Where's your inefficiency?
Efficiency and its complement of inefficiency are hot topics both in the general media and in the nitty-gritty of engineering design. No need to go over the reasons for that here: if you're reading this, you know about the consequences of inefficiency with respect to power use, run time, immediate cost, thermal load, packaging, and more.
Statistics calling out power use and therefore room for improvement are tossed around like paper clips; I've seen statements such as "60% of the power in the US is consumed by electric motors," "50% is used in transportation," and "25% is used for home heating and cooling."
Of course, these authoritative numbers usually come with an excess of significant figures, such as 24.7% rather than a more valid rounded number of 25%; I suspect that's to give them the veneer of credibility. When I add up all the claimed usage percentages, I usually end up with a total of at last 200%, sometimes even 300%. Sure, there is some overlap that accounts for part of this, but I suspect much of it is a combination of SWAG (scientific wild-ass guessing) numbers plus numbers from sources with an agenda to promote.
But the numbers do point to an issue: how do you actually determine the sources of power use (and loss) in your design. If it's an AC/DC supply supporting a fairly steady load, you can measure the AC line current and voltage, along with the load current and voltage, using a combination of basic transducers, instrumentation, and current-sense resistors, among various approaches.
Things get a little more difficult when the load or system has pulsed modes, PWM, transient operation, quiescent states, and more. Yes, there's instrumentation and techniques that will greatly help, but you still have to know more about your system's operation characteristics to make sure your data is representative and valid.
What about when you leave the electrical/electronic world? When you try to determine optical-related efficiency, you're in for a lot of surprises unless you have some experience and guidance. How do you measure the efficiency of an LED, whether individually or in an array? You could measure the DC power input and the heat given off - how would you do that? Or you could measure the light output directly, but to what off-axis angles? And for which parts of the spectrum: IR, visible, a slice of visible, or all? What about incandescent sources, with their unique thermal and optical aspects? Do you know what an integrating sphere is, anyway?
(And then there's measuring mechanical and fluid power, another very different story entirely...) Long story short: measuring power consumption and assessing points and periods of inefficiency are not trivial tasks in many cases. When you see numbers associated with these parameters, it's a good idea to investigate how they were obtained, to get a sense of their credibility and validity.
What's been your experience with measuring power and determining efficiency? Did you ever find out afterwards that your set-up and numbers were not as valid as you initially thought they were? Did this change your design decisions?