Any nonlinear component can generate EMI through harmonic generation, both during switching action when pulsed, or when driven to saturation with an AC signal. The two most common nonlinear components where EMI can appear on a simple AC signal, either through switching or saturation/rectification, are diodes and FETs. Switching action, nonlinear saturation, or both will lead to conducted and radiated EMI.
Nonlinearity Generates EMI
Any component with a nonlinear transfer characteristic will generate harmonics when it is driven with an AC signal at a single tone. Whenever a component has a nonlinear transfer characteristic, either through rectification or saturation, the transfer curve (output voltage/current vs. input voltage/current) can be approximated as a Taylor function around some nominal voltage:
The function f(V) is the transfer characteristic, while the nominal voltage at which we consider AC voltage swings could be a DC offset, an average voltage, or 0 V. The asterisks on f(V) indicate derivatives. It is the higher degree terms (n > 1) that generate all harmonics of the input signal through self-mixing, and the result is the appearance of harmonics in a radiated or conducted emissions measurement.
The EMI that is generated could be radiated or conducted (or both), and generally the power spectrum can be similar due to the manner in which harmonics are generated through saturation and switching action. The table below outlines the common instances where FETs and diodes can generate EMI through harmonic generation.
Power regulator circuit in an SMPS
More generally in analog/power electronics, where the output exceeds the load line limits
AC/DC input rectification stage on power converters, rectified output from transformer-coupled SMPS circuits, and reverse current protected circuits
The typical situation in a diode is that an input waveform is being rectified such that it becomes monodirectional. All diodes are nonlinear components, and due to their exponential transfer function, they will generate harmonic distortion in the input wave. Conducted emissions will occur as the diode is switched from forward to reverse bias by a driving AC waveform. As the AC waveform changes polarity, the diode requires some time to switch from forward to reverse bias due to discharging of its junction capacitance.
One of the important specifications that determines the emissions is the reverse recovery time, and the need for fast or slow reverse recovery will vary depending on who you ask. During the recovery phase, a fast transient response can be excited, and this could exhibit ringing if there is excess inductance along the current path.
The expected behavior of a diode when forward bias ends and reverse recovery begins is shown in the following image. When the reverse recovery phase begins at time T, the current drops at some rate (di/dt) that can excite a transient along the current path. The result is conductive EMI observed at the load, and radiated EMI observed around the circuit.
Unfortunately, there is nothing you can manipulate in a diode that would allow the diode to operate differently in terms of its EMI generating characteristics. The solution is to add to the circuit that involves a diode or to select a different diode that will generate less EMI. There are two broad cases where a fast recovery diode vs. Schottky diode should be used:
- Diode driven unidirectionally with pulses: Slow recovery might be preferred as this can slow down the pulse driven through the diode.
- Diode array used for rectification with AC wave: Fast recovery is preferred as this prevents a fast edge from arising as the diodes switch from forward to reverse bias.
In short, if the diode will quickly encounter the reverse polarity portion of an AC wave, it is best to use a fast recovery diode. If the diode will only be driven in forward bias with pulses, then a slow recovery diode is acceptable.
Saturation in a MOSFET will produce harmonics in the same way as a diode. Saturation causes clipping of the output signal in the time domain, which requires that the clipped signal contain additional harmonics that were not present on the input. This is all visualized in the load line diagram below.
Conducted EMI will arise when the MOSFET is run at the extreme edges of the load line. If there is excessive harmonic content in this case, the transistor could be swapped or the operating voltages can be adjusted.
In the case of fast switching, EMI can be conducted and radiated due to the fast edges of the driving pulse and the current delivered to the load component. The inductances in the MOSFET package, parasitic capacitances, and the inductive loop formed by the current path all contribute to generating EMI. This is one reason switching noise can be so intense in switching DC/DC regulators.
The mechanism of EMI generation in this case is through excitation of a transient response. In switching FETs, we normally opt for fast response and high efficiency (low DC loss), which causes the transient response to be underdamped in almost every case. The result is a ringing waveform that can be measured on the output from the switching circuit, or it can be measured as radiated emissions in the near field regime.
Switching node waveform in a typical buck regulator circuit.
The solution in this case is to modify the PCB layout so that inductance in the current path is minimized. In the case where package parasitics are excessive, inductance at the switching node can be reduced by adding a small amount of resistance, such as with a snubber to ground.
Don’t take chances with EMI in your PCB, use the CAD tools in OrCAD from Cadence to build your PCB with low EMI. OrCAD is the industry’s best PCB design and analysis software with utilities covering schematic capture, PCB layout and routing, and manufacturing. OrCAD users can access a complete set of schematic capture features, mixed-signal simulations in PSpice, and powerful CAD features, and much more.