Issue link: https://iconnect007.uberflip.com/i/1315894
36 DESIGN007 MAGAZINE I DECEMBER 2020 ics with Calibre and xCalibre, for example, one of the things that the product team did was to work with the semiconductor fabs to create rule decks that captured a particular process. It was a static thing. That's where we've been in PCB when it comes to DFM up until now; it has been a static set of rule decks. Ford: Exactly, there are two problems with rule decks. One is that they're never accurate because manufacturing technologies continue to evolve. The second thing is the rules become so complicated that nobody can understand or remember how to keep them updated. We have to find something which is more mean- ingful yet simple. We have to find the spe- cific language that we would use to exchange data from manufacturing and design, which is exactly what has been defined within the IPC- 2581 standard. It's a really exciting project. Holden: The other problem with rule decks is that they are a slope. In other words, make it robust and large, and it's easy, quick, and cheap. When you get toward the limit of technology, it becomes long, expensive, and complicated. Where do you put this rule, and where do you set the thing down on the rule? Most things don't accept rules as "it depends." Ford: Definitely. People never seem to agree on what those rules are. They will always want it to be a little different because their conditions are different, perhaps unique. What we would really like to see, however, is the common plat- form on which all perspectives sit, such that each solution remains interoperable. This has been the intent of the IPC Digital Twin. We've all seen digital twin-based solutions being advertised. Everybody is claiming a "digital twin" these days in some form as part of their solutions. The problem with solution-based digital twins is that they are all bound by the extent of their solutions. Within the proprietary bubble, they can do anything they like, but they can- not address or utilize anything beyond. Their contextualization is limited because if you can- not qualify events or changes that are triggered outside of the bubble, then how can you take those factors into account? For example, if you measure deviations in X and Y of component placements on a PCB relative to the surface of the board using the AOI machine, but you didn't see the fact that the batch of the PCBs has changed, which brings with it a new set of variances, how does that impact your statisti- cal analysis of the trend that you see for place- ment accuracy? Johnson: How are the EDA vendors responding to this moment in time? Ford: They're extremely positive. The IPC- 2581 committee is made up of a lot of these key design software vendors. The breakdown of these artificial barriers to interoperability is driving the industry forward. One of the great examples that I saw, getting back to assembly with CFX, is that you had the scenario before where individual machine vendors had their own flavors and formats of data, and nobody could understand anything from anyone else. Suddenly, four years ago, all of the machine vendors sat down together in an IPC meeting and said, "We need to share production infor- mation with our customers in an easier way. Machine vendor IP isn't contained in the data or its format; it is actually how the machine is designed. We want to have now data interoper- ability." Johnson: Andy, what Michael just said lines up with the "a-ha moment" we had yesterday. A marketing director at Altium made the com- ment that designers used to have a year to work on a power amplifier, and now designers are doing multiple designs in a month. There was plenty of time to stew on it and to think of the subtleties and the nuances in your design. If you're doing two a month, you just don't have that. Holden: Michael used the right word: collabora- tive. What we need and what IPC-2581C and future revisions are pointing to is kind of the Zoom screen for collaborative design in which you have electronics layout, including signal