Common Design for Assembly ‘Gotchas’
DfM means design for money. If it can’t be built, that’s a waste.
Great ideas come together with great timing; what’s left is great execution. Flipping the switch that sets the factory in motion causes a few pain points. These “opportunities for improvement” will dictate your agenda down to the minute with all the little things that go wrong. Let’s say there is a factory downstairs from you. Further, the factory is doing slow and laborious rework on old printed circuit boards, aka PCBs. There is a solution! New PCBs. We’re going from P0 (zero) to P1 (one). That’s where we, the designers, come in.

At new PCB time, the first order of business is improving the electronics in some way. The fix could be better performance, lower cost, higher reliability and, in some cases, all the above. Venturing into wireless technology and gaining FCC approval to play in its allocated spectrum is no slam dunk. Beefing up the power grid is a typical step. A good power distribution network has been known to cover for otherwise iffy routing. Every engineer will have some considerations carried forward.

Remember those last-minute patches to fool the design rule check (DRC) into giving a pass to the intentional short from one ground domain to another? No? Now would be an appropriate time to address that. Test access, anyone? More hooks to keep tabs on current density are always a welcome addition.

The second order of business is getting the fab shop to build to the print. The P1 version is when the final form factor is in play. That only matters to the extent that it sets the size limit. For your information, the limit shrinks constantly. Having collected data from the P0 build, there is some confidence in the all-important stack-up data, along with impedance calculations. Capturing all that for posterity is a big deal.

A perspective I like is to think of DfM as design for money. Manufacturing is where money changes hands. Product goes in one direction, and money flows back toward the source of goods. If you’re any kind of capitalist, keep that machine running. Do not turn it off or even slow it down.

Some assembly required. Finally, handing a set of clear plans to the assembler is way more important than building a flatpack shelving unit. Numbers of holes on the drill chart need updating. X-y machine data. Critical dimensions and other mechanical relationships are added for clarity. Anything that bubbles up out of the original design gets squared up on the second revision. In some cases, retrofitting new data can take more effort than the original creation.

The simple things. Some smart aleck said a man with two watches never knows what time it is. Producing data in two popular formats is tantamount to giving your vendor an extra watch. Once and only once. That is to say, all data must be congruent in terms of part numbers and revisions and properly put to bed. One complete set of information without overlap/repetition of the data.

An assembly drawing is just that. The soldering profile, stencil thickness, paste composition and so on are relevant, but they ought to belong to a process document. This advice is more about passing ISO audits than helping the assembly line worker do their job with a one-pager on a particular PCB assembly. Separate “what is” from “how to.”

If you choose the other path, circle back to your assembly line to see if any alterations were made to the stencil artwork. Any number of global micro-edits are at their disposal when it comes to dialing in a stencil. I try to avoid getting the assembly drawing involved with the churn of process methodology on the assembly floor.

A close-up of a circuit board
Figure 1. A cap-hungry MCU lives on the other side.
Close-up of tooling holes
Figure 2. How many tooling holes do you prefer?
Semiautomated assembly in a factory
Figure 3. Semiautomated assembly.

If there’s a conflict between the information on the PCB silkscreen marking and the information on the assembly drawing, it can disturb the people involved with making those products. Your only defense against this type of transgression is to pour over the details of file names to make sure nothing is ambiguous.

I don’t know who said it, but a little gem someone gave me is this: “If something can be misunderstood, then it already has been.” The automotive industry borrows a word from the Japanese: poka-yoke (ポカヨケ). The essence of this concept is something will fit together if and only if it is being assembled correctly. It could be the reason three tooling holes are preferred to four on the assembly subpanel. In English slang, “idiot proofing” works as a harsh translation.

So, first electrical debug, then mechanical fit, and finally placement, as it relates to assembly concerns, are solved. The reason for this hierarchy, I think, is the immediacy of the team members. Inevitably, PCB designers are first linked to electrical engineers, even (or especially!) when their background is mechanical engineering. Our world is a blend of three things that are occasionally (read: always) at odds with one another.

Because assembly takes place at some distance in both space and time, it is easier for that facet to fall off the radar. The board designer is their champion, from the smallest component placement decision up to the final design review. We’re at our best when we can integrate all these, so the line keeps moving.

John Burkhert JR.
John Burkhert JR.
is a career PCB designer experienced in military, telecom, consumer hardware and, lately, the automotive industry. Originally he was an RF specialist, but is compelled to flip the bit now and then to fill the need for high-speed digital design. He enjoys playing bass and racing bikes when he’s not writing about or performing PCB layout. His column is produced by Cadence Design Systems and runs monthly.