PCB007 Magazine


Issue link: http://iconnect007.uberflip.com/i/1309864

Contents of this Issue


Page 31 of 123

32 PCB007 MAGAZINE I NOVEMBER 2020 Can we make a three- millimeter or five-millimeter HDI board?" That's very dif- ficult. The limitation of the OEM migrating to smaller pitch packages, but those switching applications are very difficult. They cannot go to a half-mil- limeter package and build a PCB around it. The technology is simply not mature enough, and it's still not today. Even for skin effect reasons, you have to widen your in- ner layer trace routes. You can- not widen your trace routes if your dielectric is only 75 microns because that's a rule of thumb. Your trace width is near the same value as your dielectric thickness to match the 50-ohm impedance or the differen- tial impedance of 100 ohms. You have a physi- cal limitation of what you can do with HDI in terms of high-speed circuitry, and that was a trigger to develop the technology. From that point, we started looking at what can we do. We already did a lot of other things in the past when Joe worked at Cisco Systems in a previous life. Together, we developed our ideas around splitting holes and doing other types of circuitry. Those ideas are picked up again after so many years and developed into a new type of circuit technology. Matties: It's interesting that you put the bound- aries of being able to integrate this into an ex- isting facility. What sort of modifications or in- vestments do the fabricators need to make to implement this technology? Tourné: At the moment, it's for the capabil- ity they need to have a decent CNC routing machine. There are special routing machines now on the market developed for it with opti- cal alignment, so you have pattern-to-pattern alignment, higher RPM spindles, and well over 100K RPM spindles. Your routing gets more ef- ficient, and you get a better quality of rapid slot. They also need advanced vacuum via fill- ing or slot filling capability. with laser drilling, you had only a few hundred holes a minute at that stage. In the end, reliabili- ty won. That's why everyone is using lasers right now. The idea was to use micromachining tech- niques to make the Z-axis con- nection in a different way. We want to use all materials available in the industry. We don't want to have material lim- itations, and we want to simpli- fy plating as well because plat- ing lines are really complex in terms of the highest aspect ratio plating. There are a lot of limita- tions. We cannot do blind holes; that's what we want to address with VeCS as well. We want to make a blind connection without going to very complex sequential build-ups. We want to cre- ate a technology where you can make a blind connection very deep into the PCB. Typically, a blind hole can only be plated with an aspect ratio of a maximum one to one hole depth versus hole diameter. With VeCS, we can go easily to 10 to one or maybe even to 20 to one. It depends on the length-to-width ratio of the slot. At the same time, the problem with the via hole is that it is very capacitive if you look at the time domain result. We want to have a vertical connection that we can tune the impedance of. We have the same impedance as to trace width, so we don't have any reflec- tions. We can have a higher performance cir- cuit than what we can do with a normal plat- ed through-hole. Those were the main items to develop the technology. Barry Matties: What was the goal—reliability, cost, etc.? Tourné: The goal was, "Can it be manufac- tured?" We see a lot of microvia HDI technology in cellphones. But if we look at the data comm computing market, it's very thick boards, up to three millimeters or even higher. That was real- ly the feedback I got from the industry, includ- ing Mr. Joe Dickson, who said, "How can we default those circuits for the future? Joan Tourné

Articles in this issue

Archives of this issue

view archives of PCB007 Magazine - PCB007-Nov2020