Engineers working together to get it right first time
12 February 2009
Whilst design engineering is ultimately about producing a design that works, production engineering is primarily about managing risk, state Kelvin Clayton and Nigel Palmer.
Design engineers and production engineers have a very different outlook on life, although both disciplines are in the same branch of engineering. Whilst design engineering is ultimately about producing a design that works, production engineering is primarily about managing risk. Each one of the design rules that production departments produce is aimed at reducing that risk to improve production yield. In aggregate, they make a huge difference to the production cost by reducing the number of faulty boards leading to scrap, rework or failures in the field. Although each and any of the rules can be bent or broken to an extent, there is always a price to be paid in terms of manufacturing cost and/or production yield.
Many designs end up with areas where the required size, performance, cost, or power consumption can only be achieved by operating at the limits of the manufacturability guidelines. In partnership with the manufacturing team, it is often possible to identify ways of easing the constraints elsewhere to minimise the impact on yield and maximise reliability.
The vexed area of test access is a case in point. In-Circuit Test (ICT) still provides the fastest and cheapest way to verify that a board has been assembled correctly, and provide a clear diagnosis of any issue to enable rework. Its use requires that an area of the board is dedicated to providing access for the fixture pins. Test engineers everywhere repeat the mantra ‘build test access in at the start of layout not as an afterthought’ twice daily. Despite this, test access continues to be sacrificed on the high altar of design constraints and cost.
The spacing between adjacent test points should ideally be 2.54mm on the solder side of the PCB, although they can be staggered to achieve this pitch, saving a little valuable space. This enables the use of 100mil probes that are stronger and more reliable. Smaller probes can be used down to 50mil. Test pads can be given square or diamond shapes, so that they stand out clearly should any board redesign be carried out, though this does make it harder to place them.
There are, of course, alternatives to ICT, but each has potential issues. Flying probers come closest to matching the performance of fixture-based ICT – and indeed for low-volume designs are a more cost-effective solution as they don’t require the initial investment in a costly fixture. With their angled probes, flying probers are also able to reach into awkward corners of the board but with large and complex boards they are slow. AOI is an effective tool requiring no test access at all, but is unable to fully verify a sound electrical connection or to check sub-surface areas of the board. Functional test is an excellent final verification of a working product, but provides little or no indication of the nature of the fault if there is one. For ‘cheap’ goods where a level of scrap is acceptable, this isn’t an issue, and often a flying prober can be used at the rework stage to identify the cause of the failure. X-ray is more a production profiling and debug tool than a true in-line strategy, although it has yielded interesting results in identifying faulty components.
Experienced test or DFM engineers can often address gaps in the test coverage. For example, if a specific critical resistor R cannot be tested, there may be another pull-up resistor elsewhere on the board that can be replaced with a component of the same value as R, allowing the tester to verify that at least the correct reel of components was used in manufacturing the board. AOI can further enhance test coverage by verifying that the critical resistor is there and has been soldered correctly. In extreme cases, we have soldered a ‘dummy’ resistor to a breakout area of the board, although the cost of manufacturing will increase by additional component placement.
Tooling holes
The location and size of tooling holes are critical to holding the board accurately during manufacture and test, but forward planning can allow the same hole to be used for other purposes, such as attaching the board to the case.
The specification for these holes needs to be fairly precise. They must have a minimum diameter of 2mm (preferably 3mm), they must be drilled at the same time as any component holes, and they must be as far apart as possible, with a tolerance between the holes of +/-0.05mm and at least 5mm from the board’s corners. Despite this, designers are still offered huge latitude on the size and location of the holes. The biggest constraint is probably that the holes cannot be through plated as they could become unsuitable to ground the board to the case. Again, this can be overcome by via-stitching around an unplated hole, then creating a ground connection through the mounting bolt.
Staying single
Although double sided placement is no longer an issue to manufacture or test, there is still a cost penalty in process costs and fixturing costs. It is therefore worth attempting to get the whole design on one side. Starpoint (now part of Suzo-Happ) has kindly given us permission to discuss a project we worked on with them where we were able to achieve just that. The product, a module for a vending machine was aimed at a price sensitive market, and the initial design was also slightly larger than the mechanical module it replaced. This issue was resolved by a focussed partnership effort between Starpoint’s design team and TT electronics test and DFM engineers, to create a design that achieved all of Starpoint’s objectives and was cheaper to build in volume. The original design was double sided and used 0805 and 1206 chip passives. A useful cost saving was achieved by fitting the design onto one side of the board. The major steps we took were recommending the use of smaller chip passives and combining components of the same value into space-saving network devices. In discussion with the design team, it also emerged that a part of the board’s functionality was an interrogation unit to support field maintenance of the system. By transferring this part of the circuit onto a dongle to be carried by the service engineers a further size reduction was achieved. The resulting design gave Starpoint a much stronger competitive position in their market and a 20% cost saving.
There are a number of other useful tricks in PCB layout, each of which can reduce the manufacturing risk. For example, if chip passives are placed in line with the board flow, the risk of ‘shadowing’ (insufficient solder placement at the rear side of the component) can be reduced. Multi-pin components should be oriented narrow-edge to wave. Where polarised components such as leaded capacitors need to be inserted by hand, there is a measurable improvement in yield when they are all the same way around. Components close to the edge of the board may be stressed during break-out, creating hair-line fractures in the devices which are hard to detect and may only become apparent when an early failure in the field occurs. The recommended distance is 2.5mm, rising to 5.0mm for ceramic capacitors. If the spacing for ceramic capacitors can not be achieved, it is still possible to remove the stress by routing a slot along the v-score line by the capacitor’s location at PCB manufacture stage.
To put the contrast at its starkest, a designer has to get the design to work once, a production engineer has to get it to work every one of the thousands or millions of times it is made for, over a period of perhaps years, at potentially multiple manufacturing sites. Fortunately, none of the design engineers we work with define their jobs as narrowly as that, which is just as well, because a successful production design is as much about the relationship between individuals as it is about engineering. If a design is pushing the manufacturability envelope in one area, discussion can often identify ways of mitigating the risk by easing constraints elsewhere.
Contact Details and Archive...
Related Articles...
Most Viewed Articles...