Over many years, I have seen some elaborate PCB library systems. However, the best ones were those not based on the size but rather the quality of the information. That old axiom is definitely “not quantity but rather quality.”
I recently sat with a VP of engineering, and I asked what he thought of his PCB library. He thought for a moment and said, "I believe it is pretty good." I followed up, “But do you know?” To me, there is a difference. If I tell you that the stove is hot and you touch the burner, it will burn you. At that point, you only believe it. Now, if you go and feel the hot stove and it burns your hand, what just happened? You went from believing to knowing. Within an industry that demands perfection, it is not good enough to just "believe" that your library is good enough; instead, you must know. Really, what is “good enough?” What is that? We got the board through fab and assembly, and maybe it went through some compliance checks. So, it must mean that it is correct, right? No, it could mean that you were just lucky. Taking some advice from President Ronald Reagan, "Trust but verify." Most problems that occur are directly linked back to making assumptions somewhere along the process.
I-Connect007 recently conducted a survey that asked about the challenges that designers face. One of the problems mentioned in the survey responses was issues with component footprints. I can fully understand why. Starting a design with footprints that you only believe are good enough rather than knowing is a recipe for disaster. There is no recovery from harmful components in your design.
Off the coast of a somewhat rugged area of North Carolina, there stands three lighthouses. It is a harsh area for naval vessels, and is a well-known area for shipwrecks. Contrary to belief, though, a ship captain coming into port will not go from lighthouse to lighthouse. Instead, he lines up all three lighthouses where he only sees a single light. Much in the same way, during the component QC auditing process or review, we verify that all the lighthouses lined up and we should only see a single light. Simple rule: If you see more than one light, you are off course, and something is wrong.
Figure 1: Lighthouse off the coast of North Carolina.
The Verification Standard
For any verification process to be successful, you must establish it on a known standard. What do I mean by standard? What is it we will verify a component with?
The first verification document for all components is the datasheet. It is usually the very first document that engineers and librarians go after. If the process does not start there, we will be setting ourselves up for major problems down the road.
But (and that is a big but), some datasheets have been known to be wrong. A single datasheet is often trusted way too much; usually, the problems with those datasheets end up on PCB designs and finally into assembly.
When this happens, and it will happen at some point in your career, document it. If you find such a problem, it could become a habit for that component vendor. No company can afford to fact-check datasheets and components by running PCB board runs. Furthermore, you might consider dropping them from your preferred vendor list. At least red flag them. You caught a problem that may exist in other parts they provide.
Another strategy is to always have multiple suppliers for any components in a design. That way, it provides a couple of things in our process. First, it gives us an alternate part in case our first choice falls through for some reason. The other advantage is it provides another datasheet to compare the component.
When multiple manufacturers are available, it is best to pull every datasheet and compare them to determine if anything is out of line—lining up those lighthouses and comparing and contrasting what we see.
Fortunately, we are not left to our own devices to determine whether a component, especially a footprint, is correct or a datasheet. Several standards are provided by the Institute for Interconnecting and Packaging Electronic Circuits (IPC), which has been the leading authority in the printed circuit board industry since its start back in 1957. I have never read a standard provided by IPC that I did not like. The criteria that would help in the review process are those in the Design and Land Pattern category: IPC-2221, IPC-2222, IPC-2223, IPC-7351C (surface mount components), IPC-7251 (through-hole components).
Figure 2: A review of IPC standards.
I would highly recommend starting with the IPC standards as the very foundation to set up your component model libraries. IPC-7351 and IPC-7251 give a very detailed breakdown of the component category.
Through what is called Standards Developing Organizations (SDOs), the standards are reviewed regularly and updated to align with changes in the industry.
Also, keep up with the revised editions of the standards to determine whether those changes affect any of the models in your library, such as the release of IPC-7351C from the previous version regarding the shape of any rectangular pad to have rounded off the corners. This change took care of sharp turns that acted like antennas and helped in routing with a 45-degree angle trace.
A good library system (meaning one managed very well) will have the characteristic that the lifecycle would automatically change to a "new" status whenever changes occur. That way, any changes done on any components will identify them as changing.
Quarantine ‘New’ Components
If you don't already do so, start placing “new” components in quarantine someplace where they cannot be used in new designs. Before they are ready for prime time, you must first do QC checks. I've seen countless times when components got created and pushed into a design without the correct QC checks. The results were disastrous. Why is it we have the time to do things over but never have the time to do it correctly the first time? Often, good PCB designs are sacrificed on the “altar of expediency.”
Lastly, follow the golden rule of review. The person who did the work cannot review it. It is essential to get another set of eyes looking at the component. It just so happens that the person who did the work does not have an objective view but rather a subjective one.
It is probably best to have a few people who would act as the review board; for example, to take all new components and the verification document and line them all up.
The problems that possibly get introduced into a design are seemingly endless. Going back to my analogy of the lighthouses off the coast of North Carolina. There is only one clear path safely into the harbor. On each side are the jagged rocks. Whenever you deviate from that course or plan, you have a considerable possibility of running aground. In your PCB design, do not let the urgency of time bring risk and possible problems. Stay the course, and you will have a good design.
John Watson, CID, is a customer success manager at Altium.