Manufacturing Matters

No two board designs are the same. We could run a thought experiment. Given a token complexity level of a generic board that mixes a bit of digital with a slice of analog, every PCB designer would come up with something at least slightly different. The differences are unimportant; more along the lines of style points. The similarities, if you overlayed 100 separate efforts, would reveal the collective knowledge of the PCB Design community.

By that, I mean that the typical design imperatives would be present in most of the artwork. Bypass caps would be near the power pin and oriented for the shortest inductive loop. Vias would share some common characteristics, but there would also be many variations. Some Designers might approach the problem with a low layer-count stack-up and HDI technology out of habit. On the other end, you’d find one-sided boards with full test pin coverage and all of the wonderful markings that an open floor-plan presents. The two ends of the spectrum converge on the happy middle ground where most of us want to live.

Where do the rules come from?

Our goal is a boring factory that does nothing but fill pallets with products to ship to waiting customers. To get there takes an awareness of the IPC guidelines but, it is more than that. A little solder paste under your fingernails might be a good thing too. If you are not the type to roll up your sleeves, you can get a tourist view. Most factories have a sales team that is proud of the facilities and would love to show you around. Those salespeople also have a budget for feeding potential customers, so schedule something for mid-day. Seriously, if at all possible, visit your existing or potential vendors.

Those vendors are just as likely to be in-house assembly lines in which case, your capabilities are the hard-stop set of rules to life. On the other hand, when it is a merchant vendor, ask for a technology roadmap. These usually come as a table with today’s mainstream capabilities plus any advanced technology available now. Other columns cover the landscape next year and a few years out, so you know where the technology is going. You can collect these online as well. Sort them by region or technology class but be aware that these are forward-looking marketing documents subject to certain blah, blah, blah…

What do I do with all of this information?

As you survey the data for trends of interest, the consensus numbers will emerge. The interesting data are any of your product requirements that push the current capability beyond standard and are not on the immediate roadmap. Some of our hypothetical technology only comes in a few packages, especially when you consider attractive pricing and lead times. Worst case, there may be only one choice, and it may be marginal for the PCB manufacturing targets. Companies with the means will look at a situation like that as an opportunity to bring something in-house as a custom design.

Owning the critical links in the path is a competitive advantage. For that reason, it may be advantageous for the board designer to also be adept at module and transposer design. These are the little go-between boards that may be made of a high-performance material that would cost too much to use as the primary substrate. Putting the “jewelry” in its own bucket and using that sub-assembly as a no-touch solution in final assembly is one way of reducing hand operations on the main assembly line. The analog clock in the middle of the dashboard of the “Duesy” is made in Switzerland, of course. The rest goes together in Coventry or wherever the BitCoiners buy their cars. Knowing what the “analog clock” is and sourcing or building it accordingly is something that comes with experience. The clock is a metaphor for whatever complications a design brings to the manufacturing table.

Ok, that brings us to story time

Stuff happens, doesn’t it? Stuff happened to me. It is usually a good idea to get off to a good start when you meet a new RF Engineer. Everything may be important, but not equally so to everyone. It depends on whom you ask. In this instance, time was the main driver. The job was this little so-called Z-jig that we would use to characterize an RF amplifier device. “It’s just a test jig to measure impedance.” That’s what he said. I said, “Yeah, ok, I’ll squeeze these two little capacitors up to the device pads, doing without a solder dam just for this test fixture.” It took about two hours to complete.

...over 20,000 potential solder defects per month. Every month!

I wish I had spent another ten minutes insisting that we follow the usual assembly clearance rules for those two caps. The device worked great the way it was at REV P1. Now, all of a sudden, it isn’t a test fixture anymore, and we’re too far in to redesign. We’re going to use one of these devices as the pre-driver and two more as the secondary stage for a little card we called an 80 Watt Stick. Now, we’re going to lay four of these sticks side by side in a multi-carrier amplifier to support the CDMA roll-out.

Image credit: Author * Subject: A later revision of the Stick, the P3 with DFA friendly solder mask for the six caps

Keeping score at home? That adds up two caps per device, three devices per board, four boards per amplifier and a design win for 900 units per month. Letting two solder joints go outside of the typical manufacturing requirements brought on 2x3x4x900 or over 20,000 potential solder defects per month. Every month! Did I mention that we’re making rework more challenging because we can not unsolder the device without also removing the caps? Yeah, manufacturing matters. Lesson learned. The PCB designer owns the process as the subject matter expert. We have to be the voice of the fabricator and the assembly house whenever they cannot speak for themselves.

Epilogue:

We eventually respun the radio board (in a rush!) and included a different approach to the placement issue. Little inductors were printed on the board between the RF device and the capacitor. The P3 version was the final production run and became something of a cult hit among radio-heads. If you do an image search for the term Spectrian, their logo will show up and so will dozens of memories of my little mistake and maybe a single channel amp that sold pretty well.

I told you this story to give the importance of not just knowing the design rules but being the champion of the path to the boring assembly and easy money. Almost anyone can play Tetris or Connect the Dots. The designer’s touch includes a profit motive. While no two boards are the same, the ones that pay back the investors all have something in common in terms of reproducibility. Those are the ones we all want to design.