Issue link: https://iconnect007.uberflip.com/i/1505220
AUGUST 2023 I DESIGN007 MAGAZINE 25 footprints and layouts, ODB++ for manufac- turing, and many others. Today, the engineer has to take all that disaggregated data and turn it into valuable information that's called the digital twin. Shaughnessy: Do designers see or call what they are doing digital twin? What's the goal? It doesn't really matter what they call it. But I bet they are doing some level of physical pro- totyping. e goal of the digital twin process is zero spin. Now, you could say one proto- type spin is really, really good. Most teams say three to four spins is normal. But zero spin is the goal. You design it, you digitally verify it for all kinds of conditions like electrical, thermal, vibration, stress, man- u f a c tu r a b i l i t y, a n d then boom, you're off to manufacturing to volume. Design teams live in a virtual, digital world, so it doesn't bother me that a P CB de s ig ner do e s n't say, "Hey, I'm doing digi- tal twin." Do I need to con- vince them that they're doing digital twin to sell my stuff ? No. But I would like them to leverage the digital twin they're working with as efficiently as possible. at means verifying the digital twin during the design process, rather than via multiple physical prototype spins. Engineering teams oen "bake in failure" by planning for three to four spins. Now, that means you were doing pretty good design work. But three is the average, and a lot of designs go through six or more spins. But the Holy Grail is having teams not only use verification tools during the design process, but also leverage generative technolo- gies and AI to enable a system that can make those decisions for them. Of course, we're years away from that. Shaughnessy: Most designers are still using Gerber, and very slowly switching over to other digital data formats like ODB++ and IPC-2581. Gerber is a digital twin model, it's just a less complete one. ODB++ is an intelligent prod- uct model, and IPC-2581 is a fairly complete model as well. at means they take less work on the shop floor to recreate intelligence for manufacturing. Any time you have to re-create something, it's indicative of a broken digital thread, and that's an opportunity to introduce errors and waste time. But those formats still can't model everything in the manufactur- ing process. ey don't have, for instance, the enclosure; it's just the board. It's just one board, so if you have a multi-board system, those are coming across as separate files. A manufactur- ing engineer oen doesn't see the complete digital twin of the system—they have to rely on different models coming across at different times. So intelligent decisions about the complete system are limited. Shaughnessy: Is there anything that designers can or should do differently regarding digital twin? Mainly, I would want designers to be aware of the other disciplines that are hanging out around them. And not just the easy one, manufacturing, but also mechanical, cabling, thermal, signal integrity, all of these different things that have an impact on a design, and a digital twin can be built up to verify that. So, it's all about learning to leverage these different models in a way that that can produce a better result in the end. A lot of companies are going through different digital transformations. It's not just in engineering; it may be in their pur- Most teams say three to four spins is normal. But zero spin is the goal.