How Do You Conduct EMC Pre-Compliance Testing?

Detailed Explanation

Pre-compliance testing is one of the highest-leverage investments in a hardware product development cycle. A formal EMC test failure discovered at the certification stage can cost AUD $5,000–$15,000 in additional lab time, delays, and PCB re-spin costs. Pre-compliance testing conducted during prototype development identifies the same failures for a fraction of that cost.

What Pre-Compliance Testing Is (and Isn't)

Pre-compliance testing is informal, in-house or third-party scanning of a prototype to identify the dominant electromagnetic emission sources before submitting to an accredited test laboratory for formal RCM certification. It does not:

• Replace formal testing at an accredited laboratory.
• Provide any certification documentation.
• Use calibrated test setups with the same precision as a formal test.

What it does:

• Identify the dominant frequencies and source locations responsible for emission peaks.
• Allow iterative debugging on the prototype: measure, fix, remeasure.
• Provide confidence that the design will pass formal testing (or won't, and why).

Most experienced EMC engineers estimate that pre-compliance testing improves the first-pass formal test pass rate from around 40–50% (for designs without pre-compliance work) to 80–90%.

Equipment Required

Spectrum analyser: The primary instrument. Displays emission amplitude (dBµV) versus frequency. A calibrated laboratory analyser is ideal, but low-cost instruments are viable for pre-compliance:

• TinySA Ultra (AUD ~$150): 100 kHz–5.3 GHz range; adequate dynamic range for near-field scanning. Not calibrated to CISPR limits but sufficient to locate dominant peaks and compare before/after fixes.
• Rigol DSA815 (AUD ~$1,200): 9 kHz–1.5 GHz; a step up in performance and reliability. Includes a quasi-peak detector, which more closely approximates CISPR measurement methodology.
• Professional lab analysers (R&S, Keysight, Agilent): reference standard; expensive (AUD $20,000+). Use at an EMC test lab.

Near-field probe set: Used to locate emission sources on the PCB by scanning the probe close to the surface. A near-field probe set typically contains:

• H-field loop probes (several sizes): sensitive to magnetic field (current loops, switching inductors, bypass capacitor current paths). Most useful for identifying SMPS and high-current traces.
• E-field monopole probe: sensitive to electric field (voltage nodes — switch node copper, fast-edge digital traces).

Commercial sets: Tekbox TBOH01 (AUD $180–$250) or Fischer Custom Communications probes. DIY: a small loop made from a 10 cm length of semi-rigid coax with the outer conductor stripped back and the centre conductor formed into a 5–10 mm loop, terminated into a 50 Ω SMA connector, provides a reasonable H-field probe.

LISN (Line Impedance Stabilisation Network): Required for conducted emissions testing (150 kHz–30 MHz via mains supply). A LISN provides a standardised 50 Ω measurement impedance at the mains port of the equipment under test. Commercial options: Tekbox TBLC08 or TBOH01 series. Required only for products that connect to mains power; battery-powered designs without mains input do not need conducted emissions testing.

Reference ground plane: A large copper-clad FR4 board (minimum 30 cm × 30 cm) placed under the EUT (equipment under test) during radiated scanning. Provides a local ground return and standardises the measurement environment.

Radiated Emissions Scan Procedure

1. Configure the spectrum analyser:

• Set frequency range to 30 MHz–1 GHz (the CISPR 32 radiated emissions range).
• Set RBW (resolution bandwidth) to 120 kHz for peak search (matching the CISPR quasi-peak detector bandwidth at these frequencies).
• Enable Max Hold to accumulate the worst-case emissions over a 30–60 second scan period.

2. Set up the EUT:

• Place the PCB on the ground plane reference board.
• Connect only the minimum cables required for normal operation; additional cables add cable antenna resonances.
• Power the board and run it in its most active operating state (highest clock frequency, highest processor utilisation, radios transmitting).

3. Perform the scan:

• Hold the H-field probe 5–10 mm above the PCB surface and slowly scan across the entire board while watching the spectrum analyser.
• Note any peaks that stand out above the average noise floor.
• For each significant peak, move the probe systematically over: the MCU and crystal area, the SMPS section (inductor, switch node, input bypass capacitor), and any connectors or cable entry points.
• The location that produces the highest amplitude when the probe is brought close is the dominant source for that frequency.

4. Correlate peaks to circuit blocks: Identify the source of each peak by checking:

• Is the frequency a harmonic of the MCU system clock? (SYSCLK = 168 MHz → harmonics at 168, 336, 504 MHz)
• Is the frequency a harmonic of the crystal oscillator fundamental? (8 MHz → 8, 16, 24, 32 MHz...)
• Is the frequency a harmonic of the SMPS switching frequency? (500 kHz → 500 kHz, 1 MHz, 1.5 MHz...)
• Is the frequency related to interface signals (USB 12 MHz, 96 MHz; Ethernet 25 MHz, 125 MHz)?

Conducted Emissions Scan Procedure

Conducted emissions testing applies to mains-connected equipment. The LISN is inserted between the mains supply and the EUT:

1. Connect mains to the LISN input; EUT mains lead to the LISN EUT port.
2. Connect the LISN measurement port to the spectrum analyser via 50 Ω coaxial cable.
3. Set spectrum analyser to 150 kHz–30 MHz; RBW 9 kHz for most of the range.
4. Run the EUT and log the peak emissions at each mains conductor (line and neutral, tested separately).

Switching power supply harmonics dominate conducted emissions. The SMPS switching frequency and its harmonics appear as narrow peaks; broadband noise from the switch node transition appears as a raised noise floor. Fix priorities are the same as for radiated: switching loop area, input EMI filter (common-mode choke + X and Y capacitors).

Comparing Results Against CISPR 32 Limits

CISPR 32 defines Class A (commercial/industrial) and Class B (domestic) radiated emissions limits at a 3 m measurement distance. CISPR 32 applies to most commercial multimedia and IT equipment — if you are unsure whether your product falls under CISPR 32, IEC 61000-6-4 for industrial equipment, or another Australian standard, confirm this before planning your test program; see which Australian EMC standard applies to your product. Pre-compliance scanning at 30 cm requires a distance correction: add approximately 20 dB to account for the 10× difference in distance (field decreases as 1/r in the far field). At 1 m, add approximately 10 dB.

This correction is approximate — near-field to far-field transition, reflection, and polarisation effects all introduce uncertainty. Pre-compliance results should be treated as directional: a peak that's 15 dB over the scaled limit needs attention; a peak that barely scrapes above is worth investigating but may pass at the formal test.

For product design decisions and EMC certification strategy aligned with ACMA's RCM framework, Zeus Design's engineering team handles pre-compliance scanning, PCB layout review, and formal test coordination — contact Zeus Design to discuss your project's compliance path.

Design Considerations

• Perform pre-compliance testing early in the prototype phase, not at the end. The value of pre-compliance testing decreases as the design hardens. At prototype stage, a 10 dB improvement from rerouting the switching loop costs nothing; the same improvement after the board is finalised may require a re-spin or an expensive shielding solution.
• Test with the real operating configuration. Radiated emissions are sensitive to cable routing, connector positions, and software state (processor activity level, peripherals active). Test with all I/O cables attached and the MCU running at full utilisation. A passing pre-compliance scan with a stripped-down configuration and no cables can fail formal testing with cables attached.
• Use Max Hold for at least 30 seconds per scan pass. Intermittent emissions (from software loops, DMA transfers, radio bursts) won't appear in a single-sweep scan. Max Hold accumulates the worst-case over time.
• Keep a baseline scan with the EUT powered off. This identifies the ambient RF environment (Wi-Fi, mobile phones, broadcast transmitters). Subtract the ambient from the EUT scan to identify which peaks are actually from the board.

Common Mistakes

• Scanning without running the MCU at full load. The peak emissions of a microcontroller are at full CPU utilisation (all pipelines active, high clock bus activity). Scanning with the MCU idle or in a low-power mode gives an optimistic result — the dominant peaks only appear when the system is active.
• Forgetting that cables are antennas. A substantial fraction of radiated emissions in products that fail formal testing comes from current flowing on the outside of attached cables, not from the PCB itself. Near-field probe scanning of the PCB will miss these emissions entirely; probe the cables directly and check for current flowing on the cable outer shield.
• Treating a passing pre-compliance result as a guarantee. Pre-compliance testing with a non-calibrated instrument, a non-standard measurement distance, and no antenna substitution cannot guarantee formal test results. It substantially improves your odds, but the formal test under controlled conditions will always differ. Build in margin — if the pre-compliance result is 3 dB under the scaled limit, fix it before the formal test.
• Fixing symptoms rather than root causes. A ferrite bead on the power supply line suppresses the symptom (conducted current on the supply) but doesn't fix the underlying cause (large switching loop radiating directly). Always identify the dominant source first; add filtering only after fixing the layout.