EMC Troubleshooting at PCB Level Using Ansys SIwave

What are antenna-like structures on PCBs?

Everyone might have seen structures declared as “antennas”. Consider the classical rod antenna, you may see at some car roofs. These antennas have the function to receive high-frequent signals from the air, sent by any transmitters also via antennas, and route them to any further detecting and decoding unit. The rod itself works – together with other electronic components – as a resonating structure for high-frequent currents flowing through, resulting in a standing wave. The currents interact with the surrounding space being the source of electromagnetic waves or – speaking of a reciprocal system - incoming electromagnetic waves would induce corresponding currents. You may wonder, why you cannot find an antenna when looking at your smartphone, which is obviously a transceiver system for such waves. What determines the shape of an antenna?

Going back in days, antennas were large structures. Even the first mobiles had little rods at their side. This gives us a clue, that geometric dimension might influence the potential to radiate. Indeed, the antenna’s length is an important factor for its resonance frequency, where the antenna has the highest potential to radiate. As a rule of thumb, we can say the longer the antenna is, the lower is its first resonance frequency. But there are other factors as well, like inductances in series with the antenna arms or capacitances in parallel. Antenna engineers effectively use these additional components to tune antennas and to make their geometric size smaller for integration into devices. The antenna has not necessarily to look like a rod but can also be a folded conductor structure like a copper trace or plane on a PCB with well-defined shape.

This is the crucial point when coming to PCB layouts. You find hundreds of conductor traces, metallic planes, capacitors, and inductors on modern PCBs. Why should not any of them be part of an antenna structure at any frequency in the wide frequency range relevant in EMC standards?

This resonator is nothing else than an antenna structure with tuning components, where the planes take the function of carrying the high-frequent currents as radiation sources, similar to the rod described in the last section. The resonance frequency of a parallel resonator can roughly be estimated to be f_res=1 ⁄ (2π√LC)). Due to this relation, the power and ground planes with their large L and C values are most critical to have their resonances at low frequencies in the MHz- to low GHz-range, where many EMC standards operate.

When measuring the input impedance Z_in of such a parallel resonator, one might see the low value of R_Source for “zero frequency” (DC). The impedance will increase going up in frequency f due to the inductance
(Z_ind~fL) and will decrease above the resonance frequency due to the capacitance (Z_cap~1/(fC)) In effect, at resonance, a peak value can be seen in the impedance profile. Figure 3 shows a typical impedance profile, measured between a supply and the GND plane of a PCB at two different positions. The position of impedance measurement is relevant to the height of the impedance peak to be seen, but not on the resonance frequency itself. The reason for this is the spatial current and voltage distribution on the planes due to the standing wave in this resonant system.

The resonant modes only show the potential to radiate, or – in other words – they identify potential antennas. For radiation, these antennas must be driven by signals. There have to be coupling structures to these antennas on the one hand, and on the other hand, these coupling structures have to carry signals containing spectral components around resonance frequencies. Coupling structures could be for example transmission lines routed nearby to the resonant planes or vias routed through the resonating planes. The spectral components driving the antennas can usually be found in any pulsed signals (e.g., bit-pattern or clock-signals), due to their short rise- and fall times. That’s why the preferred optimization approach should be to eliminate the root-cause of EMC issues, which are the antenna structures in layouts.

Figure 5 shows the impact of both variants (detuning and dampening) on the impedance profile. In case of detuning (blue curve), the corresponding peaks get shifted to higher frequencies. The whole profile changes due to the different effect on different resonances and due to changing their voltage and current distributions measured by ports. In case of dampening by a resistive value (red curve), the peaks get flattened slightly, but the peaks do not change their frequencies.

Figure 6 shows the simulated electric field strength in far-field, caused by resonances at the shown PCB. The resonances are driven with 1 Volt at each frequency in the simulation. There can be seen some peaks in field strength. These peaks lay around the simulated resonance frequencies that can be seen from Figure 3. Especially the two modes above 1 GHz strongly contribute to radiated emissions. With the quick optimization approach, the observed field strength can be significantly reduced.

In EMC test standards there are limits specified for this electric field strength, measured by an antenna in a certain distance depending on the standard. The specific values of these frequency dependent limits also vary with the standards. There are standards defined for household devices [EMCSTD] or consumer electronics as well as for industrial or medical equipment [CISPR]. What they all have in common is, that peaks like the ones can be seen from Figure 6 make EMC tests defined there usually fail.

The shown optimization approach by using simulation helps getting rid of the resonances and so of peaks in the EMC measurement. Using simulations like the resonant mode analysis synchronously to your layouting process makes you pass the EMC testing and lets you avoid cost and time intensive remanufacturing cycles. Regarding EMC, you may follow the motto “prevention is better than cure” in your design process.