IPC International Community magazine an association member publication
Issue link: https://iconnect007.uberflip.com/i/1547617
NOVEMBER 2026 I I-CONNECT007 MAGAZINE 41 foresight to include a mechanical alignment mark so the pattern could at least be positioned perfectly along the PCB's routing outline. Hard Lessons Learned Well, as any seasoned designer can imagine, what can go wrong usually does in PCB layout. I ran all my automated electrical checking routines, and they proudly flagged a flawless netlist with perfect connectivity to the top and bottom contacts. I con- fidently shipped the CAM data off to the prototype shop along with a basic stackup detail. Then, as was my naive custom at the time, I simply asked them to adjust the dielectric layers on their end to achieve my specified impedance targets. Lesson #1: Give Your Board Fabricator a Fighting Chance of Hitting Your Impedance Originally, I had chosen my trace widths based pure- ly on ideal print-and-etch DFM constraints. I foolish- ly left the other two critical variables of impedance control—distance to the reference plane and the material's dielectric constant—entirely up to the fab- ricator. I learned just how irresponsible that practice was when the CAM engineer called me back. He patiently explained that to hit my impedance targets with traces that wide, the finished PCB would need to be 0.100" thick. That was a massive problem, con- sidering the edge card standard strictly mandated a standard overall PCB thickness of 0.062". That was the exact moment I realized I couldn't just pass the buck. I needed to start utilizing at least a basic impedance calculator to choose line widths that actually complemented the required laminate material thicknesses. Back then, field solver options were incredibly sparse on the internet. Today, there is absolutely no excuse. Most, if not all, modern PCB design software suites have advanced stackup solvers built right into the design environment. It is every PCB designer's job to get the impedance math close to a solution so the board fabricator can simply dial it in from there. I re-routed the board with thinner trace widths and finally brought the overall thickness down to 0.062". I shelled out the extra expedite fees to make up for the lost time. When the fresh prototypes arrived, I stood right there over the shoulders of our soldering techs, watching them assemble the connectors and components onto the board. Everything looked great. On to the bring-up phase. Next, the engineer I had designed the board for began probing some of the assembled board's contacts. After some time, he looked at me, and in a patient, empathetic voice that I will never forget, said, "Sheesh, Kelly, I think you've got the A and B sides of this connector flipped!" He was right. I did. Lesson #2: Build Electronic Component Footprints, Not Turkey Footprints The error didn't happen because I accidentally ro- tated the pattern or reversed the pin order. It hap- pened because the A-side padstacks were placed directly on top of the B-side padstacks in the CAD library part, making manual visual verification in- credibly difficult without proper silkscreen legend or etched alpha side indicators on the outer layers. The pattern was flipped, causing top side routes to connect to the bottom side and vice versa. Fortunately, the engineering group I worked with had seen this kind of rookie mistake before, but they weren't going to let me off the hook without turning the event into a permanent teachable moment. They handed me one of the bare prototype boards, a Dremel tool, a soldering station, a spool of blue rework wire, and a female edge card connector. I spent the next two days of my life hunched over that bench, tediously hacking and jumping traces to create a hideous "Frankenstein" connector- converter just so we could test the electrical func- tionality of the rest of the PCB assembly. It was a true design turkey born of my own making, but it was an opportunity to learn a couple of lessons that would completely transform how I approached impedance control specification and CAD library discipline for the next 30 years. Gobble, gobble. I-CONNECT007 Kelly Dack, CIT CID+, specializes in DFx-driven PCB design and applications engineering at Pio- neer Circuits, Inc. To read past columns, click here. TA RG E T C O N D I T I O N

