3.3 Emissions through galvanic coupling, electric and magnetic near field coupling to cables, long nets, and mechanical structures, interpreted as antennas

Generally all emission mechanisms in this section are common impedance coupling mechanisms. Figure 3.3 illustrates the common impedance coupling mechanisms of a source circuit to a victim circuit. A common resistor R_k couples the two circuits in Figure 3.3(a) galvanically. The coupling capacitance C_k in Figure 3.3(b) couples a noise voltage to a victim circuit. Inductive coupling of a source loop and a victim loop is depicted in Figure 3.3(c). The right circuit diagram in Figure 3.3(c) is equivalent to the left diagram. It illustrates the possibility to consider the inductive coupling with a coupling inductance L_k. The victim circuit in Figure 3.3 is a sensitive sensor circuit and the measurement of the sensor voltage is denoted by U_k. However, the victim circuit may also be any other circuit on the PCB. Extended circuits on the PCB, circuits with attached cables, or sizeable daughter boards are the unintentional emission antennas which are supplied from the coupled currents and voltages. When the antenna of the victim circuit is located far away from the coupling fields, the whole coupled field will not change significantly, even when the antenna is replaced by a different one. Therefore, the coupling impedances enable the description of the source coupling to an antenna, independent of the antenna. This provides the opportunity to classify the PCB structures regarding their emissions relevance with only the coupling impedances and without considering the antenna structure. For example, the consideration of a cable attached to a PCB in an emission simulation will also show the cable resonances. However, when this cable cannot be modified by the PCB designer and, especially, when the PCB should be connected to different cables in different applications, a simulation result including these resonances is misleading. The PCB emission optimization can be performed more efficiently by omitting the cable, just by simulating the coupling impedances. A separate simulation of the antenna can be used for quantitative prediction simulations. According to [62], the maximum radiated far field from cables can be estimated roughly with a simple line resonator model. The estimated maximum electric fields from monopol and dipole antennas at their resonance are [62]

$$E_{max}^{res}\approx(60\Omega)\cdot\frac{I_{ant}}{r}\quad \Longle... ...t \begin{cases} 1.64&\text{(monopol.)}\\ 0.82& \text{(dipol.)} \end{cases}$$ (3.4)

U_ant and I_ant are the antenna voltage and the antenna current, respectively, and r is the distance of the field observation point from the antenna position.

(a) Galvanic coupling.

(b) Capacitive Coupling.

(c) Inductive Coupling.

Figure 3.3: The coupling mechanisms between an emission source circuit and another circuit. The dotted frame marks the source circuit branch.

The coupling can be interpreted as an unwanted common mode coupling. This is shown in Figure 3.3(a), where the current through the victim circuit is denoted as the common mode current I_cm, which flows in the same direction as the current in the source path I_noise. The voltage and the current driven common mode coupling from traces on the PCB to attached cables, as described in [40], are also common impedance couplings. For the current driven mode the common mode inductance of a trace has been formulated by [35], [59] and for the voltage driven mode the common mode capacitance by [38], [39], both independent of a cable attached to the PCB. Section 5.7 links the field coupling of PCB sources to the cavity field between the PCB ground plane and a parallel metallic cover plane to the common mode coupling described in [35].

In the following an example for common impedance coupling in a power delivery network is presented. Every real power supply has a nonzero impedance. Thus, currents from one device cause a voltage noise on that impedance, which is conducted to other devices connected to the same power supply. An example is an automotive control device connected to the board power net which also supplies many other electronic devices. Another example is a three-phase converter for an electric drive, supplied from a transformer station which might supply a whole village. Figure 3.4 depicts a push-pull switching stage, supplied from a battery, which also supplies a sensitive sensor circuit. Power supply noise generated from the switching circuit couples to the sensitive circuit through a non zero resistive impedance of the power supply R1 and an inductive coupling of the loops considered with a coupling factor K1 which partly couples the inductance of the source path L3 and the inductance of the victim path L6. As an alternative to the coupling factor, one could also consider the inductive coupling with an inductance in series to the resistor R1.

Figure 3.4: Example for conducted emissions. The supply noise of a push-pull stage is conducted to a sensitive sensor circuit, connected to the same power supply. Simulation setup in LTSpice [63].

Note that common impedance coupling occurs not only on the plus branch of the supply. Common impedances are also in the ground branch. Low impedance grounding and separate ground routing to a star point are measures to minimize the coupling in the ground branch. The supply in Figure 3.4 is decoupled with low inductive capacitances, both in the source and also in the victim path. Low inductance can be achieved by capacitors with low equivalent series inductance (ESL), or by a parallel connection of multiple capacitors with the same capacitance value. The simulation of the model was carried out with LTSpice [63]. After a time domain transient analysis a FFT is performed with Barlett-Hann windowing to obtain the frequency domain power supply noise. Figure 3.5 depicts the simulation result for the supply noise at the sensor supply connection and the supply connection of the switching circuit. Although decoupling significantly reduces the supply noise at the sensitive circuit, emission can be observed. Further noise reduction at the sensitive circuit requires additional decoupling and filtering.

Figure 3.5: Simulation result of the model in Figure. 3.4. Common impedance coupling at a shared power supply causes conducted power noise at a sensitive circuit.

Like in this example, standard network simulation is generally performed to describe conducted emissions from common impedance coupling. The circuit model has to consider the parasitics of the power delivery network, the models of filter and decoupling components, and models for the emission sources. Noise sources are all current switching components, such as, inverters for AC motors, switched power supplies, or core noise generated from a microcontroller. Accurate models of these sources are necessary for the emission simulation, because even a slight change of the transient current shape may change the spectrum significantly. Standardized ICEM models are used for the accurate simulation of the supply noise from integrated circuits [60], [61]. An accurate and efficient EMC modeling technique for discrete components was presented in [64]. The passive power delivery network on the PCB can be extracted, for example, with the methods in [42], [43], [65]. Models for power cabling can be obtained for uniform cabling with transmission line theory and for non-uniform cables with three-dimensional simulation, for example, with the method in [30].

The next part contains the development of a cavity model for the efficient simulation of the emissions from a printed circuit board under a metallic enclosure cover. The model enables the explicit calculation of the common mode coupling impedances from printed circuit board structures to the interface at the apertures of the enclosure. This provides the opportunity to optimize the interior of the device independent of the external environment.

C. Poschalko: The Simulation of Emission from Printed Circuit Boards under a Metallic Cover