Issue link: https://iconnect007.uberflip.com/i/1525004
62 DESIGN007 MAGAZINE I AUGUST 2024 will cause it to take longer for the field to travel. Because it has larger spaces, it takes more time to fill those spaces up, and there are usually other signals that will be in those spaces, and they'll all contaminate each other. ere will be interference with other signals. You said that when you were younger, you focused mainly on the circuit and not on the field, yet you still managed to get designs right. What changed? Years ago, when the rise and fall times were not too steep, you could get away with that approach, but now you can't. We used IC technology that allowed us to do stupid things that didn't make it break. e main focus was always, "Does it work?" not "Is that the right thing?" What didn't make it break became the rules that we designed for. As we changed the technologies of the cir- cuits, the switching speeds got faster and energy levels got higher, we didn't change the rules, and we started seeing a gradual increase in failures as this new technology permeated the products that we've designed. Now, you don't have any choice, especially with PCIE and all the other fast buses switching at 60 or 70 picoseconds. At this level, you can't screw around. I'm working right now on proces- sors that have 5 nanometer geometries. So, the energies on these devices are switching in hundreds of femtoseconds. We have a part right now that has 40 watts on the core supply. ere's a huge amount of energy that you have to deliver and because of the geometry of the transistors, it has to be really close. How close can you put those capacitors? Are they on the bottom of the board? at's a 20 picosecond round trip. Is that close enough? Do I have to put capacitors on the top of the board where I've got a 2 picosecond round trip? ose are starting to be decisions that must be made. I'm working on a few products where I've had to put the capacitors directly between the balls. I put 01005 caps between the ball sites for the power supplies, because of the nature of my structure. By the time I get to the capaci- tors on the bottom of the board through multi- ple connectors, I'm 40 or 50 picoseconds away. People want to put a 4-ohm load on a 16-ohm driver and it's trying to deliver four times more power than it's rated for. at means it's going to heat up 4x or more than it's supposed to, and it's going to let the smoke out. What advice would you give to designers who have been primarily educated to focus on circuit theory? What do they need to understand? e first thing is to repeatedly listen to my song "All About e Space" so that earworm gets stuck in your head. at was a fun song (sung by Dan's daughter to the tune of "All About at Bass"). It's not complicated. Sunlight, which is electromagnetic energy, comes from the sun. None of that is being delivered by wires. It's not electrons that are carrying energy on a printed circuit board. It's the same energy. erefore, it's still electromagnetic field and it travels in the dielectric. Is there electron movement? Certainly. e poor electrons are being kicked around by the evil field. ere is some movement of elec- trons, oen described as electron dri, but they move at a relatively slow velocity, because electrons have mass. Don't focus so much on the electrons. It's all about the space. ere's a lot of information out on the web. e bottom line is this: When I design a system, it works. It's not that I'm super smart. It's because the rules are so simple. When you follow them, the results are predictable. It will work. Good stuff. Thank you very much for your time. Always a pleasure, Andy. DESIGN007 " When I design a system, it works. "