So far our studies have focused on the components, materials and modules that are brought together to form a populated board. However, we now turn to the wider consideration of the overall system, trying to take a holistic view of the challenges of creating an electronic product.
As with all systems, our electronic system has to perform a function, have inputs and outputs, and a “boundary”. Typically in an electronics context this boundary will present as some form of enclosure containing and protecting the electronic elements. Our system may be very simple, as with a computer mouse, a package containing a few mechanical parts and a simple board assembly. But an enclosure may equally be very large and complex, with a large number of interconnected board assemblies and modules.
As we have already considered components and board assemblies in some detail, this Unit concentrates on the additional elements of the enclosure and the interconnect. We will find that, whatever the size and complexity of the system, there will always be common elements toconsider, even if the solution is trivial for the simplest situations.
WIthin our enclosure, we need to connect individual elements within the system and also make contact with the external environment. In both cases, in order to be able to identify all the options, we need to be very clear as to our objectives. For example, asking what are we trying to transfer?
If we are transferring power, either from the supply mains or internally from supplies, then we will choose a metallic interconnect, almost invariably copper. Similarly for earth connections, whether these are power returns or screening. However, the situation is less clear when it comes to transferring radio-frequency energy, where plain copper may be inappropriate: at VHF/UHF frequencies, the interconnect will need to be accompanied by screening; at microwave frequencies, wave-guides may be an appropriate choice.
When we come to transmitting data, then an assumption of a copper interconnect is dangerous, as significant data is transferred in different ways. Think about the ways in which your mouse movement is reflected by activity at the Bolton server, and make a list of possible non-copper connections before looking at our answer and comments.
So far, our consideration of interconnect has been confined to power and data, but we also have to remember that our system produces heat, so that connecting system elements thermally is important. This brings us to the second general group of items that are present in a system, as distinct from a board assembly. These are the various support elements that help the product to meet its environmental specification objectives.
For example, our system will need to build in thermal management, whether heat is conducted away from the sources via metallic connections, or transferred out of the system using a moving fluid. Most typically, this will be moving air, but liquid may be employed in some challenging cases. [Much more about thermal management issues in module AMI4817 Design for Thermal Issues]
As well as managing heat, our system may need to manage vibration and similar excitation (such as bump, shock and acceleration). Even where these aspects are not actively managed, for example by providing anti-vibration mounts, the product will at the very least need to survive the mechanical forces applied without failing. Also, in addition to thermal and mechanical management, the system as a whole will need to resist moisture, dust and similar. However, this is generally a function of the enclosure itself, rather than a support element, so is considered in our next section.
These interconnect and support elements need to work correctly:
[ back to top ]
We have talked of a system as containing components and interconnection, but always within a boundary that usually takes the form of an enclosure. What does this enclosure do? Think about this before reading our comments.
More than anything else, the enclosure will provide protection against the environment, in terms both of the traditional temperature, humidity and mechanical requirements and of EMC compliance. For this last, we need to isolate the system elements within the enclosure from the world outside because, depending on the application, a system may either be sensitive to imposed electromagnetic fields or else disturb external electronic equipment. Much more about this in our modules AMI4814 Signal Integrity & EMC and AMI4966 Design for EMC and LVD.
The bulk of our attention in this Unit is inevitably on temperature and humidity considerations because, in real life, a product will experience a range of conditions that may vary between “hot and dry” and “cold and wet”. In some cases, the enclosure may even be attacked by water under pressure!
Skim read the first part of our paper Simulating the real world in the test house to give a flavour of how temperature and humidity in particular might be specified in order to make tests representative of real life.
As you have read these materials, you will have seen how moisture and temperature resistance might be specified, and the challenges associated with the severe environment. But don’t forget that a severe environment might also embrace aggressive materials, whether dust or acid gases, and a range of mechanical tests. Also, as you will see in module AMI4957 Test Strategies, we actually make use of tests that are representative of life, but at greater severity, in order to weed out problem systems.
A practical enclosure:
But there are many different types of practical enclosure, and the application will dictate both the materials to be used and the particular “equipment practice” chosen. Let’s look briefly at a range of possibilities:
In some cases, the electronics can actually be built on the enclosure. As an example of this integrated approach, telephone handsets have been created where the electronics is assembled onto an interconnect pattern deposited on an internal surface of the handset. However, such a simplification poses practical assembly problems, so there are as yet few integrated board/enclosures.
There are, by contrast, many products that contain a single functional board within an enclosure. Bear in mind, however, that, as in the case of the portable radio, there may be just a single board assembly, but other elements of the circuit may be standalone units within the enclosure. A common example is the battery holder, which frequently is moulded into the enclosure, and connected through to the board, rather than being board-mounted to form a stand-alone sub-structure.
There is a degree of “grey area” between the single board enclosure and the equipment case, an enclosure intended to integrate the customer interface, the necessary sub-assemblies and interconnections, and a motherboard. Frequently items such as power supplies are separate, and some elements may be provided as plug-in modules. There are many examples of traditional instrumentation built in equipment cases, but the concept is also used for the ubiquitous personal computer.
With all these “equipment cases” considerable attention is paid to aesthetic design, and there is in consequence considerable variety, both in size and finish. Whilst standard equipment cases may be bought off-the-shelf for small-volume products, standardisation and modularisation are often not key features. However, as you will know from the computer world, there is helpful standardisation when it comes to products that are intrinsically single-board but may attract additional modular functionality, usually as plug-in modules.
Standardisation becomes of even more significance when we consider larger or more complex items of equipment assembled as individual card racks. The concept goes back to before World War II, with modular systems built with stepping relays for telephone exchange equipment. Many of the dimensions adopted in those early years are still with us, despite attempts at metrication. The reason for the longevity of the “19" rack” is that it is very versatile, and allows many different products to be assembled using comparatively few standard parts. And the general arrangement of having a backplane, distributing data and power, with functionality and processing power added by plug-in modules, is a way of creating a product that is easy both to service and upgrade.
The set of card trays built into an equipment rack is a common sight in control and telecoms practice, either as a single rack or in multiple “bays” of equipment. Generally the use of a backplane is restricted to the individual card tray, the bulk of the interconnection1 being carried out within the backplane, with relatively few connections made between different parts of the rack.
Whilst the individual trays within the rack can be presented in a protected enclosure to form an equipment case, and these offer significant “building space”, the overall effect of what is a very solidly-built piece of equipment is not overly attractive as a stand-alone piece of bench furniture! Which is why so many people have chosen personalised equipment cases. However, there is a move towards miniaturisation, as may be seen in the National Instruments range.
The materials chosen for the enclosure will depend on manufacturing cost considerations and the expectations of the customer. At high volume, a moulded plastic enclosure will offer cost benefits and relatively simple assembly, particularly where fixing elements such as nuts and stake points are moulded into the enclosure. And of course the material, colour and texture can be chosen so that no further finishing is required. However, the tooling costs for mouldings are relatively high, and there are some constraints on dimensions, draft angles and so on. Also, whilst a polymeric material is a good insulator and appropriate for applications such as double-insulated consumer electrical products, lack of conductivity means that the enclosure has no value as part of the EMC screen. Such screening either needs to be carried out by an alternative means, or the internal surfaces of the plastic must be coated, by processes such as evaporation or plasma spray.
The metal enclosure has no such limitation, though care must be taken to ensure that the parts of the enclosure are bonded together adequately from an EMC perspective, and not just fitted together to form a mechanically robust structure.
Carry out a web search for metal enclosures, and examine the ways in which they are assembled. Note for example the use of extrusions and castings, and the use of pre-coated metal sheets to avoid the need to apply a separate finish. Then, for EMC-equipped enclosures, look at the way in which the front panels may be connected to the rest of the enclosure to provide an “EMC-tight” enclosure that can still be disassembled readily to provide access to the internal boards.
Mechanical processes are at the heart of what is usually referred to as ‘box build’, that is the assembly of printed circuit boards with other elements to create a completed product. More and more this is the province of the contract assembler, who may even package units with manuals in individual cardboard boxes, and ship them to warehouses ready for sale to the public. The rationale for this is that it allows the assembler’s client to concentrate on design and marketing, without getting involved in any aspect of manufacturing.
The box build process involves three distinct phases, the first two of which maybe interwoven depending on the detail of the product design:
The techniques involved will depend on the materials and design of the case and also on whether the product is designed for disassembly and on the volume of manufacture. Considering each of these in turn:
The term “permanent attachment” is only relative, and everything can be disassembled if you try hard enough, albeit with some damage to the components. However, techniques such as soldering, welding, riveting or the use of adhesives are commonly employed only for sub-assemblies (Figure 1) or where the assembly is intended for disposal if faulty, and no modification or rework is anticipated. But even here there may be some elements of demountable design, for example, allowing for battery replacement.
The techniques available will depend on the base material of the enclosure, and plastic enclosures offer a number of useful alternatives such as ultrasonic welding and solvent welding, where thermoplastic materials have been selected.
Traditional box building techniques, using screws or nuts and bolts, are common for low volume manufacture because the techniques are very flexible, and the tooling investment is limited to general-purpose equipment (Figure 2). However, this type of assembly is inherently slow, so other building methods may be favoured for high volume. Even where a metal enclosure is required, as for applications needing EMC screening, parts can be secured using “snap-in” fittings or by using one of several different custom closure devices where parts are firmly secured with only a quarter-turn.
Also, as you will know from your experiences with personal computers, it is also common for enclosure components to be slid into slots in the framework, and merely held in position at a small number of points, rather than secured by screws on all sides.
Particularly where volumes are high enough to make it possible to commission custom plastic parts, the final part of the “jigsaw” that forms the enclosure can often be used to hold in place elements of the internal assembly, cutting the cost of the product.
The challenges in box build are to make products that can be assembled in one way only, as quickly and efficiently as possible, and subsequently disassembled readily into their constituent parts for recycling.The main requirement for the box build stage is that the designer should:
What is involved in the box build task will very much depend on the intended volume of manufacture: procedures that are acceptable in the ‘craft’ conditions of a prototype shop may be totally unacceptable in a volume-manufacturing environment. Consideration there has to be given to simplifying the process stages, and making as much use as possible of mechanical aids. Rather than use individual fasteners, the product may be clipped together, welded together, or stuck together, using a variety of techniques.
Don’t forget in your design to allow for the effects of the physical environment. Will continued temperature cycling, or shock, or vibration, cause parts to become loose, so that connections go open-circuit or the whole assembly falls apart? There are also safety aspects to consider, such as clearances for mains voltages, and ensuring that children’s fingers can’t access dangerous parts! And, of course, you will take into account the nature of the final application, which may suggest rounding corners and including other safety features.
Take a look at a range of electronic products, and examine:
[ back to top ]
Having looked at the practicalities of the enclosure, we now have to consider the interconnections used. Typically these will divide into internal interconnections, and the (usually fewer) connections made to the outside. In both cases, there is a choice to be made, whether to use permanent “hard” wiring, or use some form of demountable connection, allowing the system to be disassembled.
Permanent wiring is tamper-resistant and offers the highest connection reliability, but is generally too inflexible for use in other than security-related applications. It is more usual to use connectors of some form, whether in association with loose cabling or with fixed wiring, such as a motherboard or backplane.
As we have seen earlier, the physical format of the connector and the materials used for its construction (in particular for contact surfaces), will depend on the number of operations that the connector pair is required to survive. The requirements for connectors that are designed to be made usually only once, and rarely broken, are much looser than for connectors that are expected to be made and broken fairly regularly, for example for plug-in modules that may be changed each time an equipment is reconfigured.
Other than the expected life and maximum contact resistance of the connection, from the equipment perspective the key parameters of the connector are the force needed to make the connector, when inserting one part into another, and the force needed to disengage the contact pairs. Clearly having a high insertion force is not helpful for connectors where many hundreds of connections have to be made at once, and a low extraction force may cause unreliability in challenging mechanical environments.
The density of interconnect generally decreases as we move from silicon to the board assembly, to the local sub-rack, to the rack, and finally to the outside world. At the card interconnect level, the density of connection required between boards varies greatly according to the circuit function and partitioning. Where many connections need to be made, that a backplane may be a useful intermediary between the high density interconnect on individual boards and much simpler wiring to the outside world.
The backplane distributes ground and power as well as providing data and signal connections, and will often be fitted with decoupling components and test points. Increasingly, more and more backplanes will also have controlled impedance features and/or appropriate internal screening.
Backplanes are conceptually similar to motherboards, but focus more on interconnecting the assemblies plugged into them, and rather less on providing built-in signal processing or intelligence. However, it is not infrequent for backplane electronics to include such functions as power supply monitoring and data-bus drivers.
Backplanes and motherboards have another commonality, which is that they are required to withstand the insertion and removal of modules, often maintaining circuit function during the “hot plugging” of modules. Apart from the electronic design considerations needed to achieve live connection and disconnection, there is also a requirement for the backplane/motherboard to be sufficiently rigid to prevent distortion and consequent malfunction during module insertion and removal. To meet this requirement, backplanes are typically considerably thicker than other rigid boards, and they frequently need to be assembled using press-fit connectors. This is because it is difficult to heat the backplane sufficiently for satisfactory wetting to take place and, in any case, the surface tension of solder is insufficient to draw it through such a long barrel.
Whilst some simple products use single-part connectors, where the male connector is provided by a hard gold finish on “gold fingers” on the board, most modular products use two-part connectors, where one section is soldered to the board, and the other is connected to the backplane. Which half of the partnership contains the male elements will depend on the design, but most frequently the plug-in module will have male connectors. However, it is helpful to use a design of connector where the pins are protected from damage when the module is not inserted.
Making connectors simply by pressing the two halves together is not as simple a process as one might think. If you consider the simple task of plugging in a mains connector, you will know that some degree of lateral movement is usually applied in order to encourage the insertion, and inserting a module may also require a slight “wiggle”, unless the assembly has been designed with zero insertion force connectors, where the contact force is applied only after the mating surfaces have reached the correct relative position.
In the case of modules plugging into backplanes, it is important to have good alignment between the two mating halves, and the practicalities of tolerancing usually dictate that one or both halves should have some lateral compliance. Some guidance may also be required, by shaping the active pins and the opposing pin entry, by appropriate shaping of the mating halves of the connector housing, or even by using separate locating dowels and holes in order to ensure that the parts mate cleanly.
In the same way as we look for consistency of connector mating, we need to ensure that connections other than on board and backplane are made reliably. We do not want quality to disappear in a “rats nest” of wires, each of which has been damaged during assembly and is likely to fracture when the wire is flexed!
Wiring between system elements may involve a range of different wire types, all the way from the smallest “equipment wire” to heavy cable for power applications. Whilst there is much variety in the insulation applied to wires, most small wires comprise a copper core and a concentric shroud of insulator that has been applied by extrusion, and which needs to be removed selectively to allow connection to be made to the core.
To use a wire safely we need to remove the covering without cutting into the wire. Whilst some specialist transformer wires have thin lacquer coatings that are dissolved during the soldering process, in most cases the outer covering will need to be stripped mechanically. Typically this is done by catching hold of the section of coating at the end of the wire, simultaneously almost cutting it through and pulling the resulting tubular section of coating off the wire and away from the remaining insulation. This procedure uses the fact that the coated wire (which is typically tin-plated) has poor adhesion to the polymer, so the bond can be sheared. Cutters have to be adjusted very carefully, to ensure that the bulk of the thickness is cut through, but without “nicking” the body of the wire. Because wire stripping is relatively difficult to carry out cleanly, and many coatings are difficult to remove, nicked wires are commonly seen, although these impact adversely on the longevity of the joint.
Any old piece of wire will do! For some ideas about the variety of wires available and why this statement is not true, read our short paper Wires matter.
Another aspect of consistency in wire manufacture is to use “harnesses”, assemblies of wires that have been prepared and cut to length, and are often created as a pre-wired three-dimensional “connectivity kit” that can be presented to the system components and connected in a series of simple operations.
Particularly with a stranded wire, once the end of a wire has been stripped of insulation, it is fragile and subject to damage, it is likely not to be straight, and it may be too flexible to use easily. After all, this is why solid wires are normally used for “hook-up” applications. This difficulty in use means that, although certain terminals are designed to accept bare wires, it is more common to give wires some type of termination. The design of the termination will depend on the current rating and the application; light-current terminations, such as those in connectors, are likely to be crimped, whereas heavy-current terminations are more likely to be soldered. For an indication of the wide range of different wire terminations, take a look at manufacturer’s web sites such as http://www.kpsec.freeuk.com/components/connect.htm or http://www.tonesties.co.uk/cable_terminations.htm.
Wire connections are often grouped together, either in connectors or onto terminal blocks. In order to make it possible to remove individual connections, connectors are usually designed so that wires are terminated with a crimped component that becomes the pin or socket of the connector. The alternative, of soldering each wire in turn into a “solder bucket” that is part of the connector body, is a task requiring considerable skill and the careful application of heat to ensure that the insulation does not overheat and pull back from the joint. Particularly where connections have been soldered, and the wire made more rigid by solder moving up a stranded wire by capillary action, such groups of wires are best supported in some way, to prevent flexure of the cable assembly leading to fracture of individual wires.
An alternative way of grouping a number of wires is to use a terminal block. In most designs, the retaining screw makes only indirect contact with the wire end, pressure being applied to the wire through two flat surfaces. However, in some designs, the wire end may need to be formed around the screw in order to make a reliable contact. Given the reduced flexibility that comes from solder dipping a stranded wire, it is tempting to pre-tin wires used on terminal blocks, but there is a danger in having too thick a coating, since solder will creep under continued pressure, and this may lead to an intermittent connection.
The simplest form of mechanical attachment where an electrical connection is intended is a simple loop in the bare end of a wire placed underneath a washer and secured by a bolt, but solder-less connections can be considerably more sophisticated, as in the case of a press-fit connector.
However, the basic principles are the same – in order to get an electrical connection, there must be intimate contact between the mating parts, so that the electrical connection is not compromised by the build-up of oxides or other surface films. Typically this involves what is often referred to as making the connection ‘gas-tight’, where deformation and pressure combine to create an interface protected against the environment.
There are many types of joint, and correspondingly a number of ways of ensuring that this metal-to-metal connection is secure. As an example of what might be involved, consider the stages involved in making a good connection to your car battery – this involves cleaning the surfaces, applying as much pressure as possible, and finally coating the surfaces to protect them from environmental attack. Not every electronic connection is quite as gross as this, but the general principles are worth bearing in mind.
Also be aware that, with all mechanical connections, the integrity of the metal-to-metal contact, which often is almost a diffusion weld between the two metal surfaces, is compromised if the bond is broken and then an attempt is made to remake it.
In the papers at the three links, we consider different types of gas-tight connector. In Crimping, the internal barrel of a metal part is pressed against a bare wire with sufficient force to create a mechanical link and break through the oxide layers; Insulation-displacement connectors (IDCs) ‘cut through’ both insulation and oxide; Press-fit connectors use interference between pins and the insides of holes in order to create local points of bare contact.
The gas-tight connections considered so far all relate to multiple connectors. There is, however, a fourth mechanical connection used for single-wire connections that seems at first sight to be inherently unreliable, but has proven itself to be a highly reliable method for point-to-point wiring. This is what is referred to as “wire wrapping”, where a wire is stripped of its insulation and the bare end wrapped hard around a square post with sharp corners. The corners cut through both insulation or any surface layer, at the same time breaking down the oxide on the post, to create a metal-to-metal contact. By controlling the wire tension, and the way the wire is wrapped around the post, it is possible to make a joint that is reliable for long periods. Applications range from quite simple prototype boards to complex military assemblies.
We have illustrated some aspects of wire wrapping in the Unit text, but encourage you to visit the links for more detailed information.
The posts around which the wires are wrapped are either component sockets or connectors into which assemblies are plugged. The technique is therefore more widely used for backplanes than for board assembly. Posts are square and typically 0.017" square 1" high and spaced at 0.1" intervals. The better-quality posts are hard-drawn beryllium-copper with a flash plating of gold, but bronze posts with tin plating are also used where cost is an issue.
Typically the wire is silver-plated soft copper, and insulated with Kynar, a fluorocarbon resin. The normal wrap has 1½–2 turns of insulated wire wrapped around the post, in order to help keep the wire from fatiguing where it meets the post, followed by 5–10 turns of bare wire, the number depending on the gauge of wire.
The process can be carried out manually, which makes wire wrapping very suitable for repair and modification in the field; the wires need to be replaced, but posts can be re-wrapped a number of times.
Totally manual operations are prone to variation, so semi-automated wire-wrap systems, as illustrated here, are popular. As with any assembly process, the level of automation will depend on the application and funds available ranging from semi-automated systems that identify the post to be wrapped but use the operator to make the final approach, to totally-automated systems that will both strip and wrap the wire.
A number of practicalities associated with wire wrapping are quite closely associated with a harness practice. One of the few books with any content on this is Gerry Herrick’s Electronic assembly: soft soldering and wire wrapping published by Prentice-Hall in 1992 as ISBN 0132487667, though this is unfortunately out of print. However, an indication of the requirements can be gauged from this NASA workmanship standards pictorial reference.
What are the important factors to take into account when making electrical connections by mechanical methods?
Whilst single cables can be routed within the enclosure, and longer cables perhaps coiled on the enclosure floor, having some form of wire manegement is generally good practice, both in creating a consistent assembly and in making servicing easier. The simplest method uses cable ties, either to create a self-supporting wire harness, or to fix the cable to the enclosure wall. Modern cable ties employ a ratchet principle, so that they are easy to tighten and much more difficult to loosen. Typically the protruding lead will be cut short once all the necessary wires have been inserted, but it is common practice only to “close the loop” temporarily until the entire wire harness has been assembled and checked. A cable tie-wrap gun can then be used to tension and cut each cable tie.
An alternative to using individual ties is spiral wrapping. A typical spiral wrap material is made of polyethylene, with the inner edges rounded to prevent cable damage. The technique makes it easy to create a grouping of wires where individual wires leave the array at arbitrary intervals and in different directions. In this way, spiral wrap resembles the original way of creating a wire loom, which was to use a waxed cotton tape to create interconnected tight loops around the cable-form at regular intervals. Whilst, to the traditionalist, this approach still yields the neatest and highest-quality result, it is labour intensive and relatively inflexible, with the result that cable lacing with waxed cord is only seen on high-reliability avionic applications, where its light weight is attractive.
Another flexible solution to the cable control problem is the use of ducting fitted to the inside of the enclosure. By making breaks in the duct cover, wire exit points can be created where needed. Ducts are particularly useful where system changes are anticipated, as additional wires can be inserted without having to disturb existing wiring. However, the results are rather less consistent than with harnesses in terms of the relative positioning of cables.
IPCs DVD training program (DVD-60C) demonstrates the effects of processing errors on the end product and identifies “Seven Sins of Wire Harness Assembly”:
This selection suggests that the major causes of defects which occur during wire harness assembly are problems that have their roots in either training or the correct transmission of design information.
[ back to top ]
Our enclosure, filled with electronic assemblies and interconnect, presents a considerable design challenge, and one that is much more complex than designing individual boards. We should therefore consider the mechanical assembly needed for box build in the same way that we applied Design for Assembly principles to board assembly in Unit 8.
If you haven’t already done so, we recommend you to read the linked paper on Design for Assembly analysis for a description of methods that can be used to help quantify the effect of design changes on the ease of assembly of a piece of equipment. These same generic ideas that we saw used for board assembly are even more applicable at the enclosure level, the rating methods in particular having been developed for tasks broadly similar to the operations carried out during box build.
At the same time as we consider Design for Assembly, we should also consider “Design for Disassembly”, anticipating that some products will need to be accessed for servicing during life, and that, increasingly, electronic products will be returned for disassembly and recycling at the end of their life.
Accessibility for servicing means giving attention to the way in which a technician can gain access to elements of the product for both diagnosis and replacement, seeking if possible to make the task intuitive rather than needing frequent reference to the manual. In any case, what technician ever started by reading the documentation?!
Design for Servicing also means making it possible to replace individual elements without have to disassemble the whole structure. Conversely, thought should be taken to how best to reinstate the repaired product, making it as simple as possible for the case to be replaced correctly, and without leaving any parts unused at the end of the process . . . ending up with a pile of pieces that were once part of the structure, but now have no home, is potentially as bad for electronics as it is for clock or car repairers!
There are two general principles that help us here – the first is to make as much use as possible of standardisation, both of design approaches and the components themselves. For example, using a single type of screw helps everyone, from procurement right through to servicing.
The other helpful general approach is error avoidance. This has been developed under the title of Poka-yoke (pronounced POH-kah YOH-kay), which comes from the Japanese words poka (inadvertent mistake) and yoke (prevent), and is equivalent to ‘mistake-proofing’. Developed in the 1960s by Japanese manufacturing engineer Shigeo Shingo, poka-yoke aims to eliminate product defects by preventing or correcting mistakes as early as possible. The technique has been used most frequently in manufacturing environments, and is also applicable to software and development.
A poka-yoke device is any mechanism that either:
Poka-yoke devices are used either to prevent the special causes that result in defects, or so that each item produced can be inspected inexpensively to determine whether it is acceptable or defective.
Look at some good everyday examples of poka-yoke in action at John Grout’s Poka-yoke page.
Design for Assembly aims to create a design that is efficient to assemble, can be assembled without error and at high yield, and can be disassembled when needed for servicing or recycling. But is this all that an analysis can offer to the design engineer? Are there other ways in which the product design can be optimised? As one of the tools available to the design engineer, we recommend Functional Analysis, as this takes a more fundamental approach.
Whereas DfA assumes that all the functions of the product are required and provides recommendations for easier assembly, Functional Analysis questions the need for a particular component to be included, on the basis that it is a waste of time and money to assemble components that are not required to meet the customer’s functional specification. For example, components are sometimes added to a product as a quick fix to a problem: subsequent redesigns solve the problem, but, because the engineer who added the quick fix has retired, nobody knows why it was added!
By challenging the DfA assumption that every component has a critical function, Functional Analysis gives rewards that are potentially huge, compared with any of the other cost reduction and optimisation tools available, yet it would be relatively simple to incorporate it in the DfA process. All it needs is commitment, a wider vision, and a preparedness to think “out of the box”.
More information about FA and the way it is applied in our paper on Functional Analysis.
DfA in the widest sense, including Functional Analysis, is an important activity that needs to be embedded within the design/development process flow, but becoming part of a company culture, rather than just an imposed paperwork exercise. It needs a structured approach to the problem, and a consistent application of the methods, all within a team environment.
Imagine that you have been requested to design a DfA process for your organisation. The obvious starting point is to define a list of requirements for the process. Can you list the main requirements?
Compare your answer with this one.
However good the design may be from the DfA perspective, the designer also needs to remember that the assembler will need to convert the requirement into a detailed assembly process sequence. Not only does this interpretation add the possibility of error, but it is only at this detailed conversion stage that the assembler will finally establish whether or not the assembly sequence is practicable. Close cooperation between Design and Assembly, preferably carried out before the design is finalised, is recommended.
[ back to top ]
Whilst any product should be designed so that servicing and upgrading are easy, most customers will expect a product that requires as little unplanned maintenance activity as possible, consistent with the price. In other words, we need to “Design for Reliability” as well as the other “Design fors” we have considered in this Unit and earlier.
In order to design effectively for reliability, we need to establish both the likely failure mechanisms for the product and the design environment for the application. And this environment should be based on practical experience, and not just the letter of any specification.
Most products are tested after assembly under factory conditions. In what ways would you expect conditions in the real world to be substantially different? Think about a range of familiar products such as a mobile phone, a personal computer, a car engine management system and the radar in a military aircraft. Which of these environments might be expected to be the most severe?
Compare your answer with our comments about The “real world”.
You will have seen from our comments that the real world is a hard place! However, if we know exactly how hard it is, we can make judgements as to the likely failure mechanisms in our product, and how these may be influenced by the temperature and humidity of that environment.
Once we have identified the ways in which the product is likely to fail, then we can adopt appropriate strategies for minimising the failures of the components themselves, and for eliminating any failure modes that are related to the design. These aspects are considered in the three subsections that follow.
Without making this sub-section into a whole book, it is not possible to list all the ways in which failures can be induced by the effects of the ‘real world’. However, we have tried to give some insight into both the likely and unlikely expected failures, and into the way that one problem can lead to another – failure modes rarely exist in isolation.
High temperatures can cause damage or incorrect operation in both electronic and electrical components. Whilst some types of component exhibits higher resistance as the temperature increases, most insulating components show an increase in leakage current: good examples of this are diodes and capacitors. Fortunately, leakage current is rarely a cause of complete device destruction, but malfunctions can occur, particularly in high-impedance circuits.
Given a sufficiently high temperature of course, many of the materials used will degrade: thermoset polymers will char; thermoplastics will distort; solders will creep and then melt. This can be significant in the case of heat-generating components such as inductors, transformers, solenoids and motors. Here high temperature, combined with internal dissipation, can degrade and char the insulation on internal wiring, with insulation failure leading to permanent short-circuit damage.
Low temperatures can also cause components to fail due to parametric changes in electrical characteristics. However, such failures are usually reversible, and correct function is restored when the temperature rises.
One consequential failure of low temperatures that has already been hinted at is the tendency for cold components to attract moisture, frozen out of the surrounding atmosphere. Not only does this have the potential to give rise to corrosion, which only needs some ionic contamination to proceed quite rapidly, but the moisture can dramatically reduce insulation resistance across the tracks.
Low temperature also makes most polymeric materials more rigid, and care must be taken during operation at temperature extremes to ensure that embrittled parts are not shattered.
Arcing may occur whenever current-carrying components are separated, as when switches or relays are operated, and is often seen between the brushes and commutators of motors and generators. Apart from generating electromagnetic noise, arcing also progressively damages the contact surfaces, leading to eventual failure. Arcing becomes more likely, and is more difficult to suppress2, at low air pressures, since the dielectric strength of air is proportional to the pressure. That is one reason why aircraft and spacecraft electrical systems operate at relatively low voltage levels.
2 Attempts are usually made to reduce arcing by using voltage suppression components, such as capacitors and diodes, across relay or switch contacts.
For most electronic products, the most severe environmental challenges result from high humidity. Not only is humidity associated with a number of failure mechanisms, it is also essentially uncontrolled. Whilst warm air usually has low humidity (less than 50% RH), cold air may have a very wide range of humidities, from very dry (as in Arctic conditions) to very damp (“precipitation in sight” as the shipping forecast puts it!). Most severe of all is the hot and damp combination found in the tropical jungle.
Keys to the effect of high humidity are:
Laboratory tests use pure water, but the real environment often contains contaminants. For this reason, the battery of tests available includes exposure to salt spray and a range of lubricants and other fluids associated with vehicles.
We have seen already that the temperature cycling of joints leads to eventual failure through stress. For complete assemblies, there are a number of other possible failure modes associated with the incompatibility of materials. Sometimes you will just hear creaks and groans; at other times fracture and failure may occur. In the laboratory, temperature cycling is always carried out so that the final half-cycle takes the product from hot to room temperature. This is to avoid the situation where a product allowed to regain room temperature from cold might attract condensation, which could adversely impact on performance. Of course, real life is not so kind!
And there are many other hazards. We tend to forget insulators, although they are important in cables, connectors, and coils, as well as on a circuit board. Insulators can degrade and fail, usually in the long term, due to mechanical damage, absorption of humidity or excessive temperature. The last of these will cause embrittlement and then fracture, but the same effect is produced by sunlight or chemicals.
Most of these aspects are covered by standardised tests, but the cost of assessing every aspect of a product is substantial. And, however careful you are, there will be failure causes that are difficult to predict – some insulation materials are enjoyed by rodents which may infest agricultural machinery or military equipment which provides snug winter quarters!
In any design, failures can result from inadequacies in the design, or from the failure of individual components. Bearing in mind that “components” can be mechanical, as well as electronic, the basic strategy on components should be to:
In most cases, the best information on likely failure modes will come from the designer’s knowledge of the properties of materials and of the technologies used to make components. In a limited number of cases, information may be available on the distribution of failures, as in our tables in Component failure modes. However, such information can be hard to come by.
Once the likely failure modes are known, these can be allowed for during the circuit design phase. As examples:
In all cases, the extent to which protection against failure can be afforded and is implemented will depend on the likelihood of and the consequences of failure.
As well as modifying the circuit to provide protection against failure, we can also seek to improve the reliability of individual components. A common approach is to use a components that is apparently “over-specified” for the application. Read our paper on Derating for an insight into how effective this strategy is, and for practical ways in which reliability may be enhanced.
Our discussion has assumed that you are conversant with basic reliability concepts, and will appreciate the importance of failure rate, its variation with time, and its relationship with the likely life of the product before repair is necessary. If expressions such as “bathtub curve” are but a dim recollection, read our paper on Failure rate and its variation with time.
Where components are likely to fail, and a simple Pareto analysis can quickly identify the most vulnerable components, we recommend taking specific steps to reduce the failure rate of these components. The possibilities for this include the derating of components as discussed above, and the screening of any high-risk devices to remove components that are potential defects.
In order to be able to devise screens that prevent components with potential defects reaching the next stage of manufacture, or even the customer, we need to understand more about the likely failure modes. Read our paper Improving reliability by screening to see how accelerated tests are used to weed out weak components. Note that, whilst the focus in this paper is on component screening, similar concepts can be applied at the equipment level, although the tests are on a larger physical scale.
As part of the initial study for high-reliability products, designers may be required to estimate the likely failure rate for the system, building this up from estimates of the failure rate of each component. This is a complex task, and one that frequently produces unexpectedly high (and probably invalid) failure rate values. For an insight into the issues, read our paper Modelling failure rate.
Sometimes components fail for intrinsic reasons, but often the problem has design as its cause. Designers need to ensure that systems are protected both against current overload and transient voltages. Read our paper Electrical protection for guidance on some ways of achieving both, and a reminder that many destructive transients are the result of electrostatic discharge.
Part of any attempt to minimise failure rate must be to examine the detail of the track design. Ensuring that the circuit operates correctly from the signal integrity perspective is beyond the scope of this module, but you will need to make sure that the tracks you design are able to withstand both normal operating currents and any anticipated higher currents during start-up or fault conditions. The key is to ensure that the temperature rise on the conductor is kept within bounds.
Before reading further, try and identify those factors that you believe to determine the temperature rise of a conductor. Your list is likely to start with the three items on which IPC-2221 is based.
Review your answer as you read the Coretec paper at this link (PDF file, 358KB). Hopefully having read the paper you will have a better understanding of the issues. In this particular case it seems that the industry has been too conservative in its design of traces.
Alternative background on this topic can be found in Track resistance.
Another design factor important for the reliability is the spacing of conductors. Here the guidance document is Section 6.3 of IPC-2221. The clearance requirements depend not only on the peak voltage, but also:
Note that the kind of failure which IPC-2221 has in mind is corona and other forms of high-voltage, high-current complete insulation breakdown, rather than the deterioration of surface insulation resistance. Be aware that there are many high-impedance circuits where even a small reduction in isolation may result in parametric failure, so spacings well above those in the standard are recommended.
Humidity creates the potential for failure for components and systems, as well as at the board surface, so components are almost always provided with local protection, and in some cases the assembled board may also be given a conformal coating for protection. If you have not already consulted the document in Units 1 or 2, we suggest reading our paper on Humidity protection.
At the system level, protection against humidity is generally considered in the context of protection against intrusion by solid objects and protection against liquids, combined to give a “International Protection” (IP) number. As our paper Ingress protection explains, the higher the number, the better the protection given. But of course this comes at the expense of a higher unit cost. When you devise a system and select its enclosure, be careful to understand the needs of the end-use environment, and resist the temptation to over-specify.
Notice that, at the equipment level, we have not considered the possibility that an enclosure may be hermetically sealed, because there are considerable difficulties in isolating a large enclosed volume from its surroundings. In fact, it is generally good practice to provide “breathing holes” or equivalent in order to reduce the possibility of moisture condensing on internal parts, for example where an enclosure in a humid environment is subjected to a significant reduction in surrounding temperature.
Circuits and systems may fail because of electrical over-stress, or exposure to humidity, but they may also fail mechanically: some of the issues are discussed in our paper Protecting against mechanical failure. When evaluating a proposed design, one should always consider the possible ways in which failure may occur, having regard both to the intended use and possible user abuse of the product. Also, do not forget that individual board assemblies are susceptible to mechanical damage during removal from the enclosure and subsequent reinsertion.
Although detailed consideration is beyond the scope of this module, one should keep in mind the possibility of damage from repeated exposure to mechanical stress, and the way in which comparatively small stresses may be amplified as the result of mechanical resonance. Particularly in severe environments, or hand-held applications, additional stiffening may be required, combined with “cushioning” or other constraints on delicate features.
Explain to one of your colleagues how he/she might approach the design of a high-reliability communication product and the subsequent verification that it will survive its intended hand-carried military application.
Compare your answer with this one.
“If men do not learn from mistakes of history, they are condemned to repeat them”. Failures can be beneficial, because understanding the reason for failure can help us prevent it next time – merely burying our heads in the sand may result in a product that fails to meet customer expectations and will damage your company’s reputation.
So feedback needs to be in place to pinpoint the cause of defects during service as well as those that are identified during manufacture. Unfortunately, in many cases company systems are focused on repairing product (internally) or repairing/replacing product (externally) in order to meet the needs of the customer. As a result, much potentially useful information is lost forever.
Of course, making changes to culture and practice within a company may not be within your remit, though you can always encourage good practice. However, when it comes to identifying the cause of failure, you should be aware that this is not necessarily an easy task. Despite careful work, perhaps as many as one circuit in three will refuse to yield its secrets, even to a determined and structured approach. And a determined and structured approach is what this kind of analysis entails!
In case you are involved in doing this, or organising it, we have linked here (PDF file 257KB) to a brief introduction to failure analysis. You should however be aware that this is a task requiring substantial equipment and expertise. We recommend Perry Martin’s book Electronic Failure Analysis Handbook (McGraw-Hill, 1999, ISBN 0070410445) as well worth reading.
For large structures, civil engineers design for ‘fail safe’: that is, they make sure that the load can either be taken by other parts of the structure or the effect of failure otherwise allowed for, until the failed part can be detected, repaired or replaced. For example, look at how many wires join the top and bottom sections of the Millennium Bridge across the Tyne! Apart from the aesthetic aspects of the design, these afford multiple redundancy.
The idea of ‘fail safe’ also translates into ways in which we can tackle failure in electronics. Although this unit has concentrated on preventing failure, we have already seen some evidence of the fail-safe approach, such as capacitors with in-built fuses. A key tool here is Failure Mode, Effects and Criticality Analysis (FMECA) (or simply FMEA), which is a formalised design review technique that focuses the development of products and processes on areas that will reduce the risk of product field failure.
[ back to top ]
So far our discussion on the various “Design fors” has not considered safety-related features, although these are an important aspect of any product with which consumers may come into contact. The legislation that applies within the EU internal market is based on Directives 1985/374 and 1999/34, which are aimed to create what the Commission refers to as a “fair balance of risk among citizens and producers”. However, in practice, the onus is on the producer (and in this case distributor also) to place on the market only those products that are safe in normal use, or in any use that is “reasonably foreseeable”, which obviously includes likely abuse.
The producer shall be liable for damage caused by a defect in his product.
Council Directive 85/374/EEC of 25 July 1985
. . . concerning liability for defective products
The EU requirements are implemented in the UK in the General Product Safety (GPS) Regulations 2005, and applied to all products, both new and second-hand, that are used by consumers, whether these products are intended for them or not . . . scope for litigation here! Details at http://www.dti.gov.uk/ccp/topics1/safety.htm.
Not only do we have to create safe products, we also need to have systems in place to deal with any safety issues. So the legislation has procedures both for the notification of unsafe products and for product recall. You will have seen increasing use of such procedures reported in the press, although most instances are non-electronic. However, from the electronic product perspective, everyone in the supply chain needs to have sufficient traceability on the design changes and the materials used in order to be able to identify (if needed) the specific items to which any safety-related problem refers.
An additional requirement for electronics concerns the identification of products as complying with applicable regulations. Whilst you will know from your studies elsewhere that there is no such requirement to identify products as lead-free, there are many other Directives where CE marking is mandatory, the application of the mark advising the consumer that the product to which it is applied meets the necessary criteria. These criteria are articulated in a wide number of Directives, a frightening list of which is given at http://www.atlanticbridge.co.uk/5.htm.
The Directives most applicable to general electronic products are the Electromagnetic Compatibility Directive (89/326/EEC), which is a subject on its own and covered in the AMI modules on Signal Integrity and EMC, and the Low Voltage Directive (73/23/EEC and 93/68/EEC). Again this safety-related Directive applies both to new and second-user products, and is associated with an impressive list of background material at http://www.dti.gov.uk/strd/lvd.html.
Given the importance of safety-related issues, it is not surprising that this topic has risen in importance in this increasing litigious age, and designing for safety has to be built into the overall management of risk within a company. Such considerations are generic, rather than applying specifically to the manufacture of electronic products, but we have provided a short paper on Risk management which stresses the need to understand the kinds of risk and manage them appropriately, rather than just rely on insurance!
[ back to top ]
Most of the test concepts that we considered in Unit 7 are applicable to the overall system, although the actual tests are less detailed, and the scale of inspection broader. Particularly with larger equipment, the practicalities of environmental testing can require significantly large test rigs, and even simple environmental chambers are costly.
In Unit 7 we introduced the idea of test coverage, and similarly with systems we have to apply a range of different screens in order to pick up all faults, basing our choice of screens both on their cost and their ability to pick up defects. As with a board assembly, it is equally important to identify the likely fault spectrum of the enclosure.
Fortunately, by ensuring that the individual elements brought together in the final equipment assembly have themselves been tested, we can restrict our tests to the overall equipment, and the way that the different parts interact.
Note that, when carrying out an electrical test, it is important not just to verify that the interconnection works and that the overall system function is correct, but also to take account of any unwanted interactions. As examples, these might take the form of signal integrity issues or noise, caused by faulty grounding, or harder-to-trace effects, such as overheating of the entire assembly or acoustic feedback from mechanical ancillaries such as cooling fans. There is no substitute for an enquiring mind on the part of the test engineer, especially at the prototype stage.
As well as testing for functional performance, we also need to consider the overall impact of the product on the end-user. Where that end-user is a non-specialist, then the type of “customer experience test” described at the end of Unit 7 has become increasingly important. But even with a business-to-business sale, getting the packaging right and ensuring that equipment is accompanied by appropriate documentation, manuals and any specialist connectors, can yield significant benefits by enhancing the all important “first impression”.
Although less important when equipment is installed by the manufacturer and subject to formal handover to the customer, it is always helpful to ensure that the equipment has the right cosmetic appearance. Particular attention should be paid to the effectiveness of protective packaging for panels – shipping a pot of touch-up paint is often a welcome feature, but should not be necessary!
As well as these generic equipment test issues, three specific topics apply at the system level, to confirm that the complete equipment meets the required standards of safety, reliability and over-range performance.
Whilst most safety considerations can be dealt with at the design stage, it is normal to subject at least a sample of products made to “high pot” testing, to ensure there is no possibility of the product harming the user because of insulation breakdown.
There may also be a requirement to check that the equipment is tamper-proof, and that no part of the exposed circuit can be touched by the operator. This kind of test uses simulated “test fingers” (either jointed or unjointed) as specified in IEC publication 335-1. IEC specifications are not available on the ATHENS system, but your access to British Standards On-line will show this diagram on p.2 of BS6148-8.
In an earlier reference to our paper Simulating the real world in a test house, we highlighted the wide range of possible environmental conditions and the way in which these can be standardised for the purposes of test. Testing a product to ensure satisfactory operation under severe conditions is a topic that has received considerable attention over the past 60 years, particularly in relation to military and aerospace requirements.
Read the section of the paper entitled Testing the product to gain an appreciation both of what is involved in testing and of its limitations. Note in particular the difficulties associated with making tests realistic and affordable.
In many cases a full environmental evaluation of a product will only be carried out at a prototype stage, and performed by engineers rather than carried out to a formal specification. It is particularly important therefore to ensure that prototypes represent in all respects the quality and build standard of the final product, so that false conclusions are not drawn from encouraging preliminary results.
In the same way that, except in high-reliability situations, environmental testing is rarely carried out on all systems made, the reliability of the product is usually built in by making appropriate design choices rather than directly assessed.
One reason for this is that, in order to get useful information early enough in the design cycle, the reliability test needs to be accelerated significantly and the results then extrapolated to estimate the product life. The main concerns here are:
As with all testing, a balance has to be struck between the cost of testing and the cost of not uncovering unreliability. More information in our paper Assessing product reliability – even if you don’t read the whole paper, it may be worth considering the question posed in the final SAQ.
In some cases we can expect early-life failures, although this problem is more significant with sub-assemblies than with completed equipment. Where such failures occur regularly, reliability of the shipped product can be improved by screening.
As with components, “burn-in” is the most common form of screen regime at the equipment level, although temperature-cycling and power-cycling are more effective, and a very strong case can be made for Environmental Stress Screening, which combines a number of these elements, with mechanical and environmental tests. The sequence chosen relates to the product operating environment, but seeks to apply accelerated conditions so as to stimulate early failure.
More information about this in our paper Improving reliability by screening. There is a more extended treatment of environmental stress screening in AMI4957 Test Strategies, including some discussion of the idea of a “design margin”. There is also a justification of the ESS principle, on the ground that the reduction in expected life for a good unit is only minimal (of the order of one year in 25), whereas the beneficial reduction in early failures is considerable.
We commented before that the testing that is carried out will reflect the requirements of the industry in which the product operates, in the same way that each industry will have norms and expectations for design practice and build standard. Nowhere is this more true than in the automotive industry. Observers elsewhere have noted a general trend from company and national specification towards industry and international specifications, such standards as BS EN 60721-3-5, whose object is to classify “the environmental parameters and severities to which a product will be exposed when installed in ground vehicles”, specifically excludes products which form part of the vehicle.
As well as having independent specifications on environmental and other tests, the automotive industry also has its own (tighter) quality management standards. For example, ISO/TS 16949, which combines ISO 9001:2000 with automotive-specific requirements from the American QS-9000 and German, French and Italian quality standards. For a useful introduction to this topic visit the Perry Johnson site at http://www.pji.com/iso_standards.htm.
If you are interested in learning more about automotive issues, visit the sites of the Society of Automotive Engineers at http://www.sae.org/ and the Automotive Industry Action Group at http://www.aiag.org/.
The selection of tests for equipment will be subject of negotiation between designer and end-user, bearing in mind the capability of the assembly and test team. In most cases, the work does not need to be carried out from scratch on each occasion, but can reflect prior experience, mediated by the norms and expectations of the end-user as concerns design practice and build standard. Both aerospace and automotive industries offer examples of such generic standards. However, we should always bear in mind the possibility that a design may have its own peculiarities, and be prepared to modify tests appropriately.
[ back to top ]