Issue link: https://iconnect007.uberflip.com/i/1210212
58 DESIGN007 MAGAZINE I FEBRUARY 2020 they are more reliable is a strong argument to reconsider. I will present that to my customers in the future. Matties: You also gain real estate with land- less vias. The other thing that Happy started with was what he called the measured perfor- mance. I thought that was interesting because he said for designers, salaries should be based on performance. The better designers do at their jobs, the less the boards will cost, and then you should pay them more. One key mea- sure he mentioned was density. Is that a measure that you look at? Löwer: Not until today. That's something I will try to get from customers and my boards. I will go through my PCB designs and see if those measures went up. I hope to get more insight into his data be- cause I am curious about these relations and metrics on my PCB designs. Shaughnessy: Most com- panies don't, apparently. Löwer: Sometimes, there are rough estimates for the area of big ICs, so for density. I've seen a company that tried to define necessary space or real estate for each component into the li- braries. They used old designs to get this infor- mation. But during pre-planning, they failed to go for it and get the benefits out of this infor- mation. They did not go to the extent Happy did. Matties: Happy also asked if people in the audi- ence used thermal modeling tools, and every- body said, "No." He talked about the variation that you get on a board from side to side. What did you think about that point that he made? Löwer: That represented how he and other measurement people thought, coming from HP. I've seen thermal measurement or thermal simulation in products if the customer or engi- neers knew that there would be issues. There was one product in a vacuum that had these issues known early in the project. In a vacu- um, you don't have normal convection, so thermal is a big issue there. They have done a lot of thermal simulations. Roughly speaking, the simulations were shown to be correct af- ter tests. I had another customer who was design- ing their own ASICs, which were getting too hot. They had done some simulations to find the best cooling solution and die build-up. If issues are known early on, there will be simulations, but I usually don't see it being done. Matties: But if it has an impact on noise, or signal integrity, that's an impor- tant consideration when designing a board at that level. Löwer: It has to be done. But the way most boards are designed, it's easier to test it than to make a big simulation. It's a matter of cost and effort. For the simulation, you have to have the tool, which costs mon- ey, the expert for the tool, and you need time. However, the reality is that time, tools, and the workforce are scarce in the industry. Tools and people who know tools well usually are rarely available. Shaughnessy: Until recently, for the thermal management tools, you had to have a Ph.D. or a master's degree to understand it. A designer with no degree might have a hard time using most of the thermal management tools. Löwer: Also, detailed thermal simulations are not in the electrical domain. Most thermal sim- ulation tools I've seen are introduced into the