Generally speaking designers want to find the most compact, efficient, reliable, and low-cost power supply for the application in question. A logical starting point is the AC input. Nearly all switchers are now designed to operate over a universal input range of 90-264 VAC. However, it's important to check that the power supply needn't be derated below acceptable limits at lower input voltages. Some cheaper models derate by as much as 20% or 30% when operated at 90 VAC. This means that a higher power product will need to be specified, adding to both the size and cost of the solution.
In lower power units, under 100W, there is normally no requirement for power factor correction (PFC), unless the power supply is to be used in a PC or lighting application. To keep costs down, there will be no attempt to reduce harmonic current distortion; EN61000-3-2 class A is the relevant specification that sets harmonic distortion limits.
These limits are absolute and not exceeded by a power supply in this power range, but it is important to recognise that the legislation applies to the end equipment, not the individual power supply. Therefore, should a number of separate supplies be used to power different parts of the same system, the combined harmonic distortion may exceed acceptable limits. The key is to look at the total system requirement and choose a power scheme with an appropriate approach to harmonic current distortion and PFC.
ACTIVE VS PASSIVE
PFC can be applied passively or actively. Passive PFC is lower cost and simple, although circuits can get bulky and the power factor is not greatly improved. Active circuits provide power factors on the order of 0.9 – 0.99. However, they're more complex and more expensive, and they reduce mean time between failures (MTBFs) through increased component count. A passive PFC circuit will usually enable the power supply to meet minimum legislative requirements. But if a high power factor is required, then an active PFC solution will probably be needed.
Power supplies for medical applications have to meet more demanding specifications with respect to safety and EMC—specifically, low earth leakage current and better EMC immunity. The latest edition of EN60601 (November 2004) treats EMC as a safety issue and is particularly demanding in terms of dips and interruptions. It is important to choose a medical power supply that meets the latest edition of the specification.
This is largely a design overhead, rather than a product cost, and there are now power supplies being released that meet the demands of IT, industrial, and medical applications. They're produced in higher volumes than most medical power supplies, making them particularly cost-effective for medical applications while remaining competitive for IT and industrial systems. Some power-solution providers offer EMC test facilities, significantly reducing design and approvals costs.
When specifying power-supply output requirements, it is important to be accurate with respect to minimum, continuous, and peak loads. Some applications require the supply to be continuously rated to the peak output power, while others will allow peaks for as long as 10 seconds without increasing the continuous power rating. Careful consideration of the exact requirements in this respect can result in substantial cost savings by avoiding an unnecessarily powerful and bulky selection.
The nature of the load is another important consideration. Inductive loads will require peak currents (Fig. 1) and may cause voltage surges, which need to be suppressed. High levels of capacitive load require high currents at startup and may cause the power supply to oscillate. Data sheets often specify a limit on the load capacitance.
The overload characteristic of the power supply will be an important consideration where high startup or peak loads are required. Again, data sheets for some cheaper power supplies may not be specific regarding performance under inductive or capacitive loads, effectively making the specifications misleading by omission.
In multiple output power supplies, the output tolerance requirements will affect both cost and efficiency. If all of the outputs require tight regulation, then multiple regulation loops are required. This will increase cost, reduce efficiency and decrease MTBF. A fall in efficiency from 85% to 80% may not seem significant, but remember that losses, and therefore heat generation, are increased by one third from 15% to 20%. This impacts system reliability and leads to additional cooling requirements—with attendant increases in cost and size.
Thus, it's important only to specify the degree of regulation needed for a given application. In most instances, it is sufficient to have a tight tolerance on low-voltage outputs, 5V and below, while having less closely controlled outputs for output rails of 12V and above. Don't pay for regulation you don't need.
In mixed analogue and digital applications, designers try to avoid common return paths that may result in noise transferring between circuits. There are many multiple output power supplies with isolated outputs to meet these needs, though it may not be immediately obvious from the data sheet. Careful evaluation of isolation specifications between each output and earth, and between individual outputs, is needed. As a general rule, the greater the isolation provided, the greater the power-supply cost.
Today, more applications need Class II safety isolation systems where the availability of a safety earth can't be guaranteed. For example, many medical devices are used in the home, and there is no way of knowing the quality of the safety earth in each location. This applies equally to other portable electronic and electrical devices. For this reason, equipment designers need to specify Class II power supplies where an earth isn't required for the equipment to retain its safety integrity.
The need for Class II power supplies can create some EMC design challenges. These are routinely overcome in power units under 100W, but there are relatively few higher power units available that combine class II isolation and good EMC performance (Fig. 2). There is no significant price penalty here; it's simply that choice is restricted in the higher power range.
Having determined the basic electrical requirements, mechanical considerations become paramount. In some instances, the mechanical format is pre-defined, say, for external, DIN rail, or rack-mounting applications. In most others, the most compact and cost-effective solution is an open-frame construction.
In field-serviceable equipment, there may be a requirement for a safety cover. Remember, though, that this will restrict cooling of the power supply, reducing the safe output power or lowering the safe operating temperature for the equipment. In extreme instances, this could result in having to specify a supply with a higher power rating, raising the cost. Nothing should compromise system safety, but covers are best avoided unless they're needed.
A key economic consideration, often overlooked, is the cost of integrating the chosen power solution into the end equipment. The power supply needs to be mechanically secured and the various input, output, and signal interfaces connected. PCB headers and crimped housings are ideal for high-volume applications, while screw terminals or cage clamps will be most effective in low-volume applications because they avoid the need for tooling. Cheaper power supplies may require additional cooling for reliable operation, negating the initial saving in price.
Some power companies provide pre-assembled connection solutions and offer expert advice on thermal issues. These can reduce power-supply integration costs. Therefore, it's not just a matter of what power supply to choose, but which vendor can add most value in achieving system design and economic objectives.
Excessive temperature accounts for a large proportion of equipment failure, with component failure rates doubling for every 10°C rise in temperature. Controlling temperature is key to system reliability and the power supply will create heat within the system enclosure, reducing the lifetime of the entire system. The difference in losses for a few percentage points increase in efficiency is significant and can improve system reliability and reduce cooling costs, resulting in an overall system cost reduction.
In general, power supplies are rated to an ambient temperature of 50°C at maximum power. Forced-air cooling of systems can ease these issues. However, some applications can't accept the use of fans for noise, reliability, or maintenance reasons. This may mean the use of larger or higher-rated power supplies to cope with increased ambient temperatures or alternative thermal-management techniques, such as conduction cooling via baseplates.
In the case of low-power units, the fans are usually external to the power supply and part of the general system cooling. At 200W and above, fans can usually be built into the power supply, adding only marginally to the size and significantly increasing power density. Lifetime can be extended and audible noise problems reduced by choosing products with temperature-controlled fans that only operate when they're needed. This is a cost-noise-reliability tradeoff, where selection is determined by the application.
MONITORING AND CONTROL
Within the 40-200W power range, the most common signals and interfaces are "power fail" to provide a warning of impending loss of output power and "remote sense" for overcoming cable and track voltage drops at the point-of-load in low-voltage applications. These interfaces add little to the size or cost of the power supply, and can simply be ignored in applications where they're not required.
With such a wide choice of low to medium power AC/DC power supplies available, it is important to understand the key parameters needed to help make the best selection for a given application. The fastest route is to take on board the criteria laid out in Table 1 before contacting a company that can offer a range of solutions and add value with reliable technical advice and perhaps customisation services.