| || |
Quality Control in Surface Mount Technology
The Meaning of Quality
Quality is a term which is misunderstood by many. It encompasses several disciplines but what it boils down to is giving the customer what he expects. The end user is not the only customer, however. The production area is the "customer" of the purchasing department and the sales activity is the "customer" of the production area.
There is no point in designing a product which can't be manufactured and there is equally no point in designing a product which looks so bad or functions so poorly that the Sales department will never sell it. Yet a lot of designers do just that and develop products which are difficult to assemble, or virtually impossible to test, or products which can never be wholly reliable.
Basic Principles of Product Design
These design principles apply to all technologies; not just Surface Mount.
1. Must meet its specification.
2. Must be built at minimum OVERALL cost.
3. Must be simple to assemble.
4. Must be simple to test.
5. Must be reliable.
1. The specification is fundamental to the design of any product, yet many companies allow their designers to forge ahead with the development of a new product with only the hint of a specification in someone's head. Without a specification, how do you know when you've achieved the object? The project is doomed to last for eternity, becoming ever more complex and expensive as time goes on. You MUST have a specification.
Questions which need to be addressed include:
What is the actual function of the product?
What is the maximum price we can sell it for and what, having deducted our earnings margin, is the maximum build cost?
What environmental conditions must it withstand?
How long must it function reliably?
What are the constraints on size/weight/appearance?
What safety requirements must it meet?
Does it need approval before it may be sold?
How do we cope with returns under warranty or servicing agreements?
How soon must it be ready for production before the market opening vanishes?
Are we aiming to sell abroad, in which case do we make it compatible with overseas requirements or design different models for export?
There are lots more questions which must be answered before the actual design work begins but these are a few examples. The Quality of the Design relies upon its ability to meet the specification. Without a specification then, by definition, you simply can't achieve good quality!
2. There is no point in developing a product which will cost substantially more than its rival. Designers have a tendency to think only in terms of the component costs but this is a very blinkered outlook since the end cost depends upon many factors. To save a penny on the purchase cost of a component may have catastrophic repercussions if that component should fail at an assembly stage where the fault is difficult to find or, worse still, should fail in service and need repair under warranty. Quality Engineers talk of the "Ten Times" rule which states that, on average, the cost of test and repair will increase by a factor of ten at every stage of assembly. By this reckoning, our penny component could cost the firm `100 or more if it should fail after the product has been sold. Care must be taken at the design stage to ensure that only components which are inherently reliable are used. That might mean purchasing items which have been tested under rigorous environmental conditions or it may be cheaper for the assembly manufacturer to perform his own tests, on receipt, in order to weed out potentially unreliable devices.
3. Another major factor in the design of the product is the ease of assembly. For instance, it may be cheaper to use a clip costing 5p than to use a nut, bolt and washer which together cost only 1p but take 50 times as long to assemble. I know of one firm which, instead of using a double-sided plated-through P.C. Board, opted for a double sided board without plated holes and used pins soldered through to join tracks on both sides. The P.C. Board cost only 25% of the plated-through price and the cost of pins and labour accounted for only another 25%. On the face of it, this was a sensible decision which halved the cost of the board. Unfortunately the extra 200 soldered joints per board, which had to be made by hand, so badly affected the reliability of the equipment that the first production run had to be scrapped! It's not easy to make the right decision on every occasion.
4. Testing the product is usually the last thing on the Designer's mind. Unfortunately, testing is also one of the biggest sources of problems in manufacturing industry, today. Frequently, items such as test pads are added to a P.C. Board at the last moment. They seem to be stuck there as an afterthought, and it shows. If you think about it, not only is the
P.C. Board about the most expensive electronic component to purchase, but also the most costly to replace after assembly! Test pads serve two purposes. Firstly, they greatly simplify the testing of the bare board for "shorts" and "opens" by the P.C. Board manufacturer. Secondly, they permit in-circuit checking of the assembled components themselves. Strictly speaking, it is possible to do away with in-circuit component testing and use only functional test. This method has the drawback, however, that if for instance a microprocessor refuses even to run because of a faulty pull-up resistor, then the functional tester will be unable to locate that fault or any other for that matter. To some extent this disadvantage can be overcome by processor simulation but functional test still has great difficulty in identifying the exact component which is creating a problem.
The in-circuit tester, on the other hand, will readily identify a faulty discrete component in most instances. To sum up; the designer should consider making room for test connections at an early stage. He should also try to make the design modular so that the circuit is split into a number of functionally separate areas to simplify testing. Most importantly, he should discuss the test requirements with the people who will be responsible for that job, early in the development of the product.
5. The reliability of the design depends on a wide range of factors. The choice of components largely depends on the anticipated use of the product and sometimes the ingenuity of the customer can cause surprises. One car manufacturer, for instance, had a lot of warranty claims by a certain police force which had purchased a large fleet of cars. The torque converter plate kept breaking. In fact it was being wrenched right off the crankshaft. Only when the car manufacturer sent out a test driver to accompany a police patrol did they discover what the problem was. The police car driver in hot pursuit of a suspect would occasionally miss a road turning. Now, for most of us, that would necessitate braking to a standstill, a quick "U" turn and head back for the missed turning - but these were Class 1 police drivers. They discovered that the quickest method was to shift the auto box into reverse (at 50mph!) and sit with pedal to the floor whilst the forward motion was converted into rearward acceleration. This is a good example of the failure of a manufacturer to anticipate the mis-use of his product and to build in appropriate safeguards. We have now come full circle back to the Specification. Unless this mirrors exactly the anticipated conditions of use, then the Quality requirement will not be met.
Specific problems with Surface Mount Technology
There are numerous companies which are currently changing over to Surface Mount Technology and it is extremely difficult to obtain truly independent advice. Every supplier says that his components are the best and there is a notable lack of standardization in component sizes. In addition, each supplier will give slightly different advice regarding appropriate pad sizes, shapes and spacings and process parameters. Quality is very much dependent, therefore, on the control of design from the very start. It is of paramount importance that components are suitable for the eventual use of the finished equipment but designers often overlook the need for the component to match the actual assembly and test processes which will be used in production. You may laugh at this, but I've seen boards designed to take I.C. carriers, which have "J" leads out of sight beneath the plastic moulding, being soldered by Infra-red reflow methods. If you don't quite see the joke then console yourself with the fact that the designer couldn't either. In fact infra-red radiation simply will not pass through solid plastic and consequently, in order to melt the solder, the machine was turned up to maximum output. The effect was that eventually the air was heated sufficiently to melt the hidden solder but, by this time, the printed circuit board looked like a piece of burned toast! Infra-red reflow methods can be used only where the termination can be "seen" by the radiating elements. The inter-spacing of the components is important because you must avoid shadowing effects. If you really need to cram the parts on the board you might have to use vapour-phase reflow methods but this can be considerably slower and, in view of the rising price of Chlorinated Fluorocarbons, possibly far more expensive. It also raises another important design question which is - "If I cram the components closer together, will they still be able to dissipate the heat adequately?" If not, then a reliability problem exists and you will not meet the Quality requirement that the customer expects his equipment to work and to keep on working.
Shadowing effects must also be considered if you intend to use wave-soldering methods. Imagine a number of large rocks in the middle of a fast-flowing stream. The height and turbulence of the water hitting the rocks downstream is influenced by the size of the rocks in front. Flow soldering produces a similar effect and, although the problem can be minimised by use of a turbulent wave or, better still, a vibrating wave, it must still be uppermost in the designer's mind when he begins his board layout. Unless his first question is "in which direction will my board pass over the solder wave" then he is relying entirely upon LUCK. You might wonder why anyone would consider wave soldering when Infra-Red and Vapour-Phase appear to give fewer problems. The answer is that, if you intend to use conventional leaded components as well as SMDs then wave soldering is the only option other than hand soldering.
BS5750/ISO9000 Removes the need for Luck
The achievement of Quality depends on management commitment. One of the best ways to begin is for management to work towards BS5750 Certification. Part 1 of this Quality Standard helps to ensure that all design and manufacturing work is controlled by carefully laid down guidelines, often called "Procedures" or "Work Instructions". To be effective, these guidelines must be simple as well as practical; that is to say, it must be possible to carry them out satisfactorily without adverse effect on safety or on other work and they must be easy to understand. It should go without saying that all personnel should be instructed in the understanding and use of these guidelines as part of a controlled training programme. In addition, all relevant guidelines should be made readily available to the appropriate people, and the issue of such documents must be controlled in order to prevent unauthorized changes. BS5750 part 1 stipulates the ways in which such controls should be instituted to ensure that every person is aware of his task and the method whereby he must achieve his objective and also that every process is known and variability or errors minimized or, preferably, eliminated.
The factory which achieves the control envisaged by BS5750 could be likened to a ballet or an opera where every performer knows exactly where he will be at each point in the performance and where the stage lighting manager will ensure that every spotlight or floodlight is switched on or off at the precise moment that the action demands it. On the other hand, a factory without such control could be compared to a game of hockey where each player rushes about trying to hit the puck and, at the end of the game, sits down exhausted having apparently done a lot of work without necessarily achieving very much!
It must be appreciated, however, that the achievement of BS5750 certification is the easy part. The subsequent task of ensuring that the basic procedures are adhered to, improved where possible and modified to suit changing business, is an ongoing process which never ends. The system auditors must ensure that staff use and understand the procedures and must also provide feedback of any failures in the Quality System to management so that prompt remedial action can be taken.
The Importance of Process Control in SMT
Because of its relatively small mass, a SMD has a very small thermal capacity and is, therefore, quite susceptible to thermal shock because only a small amount of heat is required to raise its temperature rapidly. In addition, the passive devices no longer have the protection of bulky encapsulation and are much more prone to mechanical damage than their leaded counterparts. Another important factor is that they are much more susceptible to failure caused by ingress of moisture - passive devices because of the lack of encapsulation and active devices, because of the very small distance between the chip itself and the outer edge of its protective package. Process control, therefore, must be considered in somewhat more detail than is generally necessary for conventional components.
In the days of the 32/6d red-spot transistor it was important not to hold the soldering iron on the lead for too long or the germanium device could be damaged. With the advent of the more robust Silicon devices the need for careful handling decreased. Recently, however, the requirement for careful handling has once more come to the fore. The development of Field Effect Devices and MOS ICs has created components which can be damaged by quite low voltages and, now that sizes have decreased further with Surface Mount Devices, the need for careful mechanical handling and electrostatic protection is most important. Procedures for controlling the handling of components and assemblies are needed and the control must begin with your suppliers since many stockists still fail to understand the importance of protecting devices against electrostatic charges and electrostatic fields. Tests in America have shown that devices may be MORE susceptible to electrostatic damage when assembled on a board than when loose so precautions are necessary at all stages of production.
One of the most important processes is that of soldering which is often regarded as a black art by management and a subject to be avoided. Often the person in charge of the machine will be given only the manufacturer's maintenance sheets and left to get on with the job in his own way. The result is usually initial experimentation to produce acceptable results, followed by extreme reluctance to change anything. This "procedure" might be a good one, were it not for the uncanny ability of the solder machine to go wrong in ways which we would never dream of.
As an example, consider a solder machine in which one of the heating elements fails. The immediate, visible result is a drop in the temperature read by the single sensor device fitted to the machine. The obvious answer is to turn the knob to return the reading to its rightful value. Unfortunately, with one element not working, this action results in a massive temperature gradient across the solder machine. In the extreme, boards can emerge with one half burnt and one half unsoldered! Correct process control in this instance, therefore, requires that the temperature readings on the machine gauges are used only as a guide and that an independent method is used to measure the temperature gradients throughout the length and width of the machine by using a calibrated instrument. In addition, because SMDs have such a small thermal capacity, it will be necessary to carry out the measurements with very small probes in contact with SMD terminations. The choice of probe can be seen to be an important consideration and the measuring instrument must be calibrated to a known accuracy. In general plus or minus 2 degrees C will be a practical and achievable accuracy and will also be quite adequate for the task, since our main interest is in the repeatability of the measurements (from known, good results) and not in the absolute values of temperature. Here, we come back again to the importance of the initial design considerations because the solder machine must be set within the worst case parameters recommended by the manufacturer of the most sensitive component. Clearly, if the designer chooses a component which requires temperatures or durations which the machine can not achieve, then the result could be a lot of scrapped production or, perhaps worse still, a potentially unreliable product. Since devices require widely differing preheat and soldering temperatures a compromise may be unavoidable.
Hand soldering SMDs is feasible but requires careful control. Most manufacturers recommend a maximum bit temperature of 280'C applied to the solder pad - NOT directly to the component - for not more than 5 seconds. Whilst it is almost unavoidable for rework, hand soldering is unlikely to achieve good reliability for mass production.
Hand assembly of SMDs is used extensively in manufacturing and should not be discounted as a viable method for small production runs. It has advantages and disadvantages. Disadvantages include slowness and dependency on operator skill. Advantages include low initial investment and, if the correct tools are used, minimal risk of mechanical damage to the components. Automatic machines give a high risk of mechanical damage unless set up very carefully but are fast, accurate and need minimal supervision.
A cleaning process is usually required to remove traces of flux and soldering contaminants, although some equipment which is to be used in a benign environment might tolerate the presence of residues without detriment. Obviously you do not want the expense of cleaning if it is unnecessary. If cleaning is to be carried out, the designer should consider the fact at an early stage. To clean beneath SMDs requires an absolute minimum gap of about 0.5mm which is often difficult to achieve. It is not, however, a factor which can be ignored and if cleaning is required then it must be 100% or the effort is wasted. The type of cleaning process in use can affect the choice of components. Polystyrene capacitors, for instance, will crack if exposed to certain organic solvents. Ultrasonic cleaning methods, incorrectly used, can damage the internal bonding of some I.C.s. Water wash can increase the risk of corrosion unless care is taken to remove all traces of water immediately afterwards.
Testing of bandoliered components on receipt is made very difficult by the sealed nature of the packaging. Sample inspection is possible but 100% test is impractical to do manually and automatic equipment is prohibitively expensive. Where possible, therefore, 100% test should be performed by the placement machine. Such testing is relatively straightforward for passive components with only two contacts and for diodes but semiconductors with more than two leads pose a problem. You need to get the assurance from the supplier that the devices are 100% operational and perform your own sample test on receipt if possible. This inspection should include a solderability test since this is often a problem. Rework is so difficult on surface mount assemblies that you simply can't leave component quality to chance and no component is more prone to faulty manufacture, nor more difficult to replace after assembly, than the Printed Circuit Board itself. In view of this fact and bearing in mind that the SMD board is smaller and has fewer holes than its counterpart with leaded components, you should be prepared to spend a little more time and money in ensuring that your production line receives no faulty P.C. boards. Insist on the most comprehensive test by the P.C.B. manufacturer (who already has the equipment) and ensure that Inspection on receipt is adequate. Siemens in Germany has created the "Q95" concept whereby 95% of assembled boards pass final test without the need for rework. The concept requires an extremely low defect rate for individual components and for soldered joints and can be achieved only by stringent control of design, procurement and manufacturing processes. The extra work at the beginning, however, pays off by minimizing rework and by allowing assemblies to go straight to final test, with Automatic In-Circuit Test being reserved only for the 5% of assemblies which fail. The result is a cheaper, more reliable product. By the way: if you use a bed-of-nails test fixture, bear in mind that the pressure of the test pins and the vacuum suction might bow or twist the board and cause damage to components (some of which are very brittle).
The packaging is another process which must be considered at the design stage, if only because it is going to cost money! Surface Mount assemblies must be protected from impact and from bowing and twisting forces. Packaging must also protect electronic assemblies from Electrostatic damage without, at the same time, short circuiting any battery which might be fitted! The packaging will often contain operating and installation instructions which might require certain safety information or Approval Marks in order to comply with the law. In addition, the packaging is the first thing that the customer sees and may have to be aesthetically pleasing, as well as containing essential information relating to part numbers and warranty.
Copyright ©1988 Martin Pickering
Version 1.1 updated on 3/5/99
This file may be downloaded for private and personal use but NO part of it may be published in any form without the prior permission of the author.