Between savings and unknown expenses Peter Reinhardt, CEO Reinhardt Testsysteme

It is an undisputed fact that board assemblies will and cannot entirely be produced without defects. The failure rate depends on technology, complexity and experience of the designer and lies in a range of 3 to 40%. Over time, almost all board assemblers have thought about building their own test equipment. This one sees in the advertisements of the market where a number of firms offer bits and pieces as well as complete subsystems with software for designing such board test systems (board ATE).

As a result, highly creative engineers have gotten the idea of developing such test systems, but their knowledge of board testing involves their present technological situation at its best. This means that the result of this design procedure will probably be a test system which covers the current requirements of the actual products manufactured. These kinds of ATE systems will quickly become obsolete in light of future developments, because they can only be used for a relatively small amount of boards currently produced. With the inevitable advent of the next-generation component technology and in turn the design of new prod ucts, those board testers will render themselves almost useless relatively fast for the novel products, since they can only be used in a highly limited manner within a short time.

At the beginning of board test some 40 years ago, a very simple functional test was provided through a fixture which connected to the board-under-test by its normally available edge connectors. The fixture and its circuitry contained a number of switches and checking routines to connect the board under test to a number of different stimuli and measurement instruments. The control was done manually, and a test procedure written by the product designer was the guiding part for the procedure. Results were normally not documented, and the performance of the operator was depending on his changing fitness and awareness from one day to the next. Later, computer-controlled test boxes followed which already used commercially available board-level instruments for stimuli and measurement and switching. But programming this ATE solution was extremely time-consuming, and very often required many weeks and months to bring a result of 60 to 75 % of coverage performance (the coverage of all pos- sible flaws).

Our middle-sized company has been active in the field of automatic testing for more than 28 years now and we have sold more than 1,700 board ATE systems. Over the years, we have talked to more than 4,000 manufacturers of electronic equipment. We have accumulated enormous knowledge about board test and developed solutions which are easy to fixture as well as easy to program. In Germany we have been the market leader (in terms of sold units) for computer-controlled test equipment for in-circuit and functional board testing for eight years, and more than half of Germany’s assemblers of electronic boards use our test gear for production. The software of the present generation of our board test systems is based on 25 men years of design efforts, and with its flexibility it can virtually deal with all issues of industrial board test.

Functional test is the most used procedure for ensuring that the customer of a product gets the best performance, but it is true that functional test, whatever ATE system may be used, allows only a fault coverage to a maximum of up to 90%. In fact, we have found that most of the programs lie between a defect coverage of 60 and 80%. This small number is extremely dangerous in its results, since with it nobody can assure the performance of a product at customer site. The shortcoming of the functional test is that the single test steps can only be developed and implemented with the detailed knowledge of function and defect effects of the designated circuitry. A problem which ranks even higher is that a functional test will in many cases appear as “successful”, although there are missing components or wrongly assembled components. And the board will fail either in the long term or due to temperature changes, depending on the functional states and/or climatic conditions. If, for example, pull-up and pull-down resistors are assembled in a wrong value, the board will fail in bus systems where it is integrated later, because the necessary load or termination on a bus line, which is supplied by the surrounding electronics, will temporarily generate failures or instability.

The in-circuit test allows ensuring that all connecting I/Os and board tracks make safe contact, and that shorts or bridges are reliably detected. Component test for values and assembly direction of the parts are also possible. Only a handful of components, e.g. varistors, are not testable with this methodology. The combination of a reliable in-circuit test and a limited functional test delivers results which will be even higher, for example, in the range of 90 to up to 95%. This test approach is not available with commercially available subsystem modules based on popular bus systems used to built own board testers. Therefore, with the design of an ATE system applying those modules, a user will miss the benefits of the in-circuit test.

The in-circuit test (also called manufacturing-defects analyzer, MDA) allows an exact fault-location on component level relatively easily. This shortens the time for the complete cycle of test and repair of the board enormously. A graphical display of the circuit and the layout of the board-under-test provides the needed information for the operator. So he can easily see which tracks are shorted or bridged, and which component is missing, wrong in value or errantly rotated or otherwise out of orientation. The functional test, which is in the majority provided on the own-designed test systems, does not support any faultfinding or fault location and does not offer any graphical display of the location on the board assembly. These are serious drawbacks.

Most of the self-designed test systems require software which has to be provided by the engineer who develops the board product. In most cases, this software is not documented in its entirety. Therefore, this engineer is the only individual in a company who is able to write or modify programs. It is a fact that ongoing modifications of a test program are inevitable, in order to immediately improve the results and to follow design changes of a board. If such software is not very well or even not documented at all, the initial product designer can only do such a modification. This engineer who developed the hardware and software of a test system has hardly the time to perform such support, or he is so much occupied with this task that he is not available for the job he is supposed to do: the engineering of new products.

And if repair of such a test system is necessary, he is the only person who can do this based on his experience with this system. This engineer has to contact the supplier of the stimuli and measurement modules used in his configuration. From experience we know that these modules have relatively short lifetime cycles, which means that when he returns the subsystem, the supplier (mostly located abroad), already produces the second or third generation of the units. The vendor will therefore indicate that it will be easier and more cost-efficient to switch over to the new generation which of course features higher accuracy and performance. But in most cases, the old software cannot entirely be used. Then this ATE system has to be redesigned and the software has to be re-written partly or in total which takes another few weeks and months (costing a fortune), before a user can continue to test. The same problem applies to stimuli and measurement instrumentation with the IEEE/IEC or GPIB-bus. Very often, suppliers will tell their customers that repair will take 6 to 8 weeks, and if a user immediately needs the stimuli and measurement instruments he should buy a second and a third unit to overcome the shortage. If the design engineer of the own-built, in-house test equipment is promoted, becomes sick, leaves the company, etc., a company may be forced to dump such a test solution as there will be then nobody in the firm who can provide any support or programming. From this scenario it becomes clear that those solutions may be a short-term solution for an actual task, but can never provide a long-term answer to handle future requirements and changes.

Professional test system suppliers who have a lot of experience in delivery of 1000+ systems can give full support and training, even for those models which are 27 years old and still in use. Spare parts and service at the customer’s location is available, too. The system software to monitor, control and program the test equipment is easy to operate. For example, in-circuit tests are typically generated within 3 to 5 hours, and functional tests within 1 to 2 days of programming and debugging. New hardware and software modules are developed continuously to cover the needs of ongoing innovation.

But for a complete solution, the test system alone is not enough. It is necessary to have a test fixture/adapter which allows both functional test and in-circuit test in the same frame with the same mechanism. This kind of fixture must be reliably connected to the system. We offer our customers an individual solution for fixturing the board-under-test at extremely low cost. Our efficient, individual fixturing solutions seldom reach the range of more than 500 euro. A change of programs from in-circuit to functional or vice versa is typically done in one minute including fixture change and loading the program. With self-designed equipment this is much more complicated, more time-consuming and much more expensive.

Most manufacturers are now ISO9000 approved. This quality-management requires a very clear definition of what kind of test gear is used, when was it calibrated and: does it still meet the calibration specification? Most home-built solutions cannot be calibrated according to this kind of requirement. Therefore, a test system of this kind cannot be used in an environment which is certified according to ISO9000.

The decision to build one’s own test system is very often based on the fact that some of the engineers want to ensure their working relationship, because as long as such a system is used, they can hardly be layed-off. Thus, regarding all the facts given above, the result will probably be that the approach of designing one’s own production test equipment is in no way a profitable or practical solution.

EPP EUROPE 441

The in-circuit-test programming of the Reinhardt ATS-KMFT 670 typically requires 3 hours. A new fixture costs about 250 euro. Programming time for the functional test, depending on the complexity of the board, will very seldom exceed 2 days with total cost of 800 euro. It is no secret that the sooner a product reaches the market, the more profitable it is for the manufacturer. So an extremely short time-to-market will pay off very quickly, whereas an in-house built solution will require weeks and months before the product reaches the market. Our test system ATS-KMFT 670 and all earlier models have been supported by training, spare parts and on-line-service for a very long row of years. With the existing calibration module consisting of hardware and software, calibration of our test system can be performed in a few minutes on site. Our team of six specialists will assist users in setting up test solutions even for the latest technologies based on many years of experience. Cost is the most important part. At a first glance, a designer of home-built systems will probably believe that he can provide an answer at a fraction of the costs of a commercially available system. But the company has to realize what all the many expenses have required for software development, programming, troubleshooting and time needed to understand the instrumentation modules to be properly integrated. The final expenses can easily be 5 to 10-times the cost of a commercially available test system. In the following years of operation, these expenses can even double or triple.

Au cours des années, le prix des testeurs de circuits imprimés, disponibles pour la première fois à la fin des années 70 en version low cost, est devenu de plus en plus attractif. Néanmoins, quelques fabricants d’ensembles sont intéressés à l’intégration de leurs propres solutions sur la base de sous-systèmes matériels et logiciels. Une fois le matériel assemblé, cela ne représente toutefois qu’une demi-solution pour les testeurs de production car le système doit être continuellement adapté aux exigences et le logiciel du système entretenu. L’expérience montre qu’après une certaine durée de fonctionnement, les systèmes disponibles dans le commerce peuvent engendrer des coûts bien plus importants.

Nel corso degli anni i Boardtester, presentati per la prima volta negli anni settanta come versioni Low-Cost, sono diventati sempre più economici. Nonostante ciò alcuni produttori di gruppi costruttivi sono interessati ad integrare da sé soluzioni sulla base hardware e software di subsistemi. Tuttavia nei tester di produzione l’hardware assemblato rappresenta la “metà dell’affitto”, il sistema deve essere in continuazione adattato alle esigenze attuali e anche il software del sistema richiede una notevole manutenzione. Le esperienze hanno dimostrato, che attraverso un determinato periodo può risultare un multiplo dei costi per i sistemi disponibili a livello commerciale.

Im Lauf der Jahre wurden Boardtester, die es erstmals Ende der 70er Jahre als Low-Cost-Versionen gab, fortlaufend kostengünstiger. Dennoch sind einige Baugruppenhersteller daran interessiert, selbst Lösungen auf Basis von Hard- und Software-Subsystemen zu integrieren. Allerdings ist bei Produktionstestern die einmal zusammengestellte Hardware nur die „halbe Miete“, das System muss laufend den Anforderungen angepasst werden und auch die Systemsoftware ist zu pflegen. Erfahrungen zeigen, dass über eine bestimmte Laufzeit durchaus ein Mehrfaches an Kosten für die kommerziell verfügbaren Systeme entstehen kann.