Issue link: https://iconnect007.uberflip.com/i/1428512
18 PCB007 MAGAZINE I NOVEMBER 2021 Kolmodin: at's where I ask about the value of in-house vs. outsource. Do you want to keep investing in capital, or do you want to bring in an expert team that's always on top of it all? It might be a few extra pennies here and there, but it sure saves you having to put multiple millions of dollars into equipment every three years, because the acceleration of this technol- ogy curve in ICs and printed circuits is faster than what you can amortize your equipment. Holden: I had a testing challenge with some innerlayers using gold wire bond showing some shorts. en we would laminate them, mark the boards, and send them out into assembly. Eventually these boards would be tagged and rejected, and we would run over to look at the card. ey all said, "Too much time," which meant testing could not find the bad compo- nent and the time being spent on it exceeded the value of board. If we have a whisker short on the bare board and it's been laminated in, the tester can't discover it. To the test depart- ment, it's too much time. And too much time doesn't point to the actual problem on printed circuit board. It's a profit center decision. Kolmodin: Right. We call that whisker a micro short. We developed an algorithm and tech- nology with our equipment called micro short detection, which is basically those slimmer shorts on an innerlayer from a clearance to a plane or something like that. If you do your nor- mal voltage test you will actually fry that thing. e tiny whisker will pop like a fuse but won't be detected as a short because it was gone so fast. e problem is that it leaves a metallurgi- cal signature that could actually cause a latent short down the road, which is what Happy was saying. Holden: Yeah, the data said that was just passing the problem down. Kolmodin: Especially with the density and spac- ing that you have on these innerlayers. You've got metal there and you might burn it open for the moment but, le over time, heat, shrink- age, expansion, and some funky metallurgical stuff goes on in there where you can actually grow that short back. It's expensive from an assembly side when you have a $3,000 board out there that fails. Getting a return is always bad news but a lot of times OEMs aren't going to spend the time to destructively analyze the board or remove a component to say, "Okay, we have to actu- ally go down to a bare board level problem." ey'll just send the board back to the manu- facturer, marked as failed, and let the manufac- turer's lab deal with it. But we do see some of the bigger OEMs and assemblers wanting root cause of analysis. Holden: I think test will become a bigger thing with EV because I've found automotive to be far more critical about reliability and perfor- mance. If you have a single failure, you really have to jump through hoops to the root cause because they're worried about the warranty costs. Kolmodin: Oh, sure. Holden: If these things come back under war- ranty, that's a big expense. Kolmodin: Just from a mechanical standpoint, we saw the horrors of the Takata airbag recall and that was huge. We're right at the time of year where the new model year comes out, so if they've got a component problem in a car, that's a huge recall, and that's big money. Shaughnessy: Todd, thanks for speaking with us today. It was a good discussion. Kolmodin: You're welcome. PCB007 Todd Kolmodin is an I-Connect007 columnist. To contact Kolmodin or read past columns, click here.