Stacking The Odds In Your Favor: Mastering The Art of Stackup Design
As everyone knows, the PCB stackup is the substrate upon which all design elements are assembled. And, it is incumbent upon the product development engineer to own the PCB stackup process which includes calling out the exact type of laminate, prepreg and copper needed in every layer of the stackup as well as taking into account all of the electrical characteristics of the PCB to ensure a design that is fully functional from both a mechanical and electrical standpoint. In the final analysis, a poorly designed stackup using inappropriate materials can degrade the electrical performance, the power delivery, the manufacturability and the overall reliability of the finished product.
This blog will address the history of the PCB stackup process, the challenges associated with a poorly executed stackup, and the ways to ensure that your stackup is optimized for all product development operations.
When Stackups Became a Concern
In the early days of PCB design, all that mattered was that the final manufactured board be of the right thickness at the lowest possible cost. Within these loosely-structured parameters, material selection was left up to the fabricator. There were two materials of choice: FR-4 and polyimide. Since the specification of these materials was so unstructured, it was possible to get a different board from every fabricator primarily because the PCB was built from whatever material the fabricator had in the Storeroom.
An interesting aside about polyimide: Everyone has always assumed that polyimide was the laminate of choice for aerospace applications because it withstands heat better than other materials. In truth, the limitation of the heat is based upon what the components can tolerate. Any of the materials that are designated as “not polyimide” are more than good enough for almost all components. The real reason for polyimide as a laminate was that the boards made from it were most often repaired in the field and field technicians weren’t expert at soldering. If a board wasn’t polyimide, when parts were replaced, the pads would come off the board. Thus, selecting polyimide was based on its tolerance for rework not because it could withstand high temperatures.
The foregoing initial practice of material selection worked fine as long as we didn’t care about any electrical aspects of the PCB such as impedance, crosstalk, loss and power delivery. Once these factors came into play, it became a whole different ballgame that required a whole new level of skill sets.
The first issue of concern was the resistance to the flow of the electromagnetic energy along a transmission line or through a component, aka impedance. And, the burden was placed on the board fabricator to get the impedance correct. As this was not within the skill sets of board fabricators (nor should it ever have been expected in the first place) the process of getting the impedance right was hit and miss. Even today, if the impedance compliance onus is placed on the fabricator, it’s still a catch-as-catch-can proposition.
As we started to go up the PCB speed curve, the next area that needed addressing was the unwanted interaction between signal wires or traces traveling in parallel—aka crosstalk or coupling. There never was a point in time during which the fabricator could successfully deal with this design consideration. Thus, this particular design aspect was the first that required engineering teams to factor stackups into their overall product development efforts.
Following impedance and crosstalk, power delivery and plane capacitance had to be accommodated as PCBs became increasingly fast and more complex. Today, we are concerned with loss in the channels and skew, and how they impact stackup operations. What all of the foregoing says is that the product development engineer has to be the master of all of these design aspects. Somewhat surprisingly, there are still some engineers who pretend that they don’t have to address these factors. Not surprisingly, those who don’t are often met with failed products.
So, all of the electrical considerations have to be engineered and they include:
- Power delivery
- Differential pairs
- Loss in high-speed channels
The key thing to remember is that engineering rules-of-thumb will never satisfy the foregoing.
But, it’s a fair question to ask where new engineers can gain these skill sets as they will definitely need them. The references listed at the end of this article are the best resources available. While there are a lot of other books in print they are often full of errors and bad advice. If they are not listed in this article’s references, it’s safe to assume that they fall into the “don’t use” category. And, it goes without saying, the average component applications notes are not good sources of information as they are commonly based on rules-of-thumb.
Once all of the above electrical considerations have been met, the following can then be determined:
- The number of signal layers needed.
- The number of power and ground layers needed.
- The thickness of copper needed on each layer.
- The type of laminate that will satisfy all of the mechanical and electrical requirements.
It might seem, based on what we’ve discussed, that the stackups for complex, multilayer boards would be the most difficult to determine and fabricate. In actuality, the hardest stackups to do are those used for 4-layer game boards. In these configurations, there is no plane capacitor (there can’t be one); there are no other layers than the surface layers that can be used for traces and very complex parts have to be hooked up together without the freedom of changing signal layers. In this scenario, there has to be the utmost cooperation between the person doing the PCB layout and the person selecting the pinout of the ICs. Without this, unscrambling a bus will require multiple signal layers.
The PCB stackup process is multifaceted and, for today’s high-speed, complex boards, is an integral part of the overall product engineering discipline. Taking all of the electrical engineering requirements into account early-on in the design cycle will guarantee a product that works right the first time while meeting all the mechanical and electrical performance criteria.
Smith, Larry D. and Bogatin, Eric, “Principles of Power for PDN Design-Simplified: Robust and Cost Effective Design for High Speed Digital Products.”
Bogatin, Eric, “Signal and Power Integrity Simplified, (2nd Edition).”
Ritchey, Lee W. and Zasio, John J., “Right The First Time, A Practical Handbook on High-Speed PCB and System Design, Volumes 1 and 2.”