Analyze it! Designers and fabs look for different things, so both bear the burden of validation. (CAM Works).

MY POSITION IN EDA affords the unique privilege of sitting on the fence, so to speak, between design and fabrication, and I spend a fair amount of time with both groups. I have noticed an interesting disconnect in beliefs when it comes to what must be done in order to validate a design (or set of Gerber files), and who is responsible for doing it.

I say disconnect because, from fabricators come complaints that designers don't do enough to validate their layouts for manufacturability prior to submitting the job for fabrication. And I hear designers insist all they need to do is "view" the Gerbers to ensure they look OK before sending them for fabrication; the shop can take care of the rest. When each side points a finger in the other's direction, what's the right way to approach this disconnect? Both sides should perform equal levels of analysis in order to validate a given design, each from their own perspective.

Fabricators don't check a design for "design errors" in the sense that a designer would. How could they? They didn't do the original layout. Instead, fabricators are typically more concerned about minimums, and whether a board can be built within a given set of shop tolerances. These minimums help the shop to determine the appropriate methodology for a given job, which plating line it should run on, what areas of the design should be watched for potential problems, how the job is to be quoted, and what turnaround times can be expected. A fabricator will check every design that comes in, regardless of whether the designer deems it perfect, because it has to make sure the boards can be built. But this doesn't preclude a designer from performing a manufacturability validation of their own, as they have a different set of problems to look for.

Unlike the fabricator's checks, a designer is typically concerned with issues that might affect the overall connectivity and working aspects of a layout--whether a matched length differential pair is no longer following its assigned 0.005" trace-to-trace spacing difference, or whether two pads are far enough apart to slip just one more trace between them and still meet the design rule requirements. But what about the manufacturability aspects, things like 90[degrees] routes from via pads that will cause etching problems, or the soldermask clearance that is now too tight because of that extra trace slipped in between those pads? These are simple examples of etching and mask problems that aren't found by "viewing" manufacturing data in a shareware viewer (or even one that was paid for). The human eye cannot reliably detect a 0.004" acid trap by scanning visually over via pad locations, let alone a 0.002" soldermask clearance between a pad and its neighboring trace. Yet these problems exist on the majority of boards built today! By not running more in-depth manufacturability analysis on layouts, designers leave themselves open to the potential for scrap, slower turnaround times and generally poor yields.

I have known a number of designers, at sites such as Motorola, Ericsson and Avaya, who have dramatically improved fabrication throughput, turnaround and yields by conducting manufacturability analysis prior to sending designs to fabrication. Some of these same people have also seen improvements in their relationship with a given fabricator, with better communication and less conflict when problems are identified.

What types of problems are these designers targeting? For the most part, etching and mask problems that are not found by most CAD systems. For etching this would include acid traps (or photoresist slivers), copper slivers, pin holes, and starved or isolated thermals. For masks some of the more common issues are soldermask slivers, soldermask-to-trace spacing and solder bridging. These represent the entry-level issues that can potentially be checked; however, starting with these will improve board quality during design and permit the fabricator to focus its time and energy on building the boards, versus the go-arounds with designers on corrections and new datasets.

Here's my message to designers: Don't let manufacturability problems plague you; take control of your designs, deploy the CAM or DfM tool and validate your manufacturing data before they go to fabrication. Don't wait for the phone to ring or boards to come back to learn the design had problems. It is true that the fabricator will probably find all of a given design's issues and more. But is it realistic to assume that the fabricator will also fix those problems when it really doesn't know anything about the purpose of that particular design? Better yet, should the fabricator be held responsible for fixing them at all?

Looking ahead, my next column will help you better understand the etching problems discussed here. We will look at what acid traps, pin holes, copper slivers and thermal problems really are, and what you can do to fix them.

JEFF MILLER is vice president, sales and marketing at Wise Software Solutions (wssi.com), He can be reached at [e-mail]. His column appears quarterly.