A common misconception in the industry is that if all the components designed into a system pass different agency standards for radiated, conducted, and common-mode noise, then the system will also pass the required noise standards. Wrong! This lesson applies in particular to dc-dc converters. Just because the converter passes agency approvals for noise doesn't mean that the system it's used in will also pass.
First, let's brush up on the basics. Noise in a power converter is either conducted, radiated, or common-mode. Conducted noise is observed on the conductors that connect the converter to the input power source, to the load, or to any control signals. Radiated noise is the electromagnetic noise radiated from the converter due to the switching of voltage, current, or magnetic flux. Common-mode noise is common to the input and output of the converter due to parasitic elements like input-to-output capacitance, leakage inductance, and isolation resistance.
With ever increasing switching frequencies and power densities, the detrimental effects of the parasitics on the power converter have driven power designers to implement resonance converters and other designs that make parasitic elements part of the circuit. Switching during zero voltage or zero current in the input switching transistor reduces input noise and improves efficiency. Converters with zero-voltage or zero-current switching modulate the switching frequency to maintain regulation. Thus, they require large input filters.
Cuk's converter employs integrated magnetics that use magnetic coupling to reduce or eliminate input and output noise. Phase modulation and other topologies use parasitic elements to achieve resonance while holding the switching frequency constant. Most often, the operating conditions in a system differ from those of the test labs that approved the components. Switching loads and poor system layout can make even certified power converters look bad. Even if a system with a poor layout works, it may not pass final noise specifications without the addition of large, costly filter components. Frequently, even the added components won't do the job.
System designers looking for help can turn to the dc-dc converter manufacturer, who may be expected to provide information on power converter topologies as well as marketing jargon and sales pitches. However, the converter manufacturer can also offer the relevant standards information as it relates to getting agency approvals. This is particularly important because a system designer normally does not have enough time to go through every standard and find out about their revisions. Moreover, the power-supply industry leads efforts to define the standards and plays a role in interpreting them.
A few basic tips can help system designers address noise problems. If the system requires minimized radiated noise, the designer should use a six-sided, shielded converter. If the dc-dc converter has a wide input range and high input impedance, the converter may loose stability and generate high levels of conducted and radiated noise. To prevent this noise, the designer should minimize the power source impedance by installing low-ESR capacitors close to the input pins of the converter. If any common-mode, input-to-output noise exists, bypass the input pins to chassis through one or more small capacitors.
For conducted, common-mode noise, which may differ from input-to-output noise, use a common-mode choke. Contact the vendor of the magnetics component or the converter for assistance. Finally, the heat generated in the system by all the components (and not just the converter) must be considered. System hot spots not only affect the power converter's efficiency, but could also increase noise, or even worse, produce a catastrophic component failure.