An important indicator when testing power supply performance: noise measurement

Today, I want to talk about another important metric to consider when testing power supply performance: noise measurement.

Where does power supply noise come from?

Power supply noise is generated from a number of different sources. Like any amplifier, the power supply generates various types of noise, and switch-mode designs also need to deal with the inherent switching noise that occurs. Switching supplies can be designed to minimize their switching noise, and output filters can be incorporated to further reduce this noise. However, the exact level of noise generated by a power supply can only be known after actual measurement.

Transient ripple noise

Why measure noise?

Bias voltages within any system are, as I see it, the foundation of electrical circuits. All systems can interface with these supplies and must deal with the noise associated with them. If the noise generated from (or through) the supply exceeds the tolerance of the circuit, the system will automatically fail. The problem with noise is that it may not (at least not always) cause catastrophic failure issues, but sometimes it can cause intermittent errors that may not be obvious when the system is thoroughly tested in a specific environment with a range of typical component values, but can cause all kinds of problems later. In some cases, I have seen software patches written to "mask" temporary system errors that may have been caused by power supply noise issues. Is it the best way to fix problems in software that may have been fixed in the power supply? I beg to differ, but I don't want to turn the topic into a philosophical discussion, so I'll save that for a follow-up article, "Fixing Errors, Software vs. Hardware"...

I have sometimes seen noisy power supply designs cause systems to fail EMI testing and slow product releases. These slowdowns could have been avoided if the noise issues were properly tested and addressed early in the design process. When it comes to biasing analog circuits, power supply noise can cause poor system performance, potentially reducing the end customer value of the product. Think about the power supply issues in the analog signal path of some sensors. In such systems, noise will directly affect the sensitivity of the system, and the higher the noise floor, the lower the sensitivity. If designers take the time to actually measure and analyze the noise generated by the power supply, they can either accept the performance or often make changes to avoid system-level problems later, at little additional cost. The cost of testing and potentially modifying the power supply is often much less than the cost of debugging and modifying the system level later, or the cost of poor performance after the product is released.

It's this attention to detail that makes the difference between an average product and one that delivers the highest performance and reliability.