Efficiency of a power supply

I need some info on efficiency of power supplies. Imagin a power supply with 91% efficiency that can provide a maximum of 800watts.

Does the rated 91% efficeincy is for all loads that going out from this source? I mean if it is providing 500W, the efficiency is still 91% or less or higher? is there a formula between load amount and efficiency?

Thanks.

-

There is no one formula. This is individual to particular power supplies. Good power supply datasheets will show a graph of efficiency as a function of current. Sometimes you might get a minimum efficiency spec over a current range (assuming fixed voltage supply). Sometimes you only get the efficiency at the maximum output power, which is usually the condition for maximum dissipation in the supply.

The reason there is no one standard formula is that there is no one standard power supply topology, especially when you look at the details. There is much choice of switching frequency, modulation technique (frequency, pulse width, both), synchronous rectification yes/no, type of magnetic core, etc. There are many other tradeoffs possible between size, weight, cost, transient response, output ripple, idle power, efficiency, and more.

-
As a note it can be determined and a formula generated if you know the controller chip used, associated components and their topology, but this is, as Olin points out, pretty much going to be chip to chip specific. – Kortuk Jul 18 '11 at 12:40
@Kortuk - it's a lot more than a formula. I've seen hundred-page MathCAD power supply calculation documents which still have an error on overall efficiency prediction by 0.5%. – Adam Lawrence Jul 18 '11 at 15:57
@Madmanguruman, it must have been a significantly more complex buck/boost then what I have seen. I have seen full models for buck/boost that has just the simple few components and external FETs. Also simple control. He already received my +1, but I was just trying to add that as extra information. – Kortuk Jul 19 '11 at 14:54
@Kortuk Yep - a complete AC-DC power supply with PFC and auxiliary converter. Just trying to illustrate that an efficiency calculation can approach an infinite number of terms if 'real-world' predictions are required. – Adam Lawrence Jul 19 '11 at 14:59
@madmanguruman, Thank you for helping add even more information. The more people can learn the better! – Kortuk Jul 19 '11 at 15:03

This is something you have to look up in the vendor's data sheet. Usually the efficiency is best at the rated power (800 W in your case), degrades a bit at light-load (500 W) and is 0 % at no-load (still some demand for household, but not output at all).

See fig. 9-1 of this data sheet for a typical curve: http://www.pulspower.com/pdf/cs5_241.pdf

-

The efficiency will vary with output power, because the supply will always consume some minimum energy itself. There should be a graph provided with the power supply which shows you the efficiency against power output. What is quoted as a single figure in the headline is best case, usually at maximum power output.

-

There's no simple relationship between a regulator's load and its efficiency, so no formula. In any case it can't be constant: with lower loads regulator's losses become more important, and at zero load the efficiency is also zero, no matter how well the regulator is designed.

A good datasheet should give you a graph of efficiency versus load, like this one for an application of an arbitrary Linear switcher (i.c. LT1912).

The graph is constructed from measurements, rather than from a formula.
This kind of graph is typical for a switcher; the design is optimized for a certain load. Other types of regulators may have monotonic graphs.

-
Many of these graphs may be monotonous, but only some are monotonic. – Olin Lathrop Apr 8 '12 at 14:22
@Olin - oops, small error. I'll fix it right away. Thanks. – stevenvh Apr 8 '12 at 14:24

You're going to find lots of variation in power supply efficiency as a function of both input voltage and output loading.

If the manufacturer only specs efficiency at one input/output condition, don't expect that number to hold anywhere else.

There are some newer directives out there such as 80 PLUS which, if a product is compliant, provide a standardized measurement methodology as well as efficiency minimums for various loads. 80 PLUS Gold for server power supplies guarantees 88% efficiency at 20% load, 92% efficiency at 50% load, and 88% efficiency at 100% load at 230V / 60Hz input.

-