Controlling Multilayer PCB Registration With X-Ray CMM

How does one optimize the multi-layer fabrication process for best registration of layers and drill patterns? As advances in all aspects of circuit and component technology increase the density of interconnections, more accurate means of layer registration are required. At these densities, the accumulated errors – from plotter calibration and film stability – from artwork and post etch tooling – especially from lamination – and also from drilling – require careful attention to maintain yields at a profitable level.

When assessing the quality of panels in trial runs, the timeliness and accuracy of corrective measures are essential. Leading PCB manufacturers are now turning to automatic measuring systems employing X-ray imaging to quickly determine the corrections needed. Once those corrections are applied, verification is required to maintain quality and catch process drift before it impacts yields.

Once the panel is laminated, the relative movements between inner-layers are fixed; the sooner the amount of mis-registration is determined, the sooner corrective action can be taken.

Historically, multi-layer PCB fabricators have attacked the registration problem from several directions and achieved success in varying degrees. Tooling can only maintain registration to a certain precision. During the lamination press cycle, the layers shift and stretch – often dramatically. Uneven temperature and the dynamics involved when the pressure phase is initiated can cause problematic distortion of the layers.

A range of techniques has been devised to quantify the amount of mis-registration present in a laminated panel: Optical and laser-based probe techniques are of no use, since they cannot penetrate the copper bottom and top layers.

The time required to prepare and measure cross-sectioned samples can be daunting. At the least, the sampled sections must be prepared and measured for both X and Y axes. Proper  preparation of cross-sections is something of an art. When the best skills are not applied, the results can be misleading. When using measuring microscopes to determine the amount of mis-registration, human error further reduces measurement accuracy.  Operator guided measurements always incur some degree of subjective judgment. Short-cut methods that skip the potting and polishing very often give deceptive results. Even those methods that involve angled cutting through coupons to reveal the layers have their problems. The inherently “noisy” imaging characteristics of the cut material make measurements uncertain. The copper is often smeared while being cut, further reducing the accuracy of the measurement.

Several schemes employ X-ray viewing and special artwork geometry as a means of verifying layer and drill pattern registration.  After lamination and drilling, the panels are inspected with an X-ray viewing system.  Usually quick subjective judgments are made as to the acceptability of the board.  With today’s tolerances, this verification method puts a strain on the judgment of the operator.  If the operator doesn’t place the coupon exactly in the center of the X-ray beam, parallax distortion will cause a shift of the layers relative to each other in the image. Any judgments made on such distorted images will be misleading, at best. At worst, parallax distortion can make a good board look bad, or mask the mis-registration in a bad board.

If the X-ray camera’s field of view is large, it makes it easy to find the coupon. But the image resolution suffers.  Even with some measuring capability added to the X-ray viewer, human error and parallax distortion are significant, as above. The image quality and field of view provided by many X-ray viewing machines leaves something to be desired as well.

Some coupon schemes are available which use electrical continuity tests to detect mis-registration. For these, the panels must be drilled and the holes plated. At best, displacement can be determined to within about 0.001″.  Over- or under-etching of the layers, drill wander, and plating process errors accumulate and degrade the accuracy of these tests.

X-ray imaging has also been used to optimize the location of the drill pattern on laminated boards.  Typically, a reference dot is placed at the same nominal X,Y position in all layers, creating a “stack” of the dots in the laminated panel. This is viewed with an X-ray camera and a “center of mass” of the overlaid fiducials is determined. The tooling holes are drilled accordingly, and the boards are then loaded on drilling machines.  This technique improves yields somewhat, but offers little or no measurement data on the amount and direction of displacement.  More importantly, in early x-ray drills there was no means of determining appropriate layer corrections. Later versions offer layer analysis by taking extra time to measure dots on the inner layers. Users have found it impractical to occupy the x-ray drill for such measurements.

The Measuring Solution

The ideal solution would be to measure the panel as soon as possible after lamination as this is the most opportune time. If corrections are needed to the phototools, the sooner they are determined and quantified, the better. The measurements should be made prior to drilling to provide the opportunity to optimize drilling and so salvage panels that would otherwise become scrap. Early measurement also allows really bad panels to be scrapped before adding any other processing cost to the panel. Ideally, the measurement should not require any preparation of a panel or sample, destructive or otherwise.  The measurements should be automatically made and the displacement of each individual layer from nominal should be reported with a resolution on the order of tens of microns.  Accumulated errors from phototooling, etching, and lamination tooling should be included in the measurement if they impact the relative positions and sizes of layers. Process errors that do not impact inner-layer registration should not cause a degradation of the measurement accuracy.  Opportunities for human error must be eliminated.

A system that addresses these concerns has been developed by Operations Technology, Inc. The OPTEK InnerVision System combines high-resolution X-ray imaging with a granite based CNC controlled transport and video measurement.  Image integration and advanced video edge detection are employed to improve measurement reproducibility from the inherently noisy X-ray image.

The OPTEK InnerVision is described as an X-Ray coordinate measuring machine. It measures and reports the positions of internal features such as coupon pads on inner layers of multi-layer printed circuit boards, reinforcements and fastener inserts in composite structures, and other such encapsulated or laminated features. The images and measurements obtained by the InnerVision reveal and quantify offset, skew, stretch, shrink, and other distortions that may affect the location of internal details. The data provided allows fabrication processes such as lamination, molding, drilling, and machining to be controlled.

Destructive sectioning of the panel is not required, because the X-ray imaging system reveals inner-layer copper features.  The top and bottom layers of copper need not be removed, as they do not interfere with the image because they simply become part of the image background. The sensitivity is sufficient to view and measure a dot’s location on an inner-layer against a background from top and bottom layers and insulating laminate materials.

Operator subjectivity is eliminated when video edge detection is used to determine the positions of the Iayers. 100 or more points are instantly acquired at the edges of a coupon dot. For these round features, the system can report the feature’s center position instantaneously. Over-or under-etch conditions on the layers do not affect the accuracy of the measurement, because the dot’s size does not affect its center position. The system provides reporting of each core’s position and scale compared with nominal CAM data.

Experience has shown that the movements that occur between the two sides of a core are negligible. For that reason, it is possible to “double” the dots by placing dots on the two layers at the same coordinates. This improves x-ray contrast by doubling the amount of copper in the dot image, enhancing the ability to measure inner-layers in thicker panels. This method also reduces the total number of dots per coupon, which translates into shorter measurement cycle time.

The preferred measurement method is based on eight coupons of round copper dots of 2 mm (0.080 inch) diameter.  These coupons are usually located in the borders of the panel, adjacent to – but not in – the corners. In this way we can construct distance measurements across the circuit area of the panel to determine the scaling error that has occurred there, rather than measuring the somewhat different movements that typically occur at the edges and corners.

The coordinate locations of the dots for each of the cores are offset from the others so that the dot image representing each core can be measured separately. The coupon area is essentially a “window” of no copper though all layers. Only one dot of copper will be present within the coupon for any given layer. In that particular location, all other inner-layers should have no copper, except of course for the layer on the opposite side of the same core. That layer would have an identically located dot. Top and bottom outer layers can remain as continuous copper foil. This arrangement assures an uncluttered X-ray image.

Once the coupons are integrated into the phototool artwork as described, the laminated panels are placed on the InnerVision machine’s transport. Automatic CNC positioning of the panel for measurement of each dot references measurements to a datum of the user’s choosing. Usually this datum is established using lamination tooling holes, or if drill tooling holes have been placed by an X-ray drill, these are used.

The panels are automatically measured, and using OTI’s Multilayer Registration System (MRS) software, the measurement data is analyzed and simplified reports are generated.

The MRS reports show graphical plots with pass/fail tolerancing of the movements so that an engineer can visualize the errors that are present. The basic MRS Scaling module creates Panel Scaling Reports that show the measured locations of the cores plotted against their nominal positions. A Batch Scaling Summary provides recommended rescaling factors to be applied to compensate the phototools and thereby improve registration in the next run.

The optional MRS Drill module generates additional reports giving recommended offsets to optimally fit the drill pattern to the measured existing panels. The Panel Drill Report plots the core positions shifted according to the offsets. In this way pass/fail tolerancing is applied based on the optimized drill solution, rather than the CAM nominals. The plots represent predicted hole-to-pad concentricity after applying the offsets to the drill pattern.

The laminated panels need not be drilled prior to measurement. However, for high-value panels it makes sense to place the panel on the drill machine and simply drill “marker” holes adjacent to  each coupon. These marker holes can be measured along with the coupon dots. The hole measurements can be used to establish an improved datum based on the drill pattern, rather than using the tooling holes, thereby eliminating any static drill tooling error from the reference datum. Naturally, this technique will improve the accuracy of the recommended Drill Optimization offsets.

In this case, for reporting purposes the datum is adjusted to fit the pre-drilled marker hole pattern and then the computed offsets are applied so that the toleranced plots once again represent hole-to-pad concentricity. Ideally, the panel should then be fully drilled on the same drill station using the offsets given in the MRS Panel Drill Report.

The OPTEK InnerVision’s accurate positioning, high resolution X-ray imaging and video edge detection are all required to obtain sufficient measurement precision. OPTEK’s MRS software provides the analysis and reporting necessary to systematically control multi-layer printed circuit fabrication to the level required to accommodate today’s high layer counts and HDI tolerances.

For more information: