Power supply specifications: what's lurking behind the datasheet?

The datasheets in power supply catalogues are normally in 'shortform' format; for some applications you may need to ask the vendor for more comprehensive data

Starting at the input

Most AC/DC power supplies are now designed for 'universal' AC input with an input voltage range of 90-264VAC quoted. Particularly if the power supply is to be used near its full rating, it's important to check that full power is still available at the lower end of this voltage range. At lower input voltages the input current rises and, particularly in power supplies with active power factor correction (PFC), the switching current in the boost converter becomes relatively high, so overall efficiency falls. The available output power may have to be reduced by up to 30% at 90VAC compared with the quoted nominal power output. If the datasheet does not state 'full power from 90-264VAC', or something similar, look deeper into the specification for a de-rating specification or graph.

Input current is stated to enable correctly rated switches and relays to be specified - it's a function of output power, efficiency and power factor and is unambiguous. Inrush current figures are more open to interpretation and 'specmanship'. This is often specified as 'cold-start' inrush current. Here, a thermistor is used to control the inrush current, its resistance falling as it warms up. However, switching the power supply off and on after it has been running for a while can produce an inrush current many times the cold-start figure. While this is not an issue that affects the power supply itself, the higher current needs to be taken into account when selecting fuses and switching components. In many higher power units the thermistor dissipates significant power, so it normally gets switched out of circuit after the unit warms up. The device then cools and works again when it is switched back in before start-up. So, from a system design point of view, it's important to know whether inrush current figures are based on a cold start or for all conditions.

Earth leakage current is normally not an issue except in medical applications, where low figures are required to meet various international specifications that have been established to protect patients. Remember to check the input voltage at which leakage current is given and that it doesn't vary too much with temperature. These factors will affect leakage current in the final application.

The only other input specifications provided for most power supplies are fuse ratings and whether both line and neutral are fused. Fuse ratings are for fuses that are not usually user-serviceable; the fuses are over-rated and are there to protect against catastrophic failure of the power supply. In-built overload protection will operate faster in all other instances. Dual-fused power supplies are usually only required for medical applications.

Outputs

Output voltage is specified at the connector of the power supply so it's important to consider the expected volt drop between power supply and the point of load, especially in low voltage, high current applications. In some instances it may be possible to adjust the power supply voltage upwards manually, typically by up to 10%, to compensate for voltage drops in the system. In others, particularly applications where the load is variable, the system designer may have to choose a power supply with a 'remote sense' function. This provides automatic compensation by detecting the voltage at the point of load and adjusting the power supply's output level to even out variations. Designers also need to be aware that varying one output rail on a multi-output power supply can result in the same percentage change in all the others. If this is a problem, independently regulated outputs will be needed, adding to the cost and size of the power supply.

Single output power supplies now rarely need a minimum load. However, a minimum load is often needed on the main output of multi-output units so that other outputs can be kept within tolerance. Such power supplies usually have only one feedback loop on the main output so a load is needed to maintain regulation on the others and prevent output voltages going low. But look out for power supplies that need a minimum load on more than one output; it may add complexity to the system to provide this. If the minimum load requirements are not clear in the datasheet, check with the vendor.

Hold-up time is an important factor if short term power interruptions, in the order of a half-cycle or cycle perhaps, are unacceptable to the system. Hold-up is determined by the input reservoir capacitor; 16ms is a full cycle in USA, 20ms in Europe. There is little room for confusion here, but look out for what happens if there is an AC supply glitch. Ideally, the power supply should ride through this without requiring a manual reset - and for the longer the better.

Line and load regulation figures can be particularly misleading. More reputable power supply vendors will state the maximum variation in output voltage over the entire input voltage range when specifying line regulation. More 'creative' datasheets sometimes state the specification over a more limited range of input voltage shown somewhere in the small print, if it's present at all. Load regulation is similar in that it should be stated over the entire specified load range; many datasheets will specify load regulation over perhaps 50%-100% load, the total possible variation being much greater than first impressions might suggest. If the conditions under which line and load regulation are specified are not clearly stated, it's worth asking the question.

Typical datasheet output specification. Final output voltages will depend on set accuracy, line and load regulation, and whether adjustment of the main output will be tracked by other outputs

Ripple and noise figures are specified under widely varying conditions on datasheets so care is needed to make accurate comparisons. Low noise figures for the power supply contribute significantly to low overall system noise and assist the process of system qualification in this respect. When you are comparing power supply noise figures, remember that noise increases with bandwidth. A typical datasheet will state maximum noise levels at 20MHz bandwidth, simply because an oscilloscope will usually allow measurement at this setting. However, some will use 15MHz, giving a misleadingly low noise figure, and a figure that many oscilloscopes cannot measure. The problem doesn't end there. Noise above 20MHz is still a problem for many systems, so it's worth asking if data is available for output ripple and noise at higher frequencies. Regulatory bodies will assess equipment for approval up to 30MHz for conducted emissions and up to 1GHz for radiated emissions so knowing the performance of the power supply across the same range is very useful. The poorer the performance of the power supply, the less specific the datasheet is likely to be. Finally, watch out for the conditions under which noise measurements are taken. Power supplies should be compared with respect to their noise performance at the output connector using a very short measurement probe to avoid radiated noise. Power supply vendors specify different measurement points and may put forward noise figures that are obtained after additional filtering components have been added. In short, the datasheet is often only a crude indication of the noise performance of a power supply. When comparing products, the best approach is to obtain samples and measure them under identical conditions - ideally conditions that reflect the application in terms of earths, shielding and the location, number and length of wires connected to the power unit.

Other output specifications are generally less critical, but here are a few things to bear in mind. The initial set accuracy specifies output voltages when the power supply settles down after it start-up phase. Line and load regulation are derived from this so, for example, if the set accuracy is ±1% and line and load regulation are also 1%, the possible output variation from the nominal specification is ±1% with respect to anywhere between 4.95V and 5.05V, rather than ±1% of 5V. It's not likely to cause a problem - but there may be voltage-critical applications where it could.

If transient response is important to the application, check the load step that's specified. Check all the details: recovery to what point, how fast, and over what size of step. (Transient response will sometimes be specified with a rate of change of current.)

Overload protection is included in every AC/DC power supply on the market. But for some applications the type of protection is important, and that's not always clear from the datasheet. The three basic types are constant current limit, fold-back current limit, and trip-and-restart. Trip-and-restart and fold-back protect the power supply from the load and prevent high fault currents in the event of a system problem. Constant current overload protection can be a requirement in applications where batteries, lamps, or high capacitive loads, are encountered.

Finally, two other start-up figures are sometimes shown on datasheets. The start-up delay indicates how long the power supply takes for the output(s) to come into specification. It's of no real concern in the vast majority of applications. The start-up rise time can be of interest in applications where 3.3V and 5V logic rails need to track, but in most cases it's not important and many companies do not specify it.

Efficiency, power density and reliability

Efficiency figures are another prime candidate for manipulation. Efficiency varies with input voltage and output voltage, by individual output, and by load. Datasheets will invariably show the best case at nominal input voltage and full load. At less than full load, efficiency falls, so it's not good practice to over-rate the power supply for a given system. Efficiency cannot be determined in isolation; it will vary with the operating conditions of the power supply, so the latter need to be taken into account when calculating the amount of heat that will be generated and determining any cooling arrangements that may be necessary.

Power density is a figure that power supply manufacturers like to boast about - the more watts per cubic inch, the more advanced the power supply - perhaps. It is based on the maximum power rating of the power supply and the volume of space it occupies. In practice it is of little interest to system designers, they simply want to know if the power supply fits the space available. However, for those that are interested, it's worth ensuring that comparisons take full account of any external filtering, protection or PFC circuits that may be needed for one power supply vs. another; it can change the results a lot.

Reliability figures are perhaps the most ambiguous of all of those found on a power supply datasheet. Reliability is normally quoted in terms of MTBF, or Mean Time Between Failures. Broadly speaking, MTBF is inversely proportional to failure rate, and assumes that failure rate can be measured or predicted. As most datasheets are prepared when products are new, the MTBF is invariably a predicted figure based on measured or predicted failure rate information for all of the components within the power supply. The first thing to be aware of is that an MTBF figure quoted in isolation is meaningless. Operating and environmental conditions will dramatically affect MTBF, so these have to be stated in order for accurate comparisons to be possible. For example, a widely accepted rule-of-thumb is that MTBF will halve for every 10°C rise in ambient temperature. There are two 'standard' sets of conditions under which MTBF is often quoted. The first is documented in a handbook known as 'MIL-HDBK-217'; the second is based on the Bellcore (Bell Communications Research) methodology. For a given power supply, the resulting MTBF figure can be as much as an order of magnitude greater when calculated using the Bellcore methodology compared with that of the MIL-HDBK-217. In both cases the predicted MTBF is only as accurate as the component data from which the figure is extrapolated, so at best it's a rough guide. However, where like-for-like methodologies have been employed, the figure does provide a useful basis for comparing the relative predicted reliability of different power supplies.

Environmental conditions vs. power

The ambient temperature in which the power supply operates will affect the available output power. Some need to be de-rated from 40°C, some from 60°C. Normally, full power will be available at up to at least 50°C, often with de-rating to half power at 70°C. However, not all power supplies need to be de-rated in this way and careful examination of the datasheet can avoid over-specification of an unnecessarily large and expensive power supply. Conversely, if the de-rating requirement is not clearly shown, misapplication of the power supply can have unfortunate consequences.

Of course, the available power will depend upon cooling arrangements. Higher power units often have their output power specified only with forced-air cooling at a given air flow. If these power ratings are given at high air-flow rates, say 30CFM, be careful; this can be extremely difficult to achieve in practice. It's safer to go for the option requiring the lowest CFM. It's worth noting that forced-air cooling can enable the power supply to operate at higher ambient temperatures before de-rating has to be considered.

Even a modest 18CFM of air flow can more than double the available output current from an AC/DC switcher but the improvement will vary with each model, as this table for a multi-output 130W power supply shows

EMC and safety

EMC and safety are usually shown on datasheets as conformance to a variety of international specifications based on IEC, EN, or UL standards. In practice, power supply manufacturers effectively self-certify their products with respect to EMC as it is the end equipment that needs to be tested and formally approved. With respect to safety, power supply makers normally obtain approval from regulatory authorities in order to quote compliance with the relevant safety standards.

With respect to EMC, conducted emissions are the key concern. As mentioned earlier, power supply emissions are usually measured with short wires connected to a resistive load. Some industry-standard specifications permit additional earths to be used to determine performance and these won't necessarily be present in the application. Furthermore, some power supplies just meet the specification while others will do so by a wide margin. Equipment makers need to look for power supplies with the lowest possible emissions in order to reduce the chances of needing addition components to get their products through final approval. The solution is to ask for a copy of the full results with details of how the tests were carried out. The likely performance of the power supply within the application then has to be assessed. Remember, all of the above relates to conducted noise; power supply manufacturers will not normally quote figures for radiated noise, except in the case of desktop or plug-top external types, because the final equipment enclosure will provide most of the protection in this respect.

Susceptibility of the power supply to ESD, surge, fast transient bursts and radio frequency interference is the other EMC issue to be considered. This is quoted with respect to pre-defined levels of performance from level 1 to level 4. There are large differences between these levels. For example, level 4 is twice as demanding as level 3. What happens to the power supply under these defined conditions is an equally important matter. Performance is defined as class A, B, or C. Class A meaning no deterioration, class B meaning it recovers from EMC induced faults automatically, and class C means that it requires manual intervention to reset. Once again, it is the requirements of the end equipment that will determine the most suitable power supply.

Products will normally be certified for safety by agencies such as UL & TUV. It is worthwhile checking the conditions of acceptability in the approval document as it may specify approval up to a given temperature or with a particular orientation of the power supply. For the safety approval to be valid, the power supply needs to be used within these limits.

Summary

Power supply datasheets contain a lot of information that needs careful qualification and consideration if it is to be meaningfully applied. System designers need to understand how individual performance characteristics are interrelated and the electrical and environmental limits under which the figures presented in a datasheet can be relied upon. Most important of all, it's vital to remember that power supply specifications are created in isolation without knowledge the end application. How the power supply will perform within the system it powers must always the most important consideration. And that's something that demands more than a cursory glance at the datasheet.

About the author


Gary Bocock is a qualified electronics engineer (HNC) and Member of the Institute of Electrical Engineers (MIEE). He has worked in the power supply industry for 20 years in design, development, applications and management roles. He has been with XP for 10 years and has held a variety of engineering and management jobs, culminating in his present position as Technical Director.