Greater efficiency in an ac-dc supply: what's not to like? Everyone knows it's "a good thing" and increasingly a priority virtue. Your decision process is straightforward: look for the supply with the dc output (volts/amps) you require, in a suitable form factor, and be sure it has the minimum efficiency your design (and regulatory agencies) require—all at a price that meets your BOM (bill of materials) costing mandate.
But before you make that decision, it's important to ask yourself two questions:
•What are the implications of an extra point or two in efficiency to design?
•What do those vendor numbers actually mean?
Efficiency's virtues—and cost
It's obvious that increased efficiency results in lower ac power costs for a given output. However, for most applications, the operating savings for a supply in the 200-500W class are quite modest (though they can't be ignored).
Instead, look at the larger system implications of efficiency by flipping the issue over to inefficiency. A supply which is 90% efficient is 10% inefficient, of course, while one which is 92% efficient is just 8% inefficient; that's basic math. Thus, a difference of just two percentage points in efficiency/ inefficiency also means that the 92% supply is one-fifth less inefficient than the 90% supply—and that's a big difference.
This apparently small difference in efficiency has system-level implications. Clearly, there's less heat stress on the supply itself as well as on the entire product, and heat is what drives down the mean time between failures (MTBF) of electronics.
Less heat also gives you flexibility in supply and system placement. Consider the increasingly popular PoE+ (Power-over-Ethernet) standard, IEEE 802.3at-2009. Under this standard, all the operating power for the remote device comes from a chassis and control unit located not at the supplied device, but at a hub—often physically situated in an office cabinet. Those extra watts of heat can complicate placement and can rule out preferred locations or require additional cabinet cooling.
The PoE+ numbers make this dramatically clear: while 12.95W per device was specified as sufficient for the original PoE applications such as standard VoIP phones, security cameras, and wireless access points, the upgraded PoE+ standard calls for up to 25.5W per device. Since a typical PoE+ system may have tens of remote devices, the total amount of power to be supplied—and, more important, resultant dissipation due to inefficiency—can be unpleasantly high.
Another major benefit of greater efficiency is that the designer may be able to go from a forced (active) cooled supply using fans, to a passive, convection-cooled unit. This reduces up-front cost, increases system reliability, and decreases product noise. In fact, designers will often select a supply which is rated with active cooling for higher power than they need, then derate it and use it with convection cooling alone.
If this is an attractive option, it's important to make sure that the supply vendor provides both forced and passive ratings. For example, the XL375 from N2Power, Figure 1, is rated at 375W with active cooling, but is also available fully characterized for convection cooling operation as a 260W unit, as the XL375CC supply.
Checking the specs
As with most numbers which define a product, there are various operating conditions and perspectives which engineers selecting supplies should take into account. This goes beyond the usual advice to compare supplies under identical or similar conditions. New standards and certifications such as 80 Plus and the Gold Standard make such comparisons easier by providing a standard specification framework and starting point.
First, where looking at an ‘efficiency up to x%’ rating, check if that's for a single load "sweet spot" and line voltage. It's possible for a supply which is 80 to 90% efficient at a higher line voltage (240Vacrms) to measure as low as 70% at 120Vac. Keep in mind that with nominal 100V (yes, there are places with this), 120, and 230Vac mains and their allowable bands, you may be looking at 85 to 264Vac operation. Also check the temperature and airflow at which the efficiency is measured; this may differ from vendor to vendor.
In addition, so-called universal supplies, which handle both nominal 120VAC and 240Vac, will likely have different ratings than supplies optimized for only a single ac line value. This is because, in general, the greater the differential between ac input and dc output voltages, the more efficient a supply can be.
Efficiency across the load range is important, as well. Most supplies are far less efficient when operating well below their rated maximum output, and achieve their highest efficiency closer to maximum output. If your application has a relatively constant load, the task of comparing supplies for overall efficiency is easier than if you have wide-ranging minimum/maximum loads. Fortunately, newer supply designs can maintain relatively flat efficiency across a wider span of their load range, see Figure 2.
Power factor correction (PFC) also affects efficiency. Better power factor performance is being dictated by regulatory requirements, and the required PF minimum increases with the supply's power rating. For most supply designs, striving for higher PFC—closer to unity—adversely affects efficiency.
Therefore, it's important to look at efficiency alongside the concurrent PFC rating. It may turn out that a supply rated at higher efficiency is doing so at a lower achieved PFC value than a supply which is less efficient. Note that efficiency is easier to achieve at higher AC mains values than at lower ones. Earlier PFC designs did better at lower mains voltages; there is a trade-off between efficiency and PFC.
Also, clarify what the ac line input to the supply look like for these efficiency and PFC ratings. Though you might assume that a pure sine wave is used, that is actually a misleading model of reality: the ac line is not at all clean and has harmonics of the basic 50/60Hz frequency. Advances in supply topologies and ICs, along with the use of digital signal processing (DSP) in the supply closed-loop control, have helped to improve overall performance.
Improved power-supply efficiency is a virtue, there's little doubt about that. But as you make your supply decision, step back and ask yourself what an extra few percentage points is worth—it may be a lot, especially if it allows you to go to passive cooling. Also clarify how the stated efficiency is measured, under what nominal line and load conditions, and what other factors may affect the validity and credibility of the nominal numbers you are quoted.