原创 Challenging the easy path of conventional wisdom

 2016-1-21 18:11  1311 20 20 分类: 消费电子

Engineering choices are usually initially dictated by conventional wisdom—after all, who has time to examine every assumption, given the pressures of time-to-market and product margins? This is true in all niches of design, including the vital and very vibrant power-related sector. While we make basic assumptions as a starting point in the design process, it can lead to decisions based on widely held ideas which are perhaps no longer entirely valid, and thus may unnecessarily constrain design options.

Here's one example: I really like low-dropout regulators (LDOs). They do so much, solve so many local power-rail problems, are so easy to use, come in so many varieties that you can almost always find one that meets your requirements, and many of them are available from multiple sources. What's not to like? Further, they are inherently low-noise DC sources, a very critical parameter in some situation. Billions are used every year, often "dropped in" close to a load to solve a local regulation issue after the initial power-distribution approach falls short.

Consider a design which needs a very low-noise DC rail. A common immediate response is to say "we'll have to use an LDO despite its inefficiency, because switchers are known to be too noisy.” Well…maybe yes, may not—it depends on the noise level you can accept. While you may need a "low noise" rail, the real issue is how low of a rail you really need and what will it cost.

If you step back and look closely at the low-noise objective compared to what's available in both switching regulators and LDOs, you may find a switcher with noise that is low enough to meet the specifics of your design situation. Keep in mind that IC vendors recognize that switchers can be noisy, so they have developed using clever topologies and designs to bring that noise down. For example, the LT8640 from Linear Technology is a 42-V input/5-A output switching regulator designed for extremely low -noise output; it can even pass the strict automotive CISPR25, Class 5 peak limits (below).

Sometimes, a reverse of the conventional wisdom of the switcher versus LDO story also needs to be examined. Rather than the specifications of the part guiding the choice, some unique aspects of the application may be the determining factor.

If efficiency is a goal – as it often is – the natural reaction is to think automatically "switching regulator" rather than LDO, since the LDOs are almost always less efficient that a well-designed switcher with comparable output.

But the choice may also depend on duty cycle and turn-on/off issues rather than innovative, improved component design. For example, the application note "Multi-string LED lighting systems and the top four linear regulator questions" from Texas Instruments shows that LDOs can be just as efficient as the switcher in some circumstances, and at lower cost, in some design scenarios.

There are similar stories with power-related components and configurations which are often but not always true, with regard to sense resistors, grounding, high-side versus low-side, sensing and sense resistors, four-wire remote-output sensing, fuses and circuit protection, among other design situations.

It's human nature to get into a mental groove (or rut, depending on your point of view), and it may take a conscious effort to step back and re-examine the situation and assumptions. However, doing so can result in a better design, often at lower cost. It's not a matter of doing an unconventional design as much as questioning conventional wisdom.

Have you ever had to challenge the easy path of conventional wisdom, whether power-related or not? How did that experience go?