Issue link: https://iconnect007.uberflip.com/i/1511130
38 DESIGN007 MAGAZINE I NOVEMBER 2023 that can drive this, which are listed in approxi- mate order of importance: • Type of laminate material (PTFE, advanced FR-4, etc.) • Etch and drill feature size and processing • Layer thickness and layer count, especially HDI build-up layer count • High BGA/QFN count • Expensive surface finishes • Specialized requirements like heavy copper Here, the key is to identify the absolute necessities that are required for your device to function to its basic specifications. ere are design choices I make out of force of habit, and I have to go back and check myself based on realistic operating requirements. Keep in mind a few simple questions here: • Do you really need a PTFE laminate in your design, or did you use it because you read somewhere that all RF boards need PTFE? • Do you really need blind and buried vias everywhere, or can your placement and routing handle larger mechanically drilled vias? • Is there an SI or density reason I need sub-5 mil trace widths with controlled impedance? • Does every stub need back-drilling? Not everyone needs heavy copper PTFE with blind/buried stacked vias and 3-mil line/ spacing. Focus on what's important and save yourself and your wallet a lot of headache. Sim- ple steps like changes to your stackup, process limits, and material selection can create big cost savings, both in prototyping and volume production. 2. A LITTLE BOARD COMPLEXITY IS OKAY is might seem like it contradicts No. 1, but I don't think this is the case. Complexity is about give and take: You add in a bit of com- plexity in one area, but it makes your life much easier in another area. ink blind and buried vias on a fine-pitch component: your design might work fine with neck-down routing and mechanical drilling at 6 mils, even if the con- ventional wisdom demands fine-line rout- ing with stacked microvias just because your design looks like it might be HDI. As an instructor, I've noticed that some new designers tend to think that anything over two layers is a bad design, or that the sudden addition of two internal ground planes will drop a nuclear bomb on board fabrication budgets. ey end up spending a ton of time on perfect placement to mini- mize ghost wire crossings, and then much more time trying to route things with mini- mal coupling or overlap. e reality is that four-layer boards are not advanced anymore. A board with two interior stitched ground planes is really an entry-level board, and I certainly would not call these stackups "complex." ey also make routing on the top and bottom surfaces much easier, as you no longer have to worry about inter- layer coupling. en there are the EMI and SI benefits; many of the most basic SI and EMI problems can be suppressed or prevented just by using the right stackup. So, if you're new to design, give yourself permission to add a cou- ple ground planes; your EMC testing results will thank you later. 3. OVER-SPECIFYING AND OVER-CONSTRAINING Over-constraining a design can back design- ers into a corner, to the point where they over-specify every net in their design rules. e corollary to this is in routing, where some EMI-worthy design features are implemented regardless of need, and they must be handled manually. Something I've seen that is related to No. 2 is in mi xed-signal design. One of the big instances where I see designers create more