I see lots of press releases for new switching regulators. The presenting companies usually talk about why their device is better than everyone else's part. The listed criteria are interesting and sometimes contradictory.
Here are some of the interesting, eye-catching claims. These are often what we call open-ended comparisons -- "better" -- better than what? It's left unstated. We could assume they mean "than the other guy's crappy parts." Thus, we have this:
- Lower cost;
- Higher efficiency;
- Better transient response;
- Higher switching frequency;
- Capable of operation from higher input voltage;
- External parts are cheaper;
- Better regulation;
- Capable of operating with a higher input-to-output ratio;
- Less EMI/EMC.
Let's consider some of these. "Capable of operation from higher input voltage" would imply that it's built on a higher voltage process. That means it won't be as inexpensive as the devices that operate at around 5V (maximum) input. So if you see "lower cost," you can assume that you cannot power the device from your intermediate 12V bus.
What about "capable of operating with a higher input-to-output ratio"? That means the input could be pretty high -- maybe 12V to 48V in -- and the output could be pretty low -- maybe 0.9V to 2.5V. Such a high ratio means that the time that the upper switching transistor is on and the lower one is off is a very short amount of time (relatively speaking). If your switcher is operating at a high switching frequency (a few MHz), your upper transistor may need to turn on, settle down, and turn off in tens of nanoseconds. Better make sure they can do that. Of course, those parts cost more.
If your switcher is capable of outputting tens of amps, those output transistors will have a large chip area and more gate capacitance. Good luck getting them to turn on-and-off quickly. So maybe operating at a lower frequency would be better.
At a lower switching frequency, especially below 1.7MHz, and especially in automotive applications, you run the risk of generating interference with the AM radio band. And at lower switching frequencies, the output inductor and filter capacitors must be larger (in value and physically). So that means more PC board space used and increased component cost.
At lower switching frequencies, the regulator's transient response is not as good -- it takes longer for the regulator to react to step changes in load current draw, partly due to the larger filter components mentioned above. Maybe you should try for a somewhat higher frequency.
But not too high, because that same problem from three or four paragraphs ago -- the transition time for the transistors to turn on and off -- will affect efficiency. As they turn on or off (but not when they are full on or completely off), their power dissipation goes way up. Better slow down again. But not too much, or the inductor will get too big, increasing its DC resistance and power dissipation.
So there's the quick overview of evaluating the manufacturers' claims and weighing them against your design requirements. How have you dealt with these issues?