Why Early EMI Testing Saves Time and Money
Failing an EMC compliance test (FCC, CE, etc.) at the end of product development is expensive and demoralizing. Board respins, additional shielding, and last-minute component changes are all costs that can be largely avoided by doing informal pre-compliance EMI scanning during development. You don't need a shielded anechoic chamber — a basic spectrum analyzer with near-field probes gives you actionable data at your desk.
What You'll Need
- Spectrum analyzer: A handheld or benchtop unit covering the relevant frequency range. For most digital electronics, covering 30 MHz to 1 GHz is a minimum; 3 GHz is better. Budget USB-based spectrum analyzers (such as the TinySA Ultra or similar) can be surprisingly useful for pre-compliance work.
- Near-field probe set: A set of small loop (H-field) and electric (E-field) probes that connect to the spectrum analyzer. These are available from several manufacturers or can be hand-wound from semi-rigid coax.
- Low-noise preamplifier (optional but recommended): Adds 20–30 dB of gain to improve sensitivity when scanning for weak emissions.
- The PCB under test: Running in its normal operating condition, powered up and executing its typical software workload.
Step 1: Set Up the Spectrum Analyzer
- Connect the H-field probe to the spectrum analyzer input.
- Set the frequency span to cover your expected problem area — start with 30 MHz to 500 MHz for most digital designs.
- Set the Resolution Bandwidth (RBW) to 100 kHz as a starting point. Narrower RBW gives more sensitivity but slower sweeps.
- Enable Max Hold mode so you can move the probe slowly without missing peaks.
- Set a reference level appropriate for expected signal levels (start at 0 dBm, adjust down as needed).
Step 2: Identify Emission Peaks
With Max Hold enabled, slowly sweep the probe approximately 5–10 mm above the PCB surface in a grid pattern. Watch the spectrum analyzer display for peaks that rise above the noise floor. Key things to observe:
- Harmonics: Emissions often appear as a series of spikes at regular frequency intervals (e.g., every 100 MHz if the clock is 100 MHz). This pattern immediately identifies the clock frequency as the source.
- Broadband noise: Wide, elevated noise floor often indicates switching power supply noise or fast-edge digital signals with poor layout.
- Isolated spikes: May indicate resonance in a particular trace, via, or connector.
Step 3: Localize the Source
Once you've identified peaks of interest on the spectrum, switch to a narrow frequency span centered on the problematic frequency and use the probe to scan for the location of maximum emission. The signal will peak when the probe is directly over the source.
Common emission sources include:
- Clock oscillators and their traces
- Switching power supply inductors and their switching nodes
- High-speed bus traces (USB, Ethernet, DDR memory)
- Crystal oscillator circuits
- Poorly decoupled IC power pins
Step 4: Diagnose and Fix
Once you know where the emission is coming from, identifying the fix becomes much more systematic:
| Source Identified | Likely Cause | Recommended Fix |
|---|---|---|
| Clock trace | No series termination, long trace | Add series resistor (22–33 Ω) near driver; shorten trace |
| Switching inductor | High dV/dt switching node exposed | Add snubber; reduce switching node copper area |
| IC power pin | Missing or misplaced decoupling cap | Add 100 nF cap within 1–2 mm of power pin with short traces |
| Connector or cable | Common-mode noise on cable | Add common-mode choke at connector entry point |
Step 5: Verify the Fix
After each modification, re-scan the area with the probe and compare to your earlier Max Hold trace. Effective fixes will show a measurable reduction in the emission peak. Document each change so you can track what worked and what didn't — this builds valuable institutional knowledge for future designs.
Important Limitations
Near-field scanning at the bench is a development tool, not a substitute for formal radiated emissions testing in a certified facility. Near-field measurements tell you about relative field strength close to the board; they don't directly correspond to far-field radiated emissions levels that regulatory tests measure. However, reducing near-field emissions consistently translates to reduced radiated emissions, making this approach a highly effective development strategy.