Issue link: https://iconnect007.uberflip.com/i/1210212
FEBRUARY 2020 I DESIGN007 MAGAZINE 15 Matties: In terms of learning how to properly design it, power delivery is extremely compli- cated to the point where it's elusive to people. Ritchey: It's not that hard. After a half-day of my two-day seminar on power delivery, when students leave that class, they can immediate- ly get rid of their EMI problems. The problem is we have more misinformation on how to do this than we have good information. Matties: That's a big problem around the world today, generally. Ritchey: Yes. When Rick Hartley set out to do this 25 years ago, our task was to get rid of the misinformation. Clearly, we have not suc- ceeded. Matties: Is there a generational shift? Is this something that the older generation under- stands better than new designers, or are they relying heavily on the EDA tools to remedy the problem? Ritchey: Tools don't do any engineering. I have some ways that I like to use when people think tools solve problems. A long time ago, I was with Amdahl when we made the first machine that would compete with IBM. It was Silicon Valley, and we got a huge amount of press. There was a press conference and a Q&A, and a reporter said, "Mr. Amdahl, aren't you afraid you're making computers so powerful that they'll replace thinking?" He responded, "What you don't understand is that what we have here is an exceedingly fast idiot." You must have an engineer. Tools don't solve the problem. Matties: The only thing that solves the problem is education. Ritchey: I agree. I still work with the university where I graduated, and I gave the professors there all this information because they don't have it. Matties: Let's assume that they build a board with an EMI issue. What's the impact of that? Ritchey: The board can do its job, but if you don't pass the EMI standards for either the CE in Europe or the FCC in the U.S., you can't sell the product. Matties: In some cases, you don't have to reach that standard, so your board may work, but product may not work very well. Ultimately, it may not perform as well, but if you're trying to reach the standard, you have to solve the is- sue. Ritchey: Correct. Years ago, when I was at 3COM, they ignored that and shipped units into Europe that failed. They were caught and fined $10,000 for units shipped. We had shipped about 200 units. Thousands of prod- ucts are shipped out of Asia that don't comply, and they have stickers on them that say they do. In Hong Kong, there's a place called Com- puter Market where one stall sold rolls of these stickers. You don't need to even put your prod- uct through the test. Holden: A lot of times, people put it inside a metal box or a Faraday cage. Ritchey: That's how you contain EMI. Shaughnessy: It's not pretty, but it works. Ritchey: The problem is I see lots of products that fail EMI tests. If you have an EMI prob- lem, you need a transmitter. That means you have a source of RF energy and antenna. Peo- ple build these nifty Faraday cages, and then they pierce them with wires that go off to a mouse, a power supply, or something periph- eral, and the connection is the antenna. I have a number of strategies I use to keep noise from getting on those as far as the peripheral. The problem is I see lots of products that fail EMI tests.