gaskets, shield cans. The last resort.

Step 6. Re-measure and record

Apply one change at a time and re-measure under identical conditions, recording the improvement in dB. Knowing which change gave how many dB lets you reuse the knowledge on the next product and supports the test report. Finally confirm that a margin of at least 3–6 dB to the limit is secured.