Design has a major effect on manufacture – how often have you heard a production manager complain about a design as being difficult and over-expensive to manufacture, whereas, with a little bit of thought . . .
Our title reflects what is often this sort of negative influence, but that is certainly not what we wish to promote as best practice, which is that everyone involved in the pre-manufacturing stages should be aware of how to optimise the performance of their manufacturing partners. The fact is that design and manufacture are inextricably linked, and need to work as a team. Sometimes this is just a matter of sharing information in a timely manner; more often the process is of discussion and compromise, hopefully carried out as early as possible in the product development process.
Understanding the needs of manufacturing partners is so important that we will be devoting the whole of Unit 8 to discussing the requirements for manufacture, looking at specific fabrication, assembly and test needs for information, and stressing the importance of everyone being fully conversant with their suppliers’ processes.
In this Unit we are taking a wider picture, starting by exploring the impact that apparently simple tasks like choosing components can have on the manufacturability of a product, and how one decision leads to others. From this topic, where the need for mutual understanding and cooperation is evident, it is no distance to looking at the Design for Excellence concept – “DfM on steroids!” – which underpins a more general discussion on best practice approaches to Design for Fabrication and Assembly.
Finally, because the world is not just design and manufacture, we are looking at some other “Design fors” of importance, such as Design for Environment and Design for Obsolescence.
In this section we are looking first at how the choice of components can affect assembly processes, board design and board specification. Then, conversely, we will see how the part chosen has to be specified to meet the requirements of the process. In some cases, this will need to be an iterative procedure.
The choice of assembly process has to be made on an individual basis, and depends most critically on the characteristics of components – their availability, cost and mechanical format and their ability to withstand processing. Some of the many possible process routes are shown in Figure 1: the option selected will also be influenced by design constraints, the capabilities of the equipment and the requirement to balance the line.
The choice of route is primarily determined by whether or not there are any through-hole components. The choice is particularly straightforward if you have:
Where there are relatively few through-hole components, and the circuit is otherwise heavily populated with a wide variety of surface mount components, your choices lie between hand soldering, selective wave soldering or intrusive reflow. As Table 1 comments, these last two processes are not always available to you – you certainly need to consult your assembler to understand their preferences and recommendations.
|assembly process||hand soldering||wave soldering||reflow soldering|
|through-hole||can be used for all types of through-hole component||all but those with very heavy leads can be wave soldered, but the parts need to be held down during soldering||intrusive reflow†|
|surface mount||difficult, especially for small components: used only for rework||feasible for smaller components, provided that they are first glued in place||preferred process, which can be carried out on both sides of the board|
|mixed technology||prototypes only||many Type III assemblies use reflow on the top side, and wave soldering on the underside|
|where through-hole components are few or large, use double-sided reflow, then hand soldering or selective wave soldering†|
† indicates processes that are not available from all assembly houses
Those who have considerable experience of electronic assembly will realise that Figure 1 and Table 1 represent considerable simplifications of the many process routes available. And it is worth keeping in mind, especially early in the design phase, that in many cases there are radically different alternatives to consider, and questions to ask. For example: Are there better ways of partitioning the circuit function, perhaps using modules? Would the circuit be more effectively made by choosing a different1 level of silicon integration?
[ back to top ]
The board design is heavily influenced by the type of component, as indicated in Table 2. Generally through-hole components are confined to low-density applications and relatively simple boards, but the comments in Table 2 on traces and layer count would not necessarily apply if the through-hole components were complex connectors: in that case many layers of interconnect may be needed, purely in order to provide access to all the pins.
|board issue||device footprint||traces||layer count|
|through-hole||each lead has to be associated with large through-hole, but dimensions are critical
need clearance around lead/body for insertion (especially automatic)
radial types need allowance for lead bending2 radius
|potential for connections from lead on every layer
generally low density → relatively simple board → wide tracks/gaps
|low density → relatively simple board leads can bridge over traces and connect between layers, further simplifying designs: single-layer can be feasible|
|surface mount||soldered area usually independent3 of through connect area of pad (and solder paste)
determines volume of solder joint, and thus dimensions are critical.
|inter-layer connections made by vias, not leads
generally high density → relatively complex board → narrow tracks/gaps
|high density → greater complexity, so more layers of finer traces|
2 Wires should be clamped next to the body, then carefully bent, in order to avoid internal shock damaging the component.
3 Unless ‘via-in-pad’ is used for high density boards. More detail in Unit 10.
Again, those familiar with high-density board designs used to accommodate Chip Scale Packages and similar will realise that Table 2 is somewhat simplified. However, the general relationship between device sophistication and board complexity still holds – using the latest component technology may have cost implications in terms of both layer count and minimum track-gap widths.
[ back to top ]
As indicated in Table 3, the choice of component reflects on the choice of assembly process, and consequently on the board specification. Higher specifications are needed for reflow soldering than for hand soldering and wave soldering. As a result, most designs for reflow soldering use epoxy-glass laminates (such as FR-4), whereas commercial applications with through-hole components will use the cheaper phenolic-paper or composite types wherever possible.
|board spec|| laminate material
|| surface finish
|assembly process||component type|
| hand soldering
||through-hole||laminate only has to withstand quite low temperatures: can use paper-phenolic types||only has to be marginally solderable, as hand process can ‘scrub’ the joint surfaces|
| wave soldering
||through-hole and mixed technology||laminate has to be flat, but only contacts the wave for a short period, so can use most paper-phenolic and composite boards, provided that they are flat||preferably should be easily solderable, but process can use relatively active fluxes, provided that residues are removed|
| reflow soldering
||surface mount||whole volume of laminate has to withstand reflow temperatures for an extended period: requires epoxy-glass laminate or better||surface has to be both solderable (limited time available to reflow) and flat (for paste printing)|
When we come to consider components that have high power dissipation, this may affect the choice of board material and of the metallising. When components are large, and intended for direct mounting to the board, such as LCCCs and flip chips, then we may need to use a construction with controlled CTE in order to make the resulting assembly able to withstand the severe temperature excursions of military life.
The surface finish of the board will be dictated both by the budget and the requirement: for through-hole components, and surface mount components of moderate complexity, a solder-levelled finish will be adequate; as soon as one moves to fine-pitch components, a flat surface is required. But whether this flat surface should be the cheap OSP, or more expensive plated surfaces, will depend on whether or not the design can be soldered in a single pass. This is a complex issue, as you will have noted from your study of Unit 1.
[ back to top ]
Components need to be able to withstand the assembly experience! Depending on the type of component and choice of process, this will involve some exposure to liquid solder and flux, and may involve immersion in quantities of liquid solder and a cleaning medium (water or solvent based).
Although the reflow process is carried out at a lower temperature than wave soldering, the time of exposure to liquid solder is much longer, and the whole of the component body reaches the soldering temperature, both of which impact on the temperature rating of the part.
Components are usually rated in terms of their ‘resistance to soldering heat’, tests of time against temperature which are carried out to international standards. In the days of tin-lead eutectic solder, typical ratings were 10s at 260ºC and 60s at 235ºC, for wave soldering and reflow soldering respectively (Table 4), but there has been a tendency for these requirements to become more stringent, because of the higher temperatures used by lead-free materials. So the T260 test has moved towards the more stringent T288. Note that not all components will withstand high temperatures, in which case hand assembly after reflow may be required, although this comes at a cost.
Other soldering constraints affect both the choice of materials and the physical construction of the part. For example, the wipers on potentiometers and trimming capacitors must protect the internal wiping parts from molten solder, and cleaning solvents must either be excluded entirely or allowed such free access that complete cleaning is ensured. DIP switches are an example of components which may be supplied with the top face of the part covered by a removable tape, which provides a contamination barrier during assembly.
|component rating||temperature rating||metallisation||packaging|
|through-hole||hand soldering without cleaning is suitable even for sensitive ‘non- wet’ components
wave soldering needs 10s at 120°C rating
solder immersion needs 10s at 260°C rating
|no special requirements for hand soldering
machine soldering needs better solderability than hand
|automatic insertion needs reel or bandolier for taped components; tubes for DIPs
otherwise can come in any form convenient for manual assembly
|surface mount||needs 235°C/60s rating||needs good wettability
exposure to molten solder is far longer than with through-hole
|wide range of standard tapes and trays suit all components; sticks also for smaller volumes|
|mixed technology||depends on soldering process (see above)|
The metallisation of the component termination is not usually an issue, provided that components are solderable, and resistant to exposure to molten solder. However, with some ceramic components the termination is not a lead but metallisation on the surface. Some parts are very similar in performance to the early chip ceramic capacitors, which lacked a nickel barrier layer and were prone to the metal content in the metallization dissolving in the solder. What remained on the component was the glass matrix of the metallisation, with little or no solderability . Where components have such a propensity to de-wet, special care has to be taken in processing, so the best way to make friends with your production team is to avoid these components at all costs!
Table 4 also includes some comments about ‘packaging’, referring in this case not to the format or construction of the component body, but to the way in which the finished components are presented to the assembly equipment.
Where it is decided to clean the assembly after soldering, there are two main implications for the components:
Variable components such as switches present the most obvious challenge from the point of view of cleaning agent ingress because contamination carried into the device by the cleaning agent can result in intermittent switch action, particularly if circuit voltages are low. For this reason, a number of products are supplied with a cover tape which has to be peeled off after all assembly stages have been carried out. The alternative, for sensitive components which are not sealed, is to treat them as what the assembly industry refers to as ‘non-wets’, that is adding them at the very last stage, and hand-soldering the leads, and at most cleaning just the pad areas by local application of cleaning agent.
Your company is redesigning an early 1990s computer-based product that used only through-hole parts on a six-layer PCB. The microprocessor was a 132-lead Pin Grid Array, with thirty support ICs in dual-in-line format; there were many connectors, several crystals and some board-mounted switches; the integral power supply contained a transformer and aluminium electrolytic capacitors.
The size of the new board is at your discretion, the product dimensions being determined by other factors. All the components are now available also in surface-mounted format except the transformer and large capacitors in the power supply, and the special-purpose crystals.
Describe to your colleagues some of the options available for the components, and explain the implications of this choice for board design and for component and board specification. If you are able, feel free to suggest areas for circuit redesign which would cut cost and/or complexity.
Component choice will not only affect the printed circuit board assembly. If we consider the mechanical components that come together to make an enclosure, to take one example, then the choice of equipment practice affects the assembly process and cost, as well as determining the final specification. “Would it be better to have a custom-moulded enclosure, or modify a standard box?” “For the small quantities we are making, should we wire-wrap the connections, or design a back-plane?” These are the sorts of question that demand dialogue between the parties, supported by an understanding of the issues.
The design itself is therefore influenced by feedback from manufacture, and not just by from process experts, because the decisions agreed on materials and components will reflect back on the design.
[ back to top ]
The interaction between design and production is very much influenced by the culture of the company, in particular their shared beliefs concerning continual improvement. We believe it important that there should be a concerted effort to use the whole raft of “design for” methods to achieve improvement, and ideally excellence. Which is why we like to use the term “DfX”: not only does it encourage a higher aim, it is a reminder that X is a variable term that can have many meanings and thus stimulates lateral thinking.
Here an example might be helpful. It comes from Dell, a well-known manufacturer of computers, an innovator in the way that it does business, and in consequence a world leader. One recent initiative has been the emphasis placed on a totally different approach to the quality of the product; Dell are reported as expecting their suppliers to deal with any failure issues, and concentrating their efforts on a “customer experience test”. Instead of examining the elements that make up the system, and focusing separately on each, Dell have attempted to replicate the customer experience of their product, by starting with the package as delivered to their customers. So the quality engineer is not looking at the build standard of the board, or the accuracy of the batch traveller, but rather seeking an answer to questions such as “Did I get the variant I ordered?” “Did the packing case contain all the bits that I needed, clearly identified?” “Was there enough information for me to be confident about connecting everything up?” “Did it work when I turned it on?” “Does the keyboard have a satisfactory ‘feel’?”
Looking at everything from a customer perspective, and being prepared to consider problems that fall outside the normal “tick boxes” are aspects of Dell’s example that we can generalise to our own situation, wherever we are in the manufacturing chain, and whoever our customer might be.
Of course, the DfX concepts is neither our own nor particularly new. However, if it is new to you, we recommend that you read our introduction to the topic, The DfX concept, which looks at common perceptions of DfX and its use as a methodology to enhance development and reduce risk.
[ back to top ]
In this section we are trying to develop appropriate DfX concepts for electronic assemblies at an outline level, intending to cover Design for Fabrication (bare board) Design for Assembly (board and system) and Design for Test/Testability in greater detail in Unit 8. These are important areas of study because of the complexity of PCB assemblies, the tight tolerances on component, board and process and the reliability demanded of the products for low profit margins.
The situation is made more confusing by the frequent use of the term “Design for Manufacture” to cover all these aspects. It can be a useful shortcut, because we end up with just one checklist, but it is easy to be blinded to the need for the detailed considerations of board fabrication, board assembly and test that will be the focus of attention by our manufacturing partners.
What does that checklist look like? In many cases a full checklist will be enormously complicated, and the list below only shows some of the more typical items, including aspects from PCB fabrication, assembly, repair and test:
With relevant background knowledge, the guidelines on this list seem obvious. But how do we get this knowledge? More important, how do we get sufficient insight to understand why the guidelines have been set up, why they are important, and the extent to which they can be “bent”? The only way of doing this is much greater involvement by all functions, but especially design, in every aspect of the whole product process, from development through production and test to servicing. Unfortunately, there is no shortcut to widening communications. For example, designers should take time to visit board shops and assemblers with the aim of talking to the fabricators and assemblers to ask how PCB design decisions can impact fabrication and assembly yields and costs. They will usually have examples to demonstrate design decisions that have a large impact on final product cost.
Understanding the guidelines as they are is helpful, but electronics is subject to continual change, with new packages and materials continually being developed, and assembly houses are constantly upgrading or adding new equipment that will impact the assembly process.
For the design fraternity, the information gained, and regularly updated, needs to be used to inform the design process. In many cases this will involve making changes to the CAD libraries, so that automatic screening by software reflects the capability of fabricators and assemblers. However, it is still important to use common sense, and realise that a capability statement is only a guideline to what an assembler or fabricator can achieve, and the limits should not be used for every design. As we will see later, pushing processes to their limits is uneconomic: the closer to the limits a design approaches, the lower is the yield, and the higher the cost, and quality may be compromised. For example, if a PCB fabricator can provide boards with 75µm tracks, then do not set this as a minimum limit, or your CAD program will attempt to route every design with substantial areas of 75µm tracking! Use the minimum dimension only when necessary, because, the thinner the track, the more likely it is that there will be breaks in the track, and the resulting poor yields will be charged to you.
Each board fabricator will have their own ideas as to the limitations on each process, and these are generally made available to customers in the form of a process capability manual that offers guidelines to designers. Many of these are available freely on the Internet. There is more information on a typical process capability manual embedded within Unit 8, but such guidelines would put figures to the following suggestions for good practice:
As with PCB fabrication, the assembly process is complicated and poorly designed boards result in poor yields, high rework and low quality. The designer should consider the assembly process during design to reduce these.
Some guidelines to consider are:
There is little point in developing an advanced product if manufacturing cannot adequately test it. Early involvement of test engineers in the product design process is essential so that a clear understanding is obtained of how the product will be tested during the production processes.
Areas that might require clarification include:
A simplified PCB design process is shown in Figure 5. The information given to layout engineers would be product schematic diagrams, a critical component list (with recommended board placement) and possibly a Bill of Materials. The layout engineer then places the critical components as requested and then places the remaining components. The engineer would then route the interconnections between the components. Finally, information such as silkscreen and solder mask is added to the PCB. During the process several analyses would be carried out.
Think about the stages of the process and describe the analysis for each stage, especially with respect to DfX.
Do make sure that you do this activity and look up the answer before reading further.
In that activity, one of the key components you will have identified in the engineering process is that of review. In our experience this is one task that only the most formally-organised companies have the resource to carry out as a team effort. However, even a review by the designer is worthwhile, looking to pick up any major difficulties or cost areas. Here a simple test is where one’s design lies in the continuum between standard practice and state-of-the-art. Like so many commonsense approaches, this is one DfX we have not yet talked about, Design for Simplicity!
Scarlett recommended striving for simplicity: ‘The first point to consider in the design of any multilayer board is whether the board should be a multilayer at all, or whether it should be designed as a double-sided board, possibly of greater surface area. Since the manufacture of multilayer boards always involves all the steps necessary to make double-sided boards, plus a number of other costly operations, a double-sided board will invariably be cheaper than a comparable multilayer.’
Whilst still true, this comment was made nearly 20 years ago, and his advice must be interpreted in the light of a need for higher interconnect densities. In practice, this means selecting solutions which involve the lowest possible technology which will accomplish the task, for example, being aware that there is a trade-off between the track-gap dimensions and layer count. Yield losses escalate rapidly at or near the limits of capability – fine tracks (but, more importantly, fine gaps) cause significant yield losses
Above all else, avoid over-specifying your requirements: Figure 1 shows a generalised yield:difficulty graph, which points up the need, as far as possible, to keep designs well within the boundaries of standard practice. Whilst we have talked about this concept in relation to board fabrication, it is very important to remember that the same issues apply elsewhere: setting a challenge to production professionals will often lead to success but at an unacceptable cost – the electronic equivalent of a Pyrrhic victory, “One more such victory and I am lost”!
At the mechanical assembly level, whether we are looking at board assembly or box build, there are well-established guidelines for optimising cost, quality and productivity that have been developed within other industries. Some sample guidelines are:
Within the wider framework of Design for Manufacture, when we are assembling products we are looking to make them easy and cheap to assemble. Because this is such a big topic, much work has been dedicated to “Design for Assembly”. DfA favours product and component designs that are easy to grip, feed, join and assemble by manual or automatic means, and its main aims are to:
DfA may be carried out manually or with the support of computer based systems available as conventional programs or expert systems. A DfA analysis could be carried out at the concept stages and DfA evaluation on completed designs or prototypes. The former is obviously preferable because changes are easier, but sometimes the DfA engineer has to work with what he is given!
As you are probably beginning to expect, the basic concepts of DfA have been elaborated into fuller guidelines of which this list is a summary of those that are most commonly quoted:
Although Design for Manufacture and Design for Assembly both have cost reduction as one of their aims, “Design for Low Cost” has been identified by some as a specific DfX. This uses the DfM and DfA approaches, but has the specific aim of reducing the cost of the product, rather than just improving its manufacturability.
To give you insight into DfLC, we should like you to review the DfM guidelines for Mechanical Assemblies given above, and suggest how they might be used to provide some guidelines for Design for Low Cost that a layout designer could use.
[ back to top ]
So far we have applied the DfX concept to the manufacture of electronic and mechanical products, concentrating on making them manufacturable by design. However, there are other Design for Xs that frequently need consideration, sometimes very urgently!
Some of those on the list below, Design for Maintainability and Serviceability, Design for Procurement and Design for Reliability, generate guidelines by a similar approach to the one we have seen for DfM and DfA. That is to gather knowledge by consulting experts in the field, and use this during the design phase. There are many guidelines, and which of those are appropriate will depend on the product and its intended use, including its environment and eventual disposal. We have included in our list below sample guidelines for Design for Environment, Design for Maintainability and Serviceability, and Design for Reliability.
Two of the Design for Xs in this list are however, considerably more significant for the electronics professional, and need more than just guidelines to implement. These are Design for the Environment, and our last Design for X, Design for Obsolescence.
Design for the Environment is a topic that has mushroomed in significance recently, as industry takes up the challenge of removing lead from electronics, mandated by the European Union Directive on Waste from Electrical and Electronic Equipment (WEEE) and its associated Directive on the Restriction of Hazardous Substances (RoHS). This is a topic dealt with at some length in AMI4982 Lead-free Implementation.
But be aware that the lead-free issue is not the only area of difficulty ahead. For example, we have already mentioned the pressures to reduce VOC emission, and there are Directives in progress on energy use and design for disassembly. The only way of dealing professionally with these is to manage the company’s environmental response in the same way that a company has a quality system and health and safety procedures. The appropriate standard in this case is ISO14001, and certification to this standard is becoming increasingly important to electronics companies.
The kind of guidelines that one might use within DfE are suggested in the list below:
It is important when looking at guidelines like this to give thought to all the processes involved, in manufacture and in service, to any waste generated during manufacture, shipment or use, and to the final disposal of the product. Here the most sophisticated users will carry out a full environmental impact assessment. If this aspect of design and management appeals, take a look at AMI4982 Lead-free implementation Unit 1: Background to the lead-free revolution and our paper on Life cycle thinking.
For more information on environmental management, read An ISO 14001 overview
[ back to top ]
Field service and logistics staff should be actively involved in the product design process so that aspects of service and maintenance are addressed. A product that is well designed for maintenance can save a great deal of money in its later life.
In some products (for example, plant equipment essential to a manufacturing line) maintainability may be one of the highest priorities, but it should be considered for all products, even if the eventual decision is that a product should not be repaired but simply scrapped if a failure occurs. Areas for consideration might include:
[ back to top ]
In Design for Procurement, product designers work effectively with suppliers and sourcing personnel to identify and incorporate technologies or designs that can be used in multiple products, facilitating the use of standardised components to achieve economies of scale and assure continuity of supply.
You won’t find too much of help from a web search, so the only way to obtain Design for Procurement guidelines is to speak to component procurement and sourcing staff, try to understand the issues they face and the problems they encounter, and discuss how to prevent these. Some guidelines are:
Finally, bear in mind that assemblers always prefer to start with a ‘clean kit’, containing all the parts needed to complete a build. It is much more expensive to add even one missing component at the end of the process, and there may also be unreliability implications in doing so. ‘For the want of a shoe . . .’!
[ back to top ]
Using simulation tools at an early stage in the product design process can model the performance of each component and module at the extremes of its environmental and manufacturing specifications. This is something that cannot be done using prototypes, so it provides a valuable insight into the reliability and repeatability of a product once it has entered manufacturing.
Areas that should be considered under this topic include:
[ back to top ]
Rather than reinvent the wheel, we are encouraging you to read the article Guarding against component obsolescence by Graham Prophet, which you will find on the EDN website at http://www.edn.com/ in the archives for November 14th 2002. The PDF version (file size 182KB) is at http://www.edn.com/contents/images/257048.pdf. This gives a comprehensive description of obsolescence in the electronics industry that you can use as a resource for the following activity, even though it was written some time ago, so prudence suggests that you should assess the information against more recent articles.
For what reasons will a component manufacturer make a product obsolete?
Traditionally, the military were the most concerned with obsolescence, due to the extended duty lives of military products (usually more than a decade). This has changed, and consumer electronics manufacturers are now just as concerned. What are the reasons for this?
What options do manufacturers have for replacing obsolete components?
What steps can an organisation take to prevent obsolescence?
It’s very easy to take the view that designing for obsolescence is of interest only to those who make products for such markets as military and aerospace, where design time-frames are extended, and the working life of a product is expected to be in tens of years rather than tens of weeks. However, the same issues are increasingly occuring in more commercial markets, and the consequences can be more serious, because there is a higher volume and less profit margin. It is not uncommon for an exciting new integrated circuit to be introduced, fail to realise its potential, cause some grief to the manufacturer, and be discontinued, all within the space of a few months. Where does this put the eager young designer who saw in that component the answer to the proverbial “maiden’s prayer”?
The fact is that, whatever one’s market, obsolescence needs to be designed for. It is true that the evil day can be postponed by designing only with components that are second sourced, are neither state-of-the-art, nor of a type likely to be superseded by foreseeable progress – bought any 16k RAM recently?! But even for components that are apparently well-supplied, the business aims of component suppliers do not always allow them to continue manufacture, and sometimes parts are discontinued with little notice.
Further information on this topic can be found at the web site of the Component Obsolescence Group (COG) at http://www.cog.org.uk/, and membership of this community is recommended for all those who are involved with long-life products. If you look at the subjects of their regular meetings you will find that obsolescence affects a very wide range of the mechanical and electronic elements that build a product. And as for software . . .
COG have kindly made a number of their publications available to students for on-screen reading. All are interesting and informative, and may be accessed at these links:
Date Coding Minefield (pdf file; 673Kb)
Long Term Storage Minefield (pdf file; 2,692Mb)
Obsolescence Minefield (pdf file; 438Kb)
Pb-Free Minefield (pdf file; 679Kb)
Redundant Stock Minefield (pdf file; 569Kb)
Supply Chain Minefield (pdf file; 662Kb)
“The threat from component obsolescence continues to increase, as equipment once scheduled for retirement still continues in service.”
“One observation I will make is the power of being pro-active. By planning ahead, managers can implement low-cost solutions early, and budget for modernisation to take care of problems whenever the business case determines it is time to refresh, redesign or modernise.”
Walter Tomczykowski, 29 June 2005 European/United States Obsolescence Conference
[ back to top ]