Issue link: https://iconnect007.uberflip.com/i/1515637
12 DESIGN007 MAGAZINE I FEBRUARY 2024 embedded resistors. ere are a lot of process- ing, reliability, and design questions. ere's the trade-off question, too: At what point does it make more sense to embed resistors vs. just shrinking down to, say, an 0105? Is it better to take the cost into the embedded process? Shaughnessy: I think some people are wary of the cost of getting involved with embedded components, even though it can save money in the long run. at's one of the trade-offs. ere's an initial design and manufacturing cost to get into the embedded design techniques. But once you're in, you will see improvements in the long run. You won't have as many failed boards; you can reduce size and item count. On the other hand, with- out embedded resistors, if I have to make enough room on my board, even using double-sided place- ment, to have 1,000 physi- cal resistors on there, how much bigger does my board need to be? How much more does my prod- uct need to cost, and what about the weight? If I can take off 1,000 resistors, even at a penny each, that's still $10 per board. Nolan Johnson: When does it make sense to go with embedded? One of the biggest trade-offs when you're considering embedded is actually the silicon geometry. e problem is that as the silicon has decreased, the rise times are faster. While you're transitioning from zero to one or one to zero, and you're either rising or falling, you're actively driving the transmission line. ere- fore, you don't necessarily need to worry about reflections or crosstalk. When you exceed that distance, now you have to worry about it. As the rise times have gotten faster, the distance the signal travels is now smaller, to the point that for some leading-edge boards, rise times were nearly 100 picoseconds. But with a BGA, you typically can't get a physical resistor close enough, because if the pin is coming out on rows 3, 4, or 5 close to the core of the BGA, the physical resistor has to be outside the body of the BGA. is means the whole distance that the signal has to travel, even on an inner layer, to get to the resistor is farther than that TTL. Now, the only way to get my signal integrity is to do the embedded resis- tor, so cost has nothing to do with it anymore. Shaughnessy: How does the mate- rial selection affect all this? e material has some influence on embedding components. If we go to the higher-end materials where the dielec- tric constants are in the low 3s and 4s, you get a little more room, but not much. Unless you start going into the RF materials—PTFE and other thermo- plastics—you don't get into the low enough dielectrics to really help much. But yes, material selec- tion will matter if you use the classic low-cost board materials that have dielectric constants in the 4.0 to 4.5 range. We're basically past the point of no return. ere are some really great materials out there, and I encourage my students to use them because they're worth it. ese right times are so ridiculously fast. ere are plenty of good white papers and design guides from the manufacturers of embedded materials. But normally, when picking a resistor or capacitor, the datasheet has a defined spec for the power dissipation and voltage rating for these devices. With the There are some really great materials out there, and I encourage my students to use them because they're worth it.