Direct answer
Locate before you modify: establish whether it is an emission overrun or an immunity failure, which frequency band it sits in, and which path it escapes by (radiated through space, conducted along cables, or through the supply). Then fix it specifically — conducted problems call for filtering and grounding, radiated problems for shielding, routing and seams, immunity problems for TVS and filter networks. Modifying the board without locating first usually takes several rounds.
Three steps to locate the problem
- Read the data: which frequency points exceed the limit, and by how many dB. A single narrow spike and a broad raised band point to completely different causes.
- Separate the paths: disconnect the harness and see whether it changes, apply copper foil and see, fit a ferrite core and see. Whichever action moves the data is the path the problem travels.
- Find the source: clock circuits, switching supplies, high-speed interfaces, motor drives, relays. Sweep area by area with a near-field probe to locate the specific region.
These three steps require no board changes and can be finished in an hour or two, but they save several later rounds of blind modification.
Directions by failure type
Conducted emission overrun
- Add or adjust the common-mode choke and the X / Y capacitors at the input
- Check the ground return path and avoid long loops that act as antennas
- The switching frequency and edge rate of the power supply are common sources
- Place the DC-DC input filter capacitor close to the device pins
Radiated emission overrun
- A housing seam approaching half a wavelength becomes a slot antenna — shorten the seam or add conductive gasket
- Keep high-speed traces from crossing a split in the ground plane
- Guard or series-terminate clock lines
- Add common-mode chokes at the harness end to reduce the harness antenna effect
Radiated immunity failure
- Add filtering and shielding to sensitive signal lines
- Add RC or TVS on critical pins such as reset and interrupt
- Check ground plane integrity so sensitive circuits do not reference a bouncing ground
- Add watchdog and state-recovery mechanisms in software
ESD failure
- Add TVS at the interface, balancing clamping voltage against parasitic capacitance
- Keep the housing ground path short
- Buttons and connector seams are entry points for discharge — address them mechanically first
Transient conducted immunity failure (ISO 7637 type)
- Combine TVS and varistors at the supply input
- Watch the TVS response speed and energy handling capability
- Match the thresholds of the reset circuit and the supply monitor
Three low-cost verification techniques
- Near-field probe sweep: locate the region of the radiating source; cheap and fast
- Disconnect the harness section by section: tells you whether the emission comes from the harness or the enclosure
- Temporarily add ferrite cores or copper foil: verify the direction is right before committing to PCB changes
When to run a pre-compliance test
One pre-compliance round during development costs far less than re-working at the certification stage. It does not need the full item list — emissions plus a few key immunity items is enough, because the goal is to find directional problems early, not to obtain a certificate.
Many projects only discover the overrun at formal certification, by which point the enclosure, PCB and BOM are fixed and the cost of change is far higher.
Where we stand
TMC LAB holds both CMA and CNAS accreditation. Automotive EMC covers the emission items of CISPR 25 / GB 18655 and the immunity items of ISO 11452 / GB/T 33014, with pre-compliance testing during development and re-testing after corrective action. Send us the failing frequency points and the test conditions, and we will help judge the direction first.
Related pages for this question
Last updated Technical criteria follow the current standards and the agreed scope; please confirm the specific project with the lab.