The fifth part of this series is concerned with the supply of voltage and current in electronics packaging systems. The particular power supply requirements of these systems vary according to specification and area of application. Various PSU types and forms of construction are available, such as AC/DC and DC/DC converters, switched-mode PSUs and linear regulators, 19" or open-frame, with various input and output voltages.

The energy supply for electronics packaging systems comes from the public mains, which in Europe, for example, is 230 V_AC, while in Japan it is only 100 V. For industrial electronics, telecoms, datacoms, automation and instrumentation, however, indeed for all applications where electronics packaging systems may be found, such voltages are too high and potentially dangerous. The electronic components require a far lower voltage, which must also be a stable DC. Function boards with their processors normally require a supply of 5 V or 3.3 V. Newer processors operate on 1.8 V or even less. For these, the PSU provides what is known as an intermediate bus voltage, e.g. 12 V. From this, point-of-load converters on the boards convert this intermediate supply to the voltage required by the processor, e.g. 1.8 V. Today's cooling fans are also supplied at 12 V or 24 V_DC. While earlier AC fans could be connected directly to the mains, voltage fluctuations in the latter could lead to inconstant fan speeds. It was also necessary to provide different fans to suit the mains voltage of different locations (100 V, 230 V etc.).

To supply the system components with the necessary DC voltages, power supply units (PSUs) are integrated into the system (Fig. 1). The functions they perform include the following: to provide a secure galvanic isolation from the mains supply; to reduce and rectify the voltage; and to regulate fluctuations and interruptions in the mains supply. The secondary-side voltage obtained from the PSU is known as SELV (safety extra low voltage). Also important for the system is an output current limiter for safety purposes. This prevents, for example, a damagingly high current flow in the event of a short circuit on a board. Other technical requirements for PSUs are overvoltage protection, lightning protection and electromagnetic compatibility (EMC). This concerns interference emissions and interference immunity. The characteristics described here are governed by law (e.g. the Directive on Electromagnetic Compatibility and the Low Voltage Directive of the EU) and are covered in Europe by the CE symbol and declarations of conformity.

In VMEbus and CompactPCI systems, output voltages of 5 V and 3.3 V are required (for processors) and 12 V (e.g. for disk drives and fans). MicroTCA systems require 12 V as the main supply and 3.3 V as the management voltage. For AdvancedTCA there are two possibilities. If the system is telecoms-based and supplied at -48/-60 V_DC, no additional PSU is built into the system, since the DC supply is already available locally. Other AdvancedTCA systems, however, draw their supply from the ordinary AC mains, and these are fitted with PSUs that provide the telecommunications standard -48 V_DC.

The specifications for the individual bus technologies vary considerably in the requirements they set out for system power supply. The VMEbus specification, for example, prescribes only the voltages in terms of magnitude, number and tolerances. The CompactPCI standard additionally defines, for a plug-in PSU, the connector to be used (pin-out, form and load capacity) and functions such as signalling and remote switch-off. As a rule, where no redundancy is required, CompactPCI systems are fitted with open-frame PSUs with power-factor correction (PFC) and a rating of about 300 W with wide-range input voltage. If redundancy is required, it is recommended that plug-in 19" PSUs are used. Then when the PSUs are run in parallel in redundancy operation (n+1), the current share bus ensures equal division of the current between them.

No definition is given in the AdvancedTCA specification as regards the AC supply. The MicroTCA specification, on the other hand, gives very detailed data on the input voltage. MicroTCA specifies a PSU in the form of a single full-size module with 600 W capacity, four separate input voltages (+/-24 V_DC, -48 V_DC, - 60 V_DC) and a wide-range input (100 V_AC to 240 V_AC). Additionally, a management controller on the PSU monitors voltages and currents; and for each individual AdvancedMC module, it enables the supply of both the main voltage of 12 V (16 outputs) and the management voltage of 3.3 V (16 outputs) on the instruction of the MicroTCA carrier hub. It follows that a fully conforming MicroTCA system requires a relatively expensive power module.

To reduce costs, however, there are also solutions that reduce the management functionality to a necessary minimum. Less expensive solutions can be realised using other forms of PSU construction, e.g. open frame. The simplest approach would be a conventional open-frame PSU with +12 V output. The 3.3 V management voltage also required would be obtained via a step-down converter on the backplane. The switching of power to individual slots would be performed on the backplane. Such a solution must however be supported by the MCH (MicroTCA carrier hub) used and by the AdvancedMCs. It is also possible to insert a power management mezzanine board into the backplane, and the MCH can then use this to switch the supply to each slot. Such a solution would however lack current, voltage and temperature measurement; the mezzanine board is not included in the e-keying, and nor is a redundant power supply possible. A further option is to create an external 12 V source and use this to supply the AdvancedMC slots via a MicroTCA module with 12 V input. A module of this type is known as a power feeder module. Here it is equally possible to do away with the management completely or at any rate to realise a reduced level of management functionality.

A fundamental distinction in PSU design is made between DC/DC converters and AC/DC units. In these expressions the first term represents the input and the second the output. As concerns AC/DC converters, these in turn can be divided into two types - switched-mode and linear regulators (Table 1).

Each of these presents advantages and disadvantages in different areas because of their different circuit arrangements and principles of functioning. The user must consider the most important characteristics for the given application and select accordingly between the two types (Fig. 2).

DC/DC converters are always switched-mode designs. They are used in all situations in which a DC supply network is already present. In telecommunications, for example, the supply is always 48 V_DC, which is then transformed down by the converter to 12 V_DC. Similarly, in automation there is a 24 V_DC supply alongside the AC mains. For historical reasons, railway applications can indeed involve a number of DC supply voltages (in the range 14 to 150 V_DC) that must all be transformed to e.g. 5 V or 12 V. Nevertheless, AC/DC converters are the classical power supply system for almost all types of application. These are also found in railway systems where, for example, a 230 V mains is present on a train or in a trackside facility.

As concerns physical construction, there are basically two types of PSU that are preferred for VMEbus and CompactPCI systems. These are standardised 19" plug-in designs on the one hand, and non-standardised, freely mounted open-frame types (Fig. 3) on the other. 19" PSUs integrate well into the system, being simply inserted into a dedicated slot from the front in the same manner as any other plug-in module. No further wiring, etc, is necessary. This simple pluggability is also an important feature for rapidly exchanging units and thus for assuring high system availability in applications that require this. Additionally, placing the PSU within the board cage is an advantage in terms of system cooling. Cooling, using fans or fan modules, is specially designed for this arrangement of the boards, and so no additional measures are necessary. The PSU is cooled by the defined flow of cold air. However, the aperture dimensions of 19" PSUs are fixed by the standards and the designer is often obliged to use costly components in order to maintain conformity.

Open-frame PSUs, on the other hand, with their non-standard construction and lower-cost components, offer greater value for money. An open-frame PSU may either be housed in a simple metal enclosure, possibly with its own fans, or alternatively may be a PCB-mounted assembly with cooling, with no housing, integrated directly into a system. Such open-frame devices are usually bolted to the inside of the rear panel of ordinary 3 or 6 U 19" systems. The mains input and the various outputs (e.g. four distinct output voltages for CompactPCI) to the backplane must be wired via cables. Consequently the open-frame PSU is hard wired into the system and cannot be exchanged as fast and as simply as can a 19" plug-in unit. It is also easier for faults to develop in the wiring. Finally, since these PSUs are not positioned within the defined airflow of the board cage, separate arrangements for cooling must be made.

The lower-cost open-frame devices are still frequently found in older VMEbus systems. However, more common in VME64x, which requires the same four voltages as CompactPCI, are 19" plug-in PSUs, which are also used in CompactPCI applications. Here open-frame solutions are only considered where cost is the primary issue and where the advantages of plug-in units, redundancy and high system availability are not high priorities.

No choice exists between 19" plug-in and open-frame for standardised MicroTCA systems. For these a standardised power module must always be used that can be inserted from the front. The module size corresponds to the standardised mounting grid of the AdvancedMC modules that are integrated into such systems as function boards. In AdvancedTCA, as mentioned earlier, no PSU is necessary if the system is used in a telecoms environment with a 48 V _DC supply. However, AdvancedTCA systems that run from the AC mains are always fitted with front-mounted plug-in PSUs. The system availability of 99.999 % specified in the AdvancedTCA specification will permit no alternative; the PSU, like all other components, must be swappable quickly and during operation.

PFC (power-factor correction) has been a legal requirement for a number of years (EN 61000-3-2). This is achieved via a passive or active filter module added to the PSU to minimise the proportion of unwanted upper harmonics in the supply and thus to bring the power factor as close as possible to 1. In non-linear consumers, sinusoidal supply voltages give rise to phase-shifted and non-sinusoidal input currents. These comprise the sum of various high-frequency elements, i.e. harmonics, and can cause interference and losses in power supply networks and in other electrical devices. It is therefore necessary to make current flow as close as possible to sinusoidal and to minimise reactive current. The PFC corrects the non-linear current consumption of the consumers and, in the case of active PFC, also compensates the reactive current.

Power-factor correction is in two types: passive and active. With passive PFC, the filter used is in principle a mains filter with particularly high inductance. Such devices are easy to manufacture, but yield only moderately good results. They are normally used only in low-power applications up to 200 W. In an active PFC system a more complex circuit is required. However, the result is a very good power-factor correction (to 0.99). A special switching-mode power supply controller determines the current drawn in accordance with the sinusoidal variation in the mains voltage. These active PFC devices consist of a rectifier connected directly to a step-up chopper that charges a large capacitor to a voltage greater than the peak value of the incoming AC. The end consumer is supplied from this voltage (typically 350 to 400 V). An active PFC has a further advantage in that it compensates for fluctuations in the mains voltage. It is often so dimensioned as to allow the devices it supplies to be used anywhere in the world, on any available mains voltage, without conversion. A wide-range input of this type is almost always requested by customers who wish to use their VMEbus, CompactPCI and other systems worldwide. The wide range of existing DC supply voltages of railway systems (14 to 150 V) can also be covered with this circuit topology and the resultant wide input voltage range.

Today's requirements for environmental sustainability mean that a high efficiency is essential. Not only is high efficiency important in terms of a high power density, but it also has a significant bearing in electronics packaging systems on the cooling or heat removal. The better the efficiency of the PSU, the less heat is dissipated into the system that must then be removed. The usual efficiencies of Schroff PSUs (switched-mode) are up to 90 %. In VMEbus and CompactPCI PSUs the figure is about 85 %. However, power supply units with only a single output voltage, such as 48 V for AdvancedTCA, can achieve efficiencies of up to 95 %. Higher efficiencies require not only higher development and product costs but also longer development times.

For PSUs, environmental sustainability also means conforming to the EC guideline 2002/95/EC that limits the use of certain hazardous substances in electrical and electronic devices (RoHS). This governs the use of hazardous materials in devices and components. Since mid-2006 it has no longer been permitted to market end products that contain hazardous substances. While systems used in industrial environments are not subject to this guideline, Schroff has taken the strategic decision, ahead of time, to convert all its products to RoHS compatibility. This also affects all PSUs manufactured by Schroff. Moreover, it is important for environmental sustainability that PSUs are not designed as single-use products but that they can be repaired and expanded. Such expandability is also necessary in the case of VMEbus and CompactPCI systems, which themselves can be expanded on a modular basis.

If PSUs are to be used in electronics packaging systems for worldwide use, they must also have certain official approvals. In the European market this is documented by the CE symbol. This also covers the EU guidelines for each of EMC, low-voltage systems and electrical safety. Additional testing marks in Europe include the familiar VDE and TÜV symbols. European railway equipment is governed by the EIRENE standard for train communications plus the standards of the individual countries. Certain northern countries such as Norway (NEMKO) and Denmark (DEMKO) have their own test marks that have been harmonised with the CE symbol. In the USA the UL certification system applies. This contains similar guidelines to those of the European CE, but without the EMC guidelines. The Canadian CSA certification has content comparable with that of UL. The technical requirements for achieving these certificates are one matter; the not insignificant costs of obtaining approval and certification are another again.

The term 'high availability' indicates the capability of a system to maintain unrestricted operation in the event of the failure of one of its components. A high availability of 99.999 % (availability class 5), as required by the AdvancedTCA specification, for example, means that a system may be down for only 26.3 seconds in a month or 5.26 minutes per year. For VMEbus, CompactPCI and MicroTCA systems no mandatory level of availability is set out in the specifications, but a certain level may be required by the user for specific applications. To ensure high availability, individual components - or all components - of the system must be designed for redundancy. They must also, for certain degrees of availability, be hot-swap capable so that they can be exchanged while the system is in operation. This naturally also applies to the PSUs used.

Redundancy operation of PSUs (device redundancy) means the use of several PSUs in parallel (n+1, where n = number of PSUs required to supply the necessary power, 1 = one additional PSU included in the system to cover in the event of the failure of one of the PSUs). Generally, all PSUs in the system are operated in parallel, and the total load is divided equally among them. As a result, the individual PSUs each operate below capacity and have a longer working life. This is achieved by what is known as current-share operation. Current sharing is made possible by means of a special line, the current-share bus (CSB). This continually measures the output currents and the PSUs are notified via the bus of the share of the current that each has to supply. This ensures an equal apportioning of the load. The user of the system does not have to adapt any other circuit. Should one of the PSUs fail, the remaining units take over its load, and so run at full capacity. In such a situation the faulty unit can, if it is a 19" plug-in device, be quickly and easily exchanged. Redundancy operation of this type is not possible with 'ordinary' PSUs, however; they must be specially configured for this mode of operation. Such PSUs, designed for redundancy operation, are normally hot-swap compatible. This means that they can be removed from or inserted into the system without affecting either the power supply or the system application. This is necessary to enable problem-free swapping of PSUs without having to shut the associated system down.

Another important point in regard to high availability and redundancy is the redundancy of the supply mains itself. An uninterruptible supply of power can be obtained by the use of UPS (uninterruptible power supply) devices. These consist of batteries, an electronic power converter and an electronic monitoring and control system. Common UPS units have capacities ranging from about 300 VA up to many hundreds of kVA. Another important performance factor for a UPS is the maximum buffer time, which is a function of the battery capacity. Depending on the requirement, this may be from a few seconds to several hours. Some UPS units allow extra batteries to be connected, thus extending the buffer time; these are normally units of 1500 VA and above. A conventional UPS draws its supply from the AC mains during normal operation, converts this into DC and charges the batteries. This internal DC is reconverted to AC that matches the voltage and frequency of the original mains. This AC is now available as input power to the PSUs in the electronics packaging systems. Should the mains supply fail, the systems quickly switch over to the AC that has been generated from the batteries.

Another type of UPS is known as a "power system" (Fig. 4). These function similarly to a conventional UPS. The difference is that the DC of the battery is not converted to AC. A power system thus supplies DC, at the voltage of the battery itself. Such power systems are used only in telecoms environments (e.g. AdvancedTCA), where the systems require a 48 V_DC supply. Because the voltage is converted only once, power systems tend to be lower in cost than traditional UPSs and to be more efficient.

Supply redundancy can also be realised by means of two separate mains networks (primary mains and emergency mains). An automatic electronic switchover device provides the changeover between the networks; this may be installed in an electronics cabinet with multiple electronics packaging systems and multiple PSUs. This changeover system functions as an electronic switch that selects between two mains networks (primary mains and emergency mains). If the primary mains cuts out for more than 10 ms, the device switches over to the second input and thus onto the emergency supply network. The total switchover time is less than 20 ms. The changeover electronics features a positively driven safety relay linked to special relays with large contact spacings. This arrangement assures conformity with the EN 60950-1 safety standard. The inputs to the changeover module are separately monitored and their status signalled. The signal outputs are galvanically isolated via two optical couplers. The output transistors of these couplers may be used for signal processing in the application. The wide input range of the module from 90 to 264 V_AC allows it to be used anywhere in the world. The maximum nominal current of 8 A allows it to operate in Europe with applications of up to 1600 watts.

For cost reasons it is beneficial to use off-the-shelf PSUs, of which there is a wide range currently on the market. Should a new design be really necessary because of particular customer requirements, this will generally be very time-consuming and costly. Modifications of standard PSUs that require no change to the PCB layout, e.g. different output voltages, different colours of indicator LEDs etc, are relatively simple alterations to make. Modifications of this type also entail relatively low fixed costs and can be made for item quantities of as little as 20 or 50. More extensive modifications, but where it is still possible to make use of existing components and designs, can be realised economically for quantities of 500 upwards. Examples might include a change to a different form factor or developing a PSU with the positive characteristics of a linear regulator and a lower residual ripple and reduced interference. Such a PSU is not available off-the-shelf; only certain elements of the requirements are available on the open market. Depending on the work involved, development times may here be between 8 and 16 weeks.