The present invention i elates to a printed circuit board test set with a test pin adapter and a method for setting the latter. This technology is used to connec the standard basic contact grid of an electronic test apparatus for single- 01 multilayet board- type interconnection substrates, such as printed circuit boards, ceramic substrates and similar, to the not necessaril grid- onded contact points of the printed circuit board to be tested (interconnection substrate, test specimen) , wherein said contact points are in the form of very small and very densely arranged metal contact areas ("pads") on one or both surfaces of the test specimen for SMD components and/or contact holes (plated-through holes) for "wired" components. The printed circuit boards are, in the case of the testing under discussion here, generally not yet loaded with such active electric/electronic components. It must be pointed out in this connection that the adapters currently preferred are so-called test pin adapters with test pins inside them, which are used in the printed circuit board test devices currently under discussion both for the single- sided and for the double-sided testing of printed circuit boards. However, other "contact elements" are also conceivable as test pins, such as so-called "vertical conductor rubber plates ", i.e. lor example rubber mats with contact wires finely distributed therein and disposed at right angles to the surface, or similar elastic mats which become live due to local pressure loads in this area, such are known in prior ar .

As regards the testing of unloaded printed circuit boards, the limit of the testability was determined in the past mainly by the configuration possibility of the contact test pins in the adapter and by the contact density of the bonding pad (basic

grid) of the printed circuit board test device. Tn keeping with the ever increasing miniaturization of electronic components, there has however occurred every year a further reduction in size of the electric conductor tracks and contact points and their density on the printed circuit boards to be tested. At the same time the size of the panels has had to be increased for utility-based production to the relatively large standard formats, i.e. the standard formats contain a number of identical printed circuit boards or printed circuits, in order thereby to optimize the economic return on the manufacture of the latter.

There therefore arises on an ever increasing scale the problem that the test pins in the supporting element of the adapter cannot be positioned sufficiently accurately in relation to the connecting structures (connecting areas, pads, holes) on the printed circuit board. This gives rise in many cases to defective and/or wrong contact points. These defective contactings make a reliable electrical test of these printed circuit boards extremely difficult or even impossible.

The causes of this problem lie (albeit on a reduced scale) in the adapter and above all in the test specimen itself. The influence of the adapter on the deviation of the contact points from the target value results both from the tolerances during the manufacture of the various components of the adapter and from the requisite clearance between e.g. the test pins and the guides provided for the latter (e.g. a number of drilled plexiglass plates as the supporting element) of the adapter:

Accuracies of typical test adapters (excluding influence of the temperature during the testing)

Deviation of the position of the tip of the contact element relative to the the centre of the guide element ± 20 μm/± 0.8 mil Overall clearance between guide element and guide hole in the insert + 25 μm/± 1.0 mil

Positional tolerance of the guide holes in the insert ± 05 μm/± 0.2 mil

Resulting in an overall deviation of ± 50 μm /+ 2.0 mil

"mil" means here a thousandth of an inch - this unit of length has acquired general acceptance in the printed circuit board industry.

The influence of the temperature during the manufacture and use of the adapter can be limited relatively easily by air conditioning measures. These measures can therefore be excluded from further conside ation.

Far more difficult to control are the differences in the geometric structures on the test specimen, i.e. on the printed circuit board to be tested. These differences can be divided into relative positional displacements of the whole image, which can be attributed to recording errors during the exposure process, and to distortions within the image, which have arisen through irreversible t liermal processes of the test specimen or of the film used for the exposure of the printed circuit. By virtue of the sequence of the production processes (drilling, exposure, hot tinning etc.) there are at least four mutually independent sources of error affecting the exact

position of the conductor tracks and points of connection on the printed circuit board, i.e. the holes in the printed circuit board, the surface structure of the underside, the surface structure of the topside and the outer contour must be regarded as four mutually independent overall structures. Each of these overall st uctures possesses its own peculiar distortion/deformation properties affecting the overall geometry of the useful object. The orientation of the overall structures to one another can be described by a displacement/distortion in X- and Y-direction and a rotation. For the individual printed circuit board in the panel the distortions can moreover be represented witli sufficient accuracy as pure displacements and rotations.

Typically a "mixed technology" is used today. In the latter both wired components for connection to contact holes and surface- ountable (SMI.) components for connection to contact areas or pads on a printed circuit board are provided. The grounds of the need for this mixed technology are that a number of electronic components, such as e.g. processors in high-pole PGA's (pin grid arrays) are not obtainable in surface-mounted form. Economic considerations or other technical factors also lead to this mixed technology.

The printed ciicuit boards manufactured for pure surface mounting can be al igned by means of conical reference pins for which the optimum position has been determined manually or optically. This kind of alignment, however, does not work for a large proportion of the printed circuit boards used today with the above-mentioned mixed technology, because the large number of conical test p ns/contact, elements provided for the contacting of the contact holes, when they are pressed against the test specimen, unintentionally displace the test specimen again into a false position, and hence cancel out the adjusting effect of e.g. two reference pins.

The attached Fig. L of the drawing shows the theoretically intended arrangement of contact elements or test pins during the double-sided testing of a printed circuit board B which is tested by means of a conventional adapting technology. The printed circuit board is here held in position by reference pins T and in the area of plated-through contact holes is contacted by relatively massive pins H and in the area of very small, cLosely arranged contact areas or pads on the board topside by thin, very closely arranged pins S.

Fig. 2 shows the defective alignment, resulting in fact from unavoidable fabrication inaccuracies during the manufacture of the refeience pins, the test pins and the printed circuit board B to be tested, of a more or Less large number of pins S in proportion to the very small contact areas for SMD components assigned to them, whi ch tesults in correspondingly defective contactings of the test pins S if the conventional adapting technology is used. Since the holes in the printed circuit board B (for forming the at least two reference holes for the reference pins T and the in most cases very numerous through-platings for the test pins 11) are produced in a different operating cycle to the "printed circuit" including the contact areas for the SMD components, it is impossible in practice, despite adequate miniaturization of the connecting structures on the printed circuit board B to be tested, to prevent the holes and the printed circuit together with the contact areas from undergoing a major unintentional displacement relative to one another, leading to the defective contacting, shown in Fig. 2, of one or more contact areas for SMD components by the pins S, which results in corresponding measurement errors in the testing of the printed circuit board B if the conventional adapting technology is used. The defective alignment described above is however not caused simply by the possible defective al ignment of the holes relative to the printed circuit, but to a large extent also by the fact that, by virtue of the production process for manufacturing the printed circuit, the latter deviates in the

final analysis from its theoretically determined or intended position and dimension, i.e. it is applied to the board B inherently deformed or distorted or askew. Since however the guide holes for the pins T, 11 and S in the board TF or BF (TF = "top fixture" - top adapter; BF = "bottom fixture" - bottom adapter; S = "SMD sample - test pin for SMD pads; H = hole sample - test pin for contact holes) cannot be produced to take account of such accidentally occurring deviations and/or are themselves subject to certain fluctuations/tolerances in their position, it becomes clear that in the case of this conventional adapting technology with increasing miniaturization of the printed circuit boards to be tested, and with the increasing density in particular of the contact areas or pads for the SMD components, it is becoming increasingly difficult or even impossible to achieve a sufficiently accurate contacting in all contact points s multaneously of the printed circuit boards to be tested.

A further cause for the defective contacting of the printed circuit boards to be tested, whicli is frequently observed during the conventional adapting despite the (supposed) fixing of the latter to a predetermined position by means of the reference pins T, is based on the fact that during the pressing of the printed circuit board in the course of the test operation onto the possibly very numerous test pins II the latter in certain circumstances simply cancel out the intrinsically provided and intended alignment effect of the reference pins. For since the exact position of the individual test pins H provided with a cone point, which engage with the holes of the through-platings, is naturally also subject to certain fluctuations, and namely not on Ly in relation to the respective position of the holes accommodating the pins H, but also because the cone points of these pins H are by no means always accurately disposed centrally, it can easily happen that the actual position of the printed circuit board B, when the printed circuit board is in the pressed state for the test operation, is determined very much more by the large mass of

the test pins H than by the in most cases two reference pins T and hence is also somewhat accidental .

A solution to this problem will be explained with reference to Fig. 3, which shows a possible form of execution of the invention by means of a cutout from a printed circuit board test device with a novel test pin adapter on the top side and underside of a test specimen B. This solution utilizes a plurality of sections of the supporting elements, which sections, e.g. subdivided into various planes and disposed on one particular side of t lie test specimen B, are e.g. in the form of adapter plates which are freely positionable independently of one another relative to the test specimen within limited regions relative to the test specimen. Fig. 3 thus shows the essence of the novel adapting technology, which is distinguished above all by the fact that the test pins S for the contacting of the very smaLl and densely arranged surface contacts of SMD components on the printed circuit board are accommodated in holes in a section of the guide element (hence e.g. in an adapter board) BF or TF, which is (are) displaceable by a significant dimension parallel with the printed circuit board plane in relation to tire t.est pins II and the reference pins T and hence to the test specimen, wherein this dimension must conform at least to the greatest anticipated defective alignment of the test pins S by the surface contacts of the printed circuit board to be tested which are assigned to them.

For the spacing of the various sections of the supporting element (e.g. of the guide plates of a test pin adapter) a sequence presents itself which takes account both of the accuracy requirements concerning the test pins guided by the latter and of the mechanical stability properties of the latter: the more slender and hence more flexible test pins for connection to the contact areas or pads are accommodated in the section of the supporting element (guide plate of the test pin adapter) which lies closest to the respective printed

circuit board, since the positioning of these contact elements/test pins K requires, by virtue of the high contact density of these small contact areas and the high flexibility of the thin test pins, the greatest possible accuracy.

For the contact elements intended to follow the loading with wired components, holes/vacancies are used whose dimensions are increased to such an extent that a lateral movement of these contact elements within the required tolerances is possible. The great stability of these typical ly more stable contact elements, combined with the sel f-centring of the conical tips in the holes of the test specimen, permits guiding without difficulty through an independent, further removed section of the supporting element. The contact elements for the "pads", which of necessity penetrate this second section of the supporting element on the respective printed circuit board side, are guided through enlarged holes/vacancies in a similar manner, so that an independent precision positioning is possible.

For the positioning of t.he test specimen in the adapter, reference holes are generally used which are produced simultaneously with the standard di illing process for through- platings or component holes. For this case it is essential that the alignment means (e.g. reference pins) required for the pre-centring during the manual or automatic insertion of the printed circuit board to be tested be positioned on the same support plate as is used for the contacting of the terminals for wired components. This is possible if all the holes in the printed circuit board, including the reference holes, possess a common distortion and displacement structure.

In special cases the reference holes are produced together with the milling of the outer contour as a last production step. In this case it may be necessary to provide a further support element section for the positioning, which can be

positioned independently of the other sections relative to the test specimen.

In each case it is highLy recommendable that a steadily pressed-in barrel be used for the guiding of the "reference pins". Such a barrel, typically reaching close to the test specimen, prevents lateral spreading/warping of this alignment means (reference pin) despite the necessary clearance for the vertical movement , normally supported by a spring, of the reference pin.

Although the reference holes produced together with the outer contour on accuracy grounds in special cases in the past theoretical ly exhibit better accuracy by virtue of the absence of a plat ed-through hole barrel, the use of piated-through reference holes is nevertheless to be recommended, because the overall work-load for the testing is reduced and the collection of the many contact elements with typically conical tip for t lie contacting of the terminals for wired components permits a substantially better centring. The law of large numbers operates here, which produces for the overall positioning accuracy the mean value of the distribution frequency, contingent on manufacturing tolerances, around the target value of the positioning. In mathematical terms there is obtained as the probabLe deviation the statistical error of the mean value. Consequently the di fe ences produced by slightly varying wall thicknesses of the pi ated-through barrels are more than offset.

Two methods can be used in principle for determining the position for the alignment of the support plates by alignment or contact elements:

The simplest (first) method is a common positioning for a lot. This method can only be used effectively, however, if the fluctuations in the distortions/displacements occurring within a lot are small compared with the total deviation to be

corrected. Elaborate measuring and subsequent positioning is inappropriate in this case. For tire test specimens lying at the extreme edge of the contact points -position errors- distribution curve there will then be added as a follow-up to the test of the lot an additional labour- and time-intensive phase, in which the p inted circuit boards exhibiting faults (typically breaks) are tested with manually optimized displacement values.

In the second method an alignment takes place for each individual printed circuit board. To enable optimum displacement values to be set directly for each test specimen in the first test, optical measuring devices, such as e.g. CCD cameras, are generally used. These cameras measure the position of marks which are provided additionally for the optical recording. These marks are present on almost all test specimens, so that the fully automatic loading devices used later can load the printed circuit board later with surface- mounted components. This means therefore that a special measuring apparatus or station is required, which prior to the printed circuit boat d test, proper measures the displacement of the printed circuit of each board by means of graphic elements in the "image" of the printed circuit board.

Since in the double-sided testing of printed circuit boards - as mentioned above - very many position error degrees of freedom exist, a corresponding number of cameras also have to be used for this second method. There is obtained as a minimum for the latter a number of two cameras on each printed circuit board side to be tested, in order to determine for each side the actual orientation in the three degrees of freedom (X- and Y-axis, angle) . In practice these cameras cannot be accommodated in the test adapter, even with extreme miniaturization, since the optical recording marks lie very close to or between the contact points to be tested. A remedy is provided here by the above-mentioned special measuring apparatus or station, in which the test specimens are measured

optically prior to the test proper. This measurement can be cairied out in the simplest manner in a fully automatic test system, because then the test specimens can pass through a synchronously operating twin station, in which in the first station measurements are made and in the second station (the test adapter) testing is carried out with optimum positioning with the displacement values thereby determined.

In the measuring station there can be used per side either one camera positionable freely in X/Y direction or two cameras recorded through exact positioning holes in the adapter. The fiist case increases the mechanical outlay on the system side (one-off procurement costs) , because a high-precision positioning device is J equ i r ed . The second case requires greater expenditure on each adapter, which has to be developed individually for each individual p inted circuit board (series) .

In a fully automatic test system, in which the conveyance to the actual test adapter takes place by means of a control system which is able to monitor highly accurately the location of the test specimen during the conveyance, alternative methods exist for measuring the displacement of the printed circuit on the printed circuit board relative to the holes. For example, the position of alignment marks can be measured in such a way that a reflexion light barrier measures the reflexivity of the surface of the test specimen either directly or indirectly across a certain distance by means of a fibre-optic light guide. The degree of intensity measured during the movement permits an accurate computer-aided calculation of the displacement values. The degree of accuracy required of the conveyance means can also be reduced if a straightforward incremental measurement between the printed circuit board print and the holes is used. For this there can be provided e.g. a hole in a preferably rectangular metal area (pad) , during the optical scanning of which an intensity profile is then obtained whose deviation from a predetermined

profile or whose asymmetry mirrors very accurately the displacement values.

In order to become as independent as possible of the contamination problems encountered in everyday production runs, use can also be made, instead of optical measurement, of mechanical scanning (pick-up) . For example, a sapphire pin or a subminiature sapphire roll with connected structure-borne noise amplifier scans the surface of the printed circuit board. The edges of a metal area on the test specimen relative to the support material or relative to a hole then give clearly di fferentiatable (sonar) pulse shapes whose symmetrical or asymmetrical position provides clear evidence of displacement (sonar pattern) .

The measuring techniques described above provide the possibility, prior to the test run proper, during which the printed circuit on the test specimen is verified by the contacting of possibly thousands of test pins, of carrying out a measurement of the displacement. The result of the measurement is then used directly for the adjustment of alignment elements in the adapter, so that the adapting device is optimally positioned individually for each individual test specimen. This advantageous result is made possible by the work undertaken during the measurement of the displacement values .

Should it be decided on economic grounds or because of reduced accuracy requirements not to use one of the above-mentioned methods, there is a further method which requires a small work-load, but can only be used "post factum" :

Use is made there of the fact that the displacement values found within a single lot vary far less than those between lots (tracking production parameters) . Within the adapter that is used for the electrical testing there are accommodated, in addition to the test pins or contact elements that are

required for the actual testing of the connection pattern or the printed circuit, further displacement test pins, which interact with a special "test pattern" or alignment marks on the printed circuit boaid. These additional, displacement test pins are so arranged that they supply, in the event of increasing displacement of the board pattern, an increasingly varying displacement measurement result. Should one or more faults be discovered during the electrical testing, therefore, there can be established by means of the separately recorded, additional displacement measurement result the direction, and in some cases a value, for the displacement correction required. After a corresponding mechanical adjustment of the alignment means in the adapter the same printed circuit board is then tested once again. The additional displacement test pins introduced for the displacement measurement can moreover be connected either to the conventional printed circuit board test set itself or to an external electronic unit and be evaluated by the latter. In special cases there will in addition be determined by means of the relationship with the mechanical displacement values the optimized position of the adapter relative to the test specimen. A structural combination of this external electronic evaluation unit with the mechanical control unit for adjusting the alignment means on the adapter makes a compact solution or assembly with user - friendly operation possible. This leads to a twin station with a Station A "Sensor" with the optical or electrical displacement- measurement and a Station B "Tester" with the mechanical setting of the adapter based on the measurement result. The electric displacement sensor operates by means of alignment marks, including for pure SMD application, i.e. without "mixed" technology with subset displacement.

The adapter according to the invention can also be constructed for a "mixed" manufacturing technology by division (partitioning) of the guide element in the plane of the guide element, if the structures on the printed circuit board exhibit adequate geometric spacings .