It's easy to repeat power supply truisms so often that they become accepted as absolute facts, but good engineering means reviewing the specifics of the situation and making a fresh assessment.
We all know the conventional wisdom and generally understood guidelines associated with circuit design, but a good engineer knows to re-check these carefully. They may be obsolete due to technical change and developments, or simply may not apply in the specific circumstances of the project.
For power supply ICs and power supplies, three often-repeated "truisms" come to mind:
Conventional wisdom No. 1: Switching regulators are more efficient than LDOs (low-dropout regulators). Generally true, yes. Yet, each year, hundreds of millions of LDOs are sold, and many new ones are introduced, so they still do have a vital role. These LDOs are not used just for those situations where the design team is stuck and says, "What the heck, we'll just stick an LDO in there and move on past that power-rail problem."
In fact, in some cases, those LDOs can actually have equal or roughly comparable efficiency to a switcher. A recent blog/application note, "Multi-string LED lighting systems and the top four linear regulator questions" from Texas Instruments, shows where LDOs can be just as efficient as the switcher in some circumstances, and at lower cost.
Conventional wisdom No. 2: Switchers are inherently noisy, so if noise is a concern, use an LDO. Again, generally true. However, some of the newest switcher ICs are extremely quiet, so don't rule them out too quickly for many supposedly "low-noise" design situations. The real questions are: How low a noise level do you need, and is there a switcher whose noise is low enough and that also has the other specifications and pricing that you need?
Look for example, at the LT8614 42V/4A regulator from Linear Technology, which has radiated noise between 15 and 20 dBµV/m from 30 to 300 MHz, which is below the CISPR25 Class 5 radiated-noise requirement. (As a comparison, its LT3065 45V/500mA LDO offers 25 µVRMS noise from 10 Hz to 100 kHz). The switcher vs. LDO decision obviously depends on how much noise you can accept, and where it is in the spectrum.
Conventional wisdom No. 3: If you need increased efficiency, get a better power supply. Whether it is done to increase run time from the battery, decrease system thermal load and dissipation needs, or save on operating costs, a more efficient power supply seems like the smart first thing to do.
But is it really? Look at it this way: Most supplies now operate in the 70% to 90% efficiency range when properly sized to their load. That means that about two-thirds or more of the power dissipation is at the supply's load, not the supply itself. Therefore, while using a more efficient supply will certainly save power, reality is that the load is where the real loss is.
Instead of looking at the supply, look at making the load -- your circuit -- more efficient; you could save a lot more power, and probably do so more easily than trying to squeeze another percentage or two from the supply.
Of course, when the load is a motor or similar mechanical device, it's often hard to boost load efficiency, as the motor has to do real work in the physics sense and therefore must receive a certain amount of power to operate. But if the load is mostly electronics, using lower-power components may allow you to cut the circuit dissipation by 10 or 20 points -- a far greater power saving than you can get from a more efficient supply.
Are there other supply-related truisms you hear or assume? Are they really true in all or most cases? Have you ever stepped back to examine them, and were surprised by the results?