The Long-Term Consequences of the Worldwide Chip Shortage

At the beginning of COVID in 2019 and the subsequent restrictions and challenges associated with it, it wasn’t clear what the long-term consequences of the pandemic would be; how long they would last; the extent of their spread and how dramatic their impact would be.

One of the more concerning impacts of COVID has been the worldwide chip shortage problem. While COVID certainly has contributed to the problems associated with the chip shortage, it’s not the only factor. This article will describe those other factors, how the impact has also affected PCB production and the different technical and business practices that can be put into place to hopefully help avoid the impact severity of another similar shortage in the future. For this article I got input from Lee Ritchey, Founder and President of Speeding Edge who has seen just about every kind of problem imaginable spanning the history of the electronics industry.

Contributing Factors to the Chip Shortage

The current chip shortage is not a new phenomenon within the industry. Over the years, for one reason or another the electronics market sector has been hit with a variety of challenges but nowhere as severe as what we have seen over the past couple of years.

The impact of COVID has been felt so heavily because it’s involved the virus perpetuating and moving from country to country or from one country to the rest of the world. As COVID numbers were racking up, companies that initially weren’t impacted were later on as the virus moved in serial mode across varying locales. Some countries closed down completely, some companies within certain countries closed down; others had severe restrictions on the number of people who could be working at any one time and still others were forced to find other means of shipping their products. But COVID alone hasn’t caused the shortage.

Ritchey states, “There have been two critical factors. First the need for chips required for the auto industry was severely undercast. The automotive industry guessed wrong and they cut back their orders thinking that sales would fall off. When those sales didn’t diminish the seeds of a shortage were implanted. The second factor is that the ICs used in today’s products are sole sourced (more about this below) and anytime that occurs, an industry can be held hostage to its own success.”

Why Applications Make a Difference

In contrast to the auto industry, Ritchey explains, “Companies like Arista and Cisco haven’t suffered the same kind of impact as those companies designing and building automotive products because they accurately forecasted their markets. If anything, the demand for the chips used in their products has gone up and they adjusted their operations accordingly.”

Certainly, COVID has played a big role in the shortage of chips available for use in automotive products but it’s not the only element in the equation. When coupled with other ingredients it created an electronics version of the “perfect storm.”

Ritchey continues, “One of the biggest challenges today is most of the components are targeted for particular applications and, as noted above, they are sole sourced. When we had TTL, there might have been five different suppliers. That dynamic no longer exists. The notion that there is a substitute for the components used in certain sectors just isn’t the case anymore. There aren’t two outfits making the same part. This is the result of everyone having to put all their eggs into one basket all the time. If you are stuck with designing a product and you can’t get the components you need, you are just going to have to sit and wait.”

Under the foregoing scenario the only leverage a company has as a customer is by telling their supplier they won’t use them on the next design. And, the only way they can approach a different IC vendor is with a brand-new design. Ritchey states, “It doesn’t help to take an existing design to a different vendor. You have to get in line behind their existing customers who are designing a similar product. And, it doesn’t help to go to new designs because you have to get into line to get the ICs often when capacity is already tapped out.”

It’s certainly true that during a shortage, some companies may stockpile the chips that they will be needing. This enables them to maintain the processes and practices they have in place for a while longer. But, not all companies have the luxury, opportunity or money to make this their long-term business model.

The PCB Side of The Equation

The PCB side of the story is considerably different with respect to the chip shortage. Over the years, some components and the materials used in the fabrication of PCBs became unavailable when new ones were introduced; some became obsolete and were no longer manufactured; some were the result of a catastrophic event completely eliminating a resource (as when the 2011 earthquake and ensuing tsunami wiped out the Nittobo glass factory that was the sole supplier of low Dk glass). In the early days of COVID, there was a shortage of the copper foil needed to manufacture PCB laminates but this impasse was essentially a “blip” when compared to the rest of the industry.

What’s important to remember is that, despite claims to the contrary, there is no such thing as a “plug and play” component that can be mounted on a PCB. It may fit within the same package but changes to the features and functions of the component necessitate changes to the PCB design. The slightest variation, even inside a package, can quickly render a design unmanufacturable and inoperable. And the small amount of wiggle room potentially available within a design can disappear when moving from one major component to another. And with demand outstripping supply in many instances, any change in component technology necessitates going back to square one with the design process and doing all the analysis, especially during the stack-up phase, that is necessary to ensure that the resulting design will be manufacturable, operable and reliable over the lifetime of the end product. Obviously, having to go back to square one means that there will be non-recurring engineering costs which can be significant, development times that are protracted and market windows that are potentially missed. In this scenario and to meet corporate demands, some companies may attempt to make design shortcuts in order to get their product to market quickly. This is an example of a “you can do it fast or you can do it right” product development scenario.

The Elephant in the Room

There’s been an attempt on the part of some to blame the current chip shortage entirely on COVID. While it certainly has focused a lot of attention on the situation, the problem itself has been brewing for some time and Taiwan is at the crux of the concern.

Ritchey explains, “The real fear centers on 60% of the world’s ICs being made in Taiwan. If China does what they want to do, they would control the supply of ICs. That’s why Intel is building a giant wafer fab in Japan and other ones in Hillsboro, Oregon, and in Phoenix. And they are bringing their plant in Albuquerque back online. It isn’t to increase the amount of stuff being made here. It’s very quiet pressure resulting from our economy and our government saying we have to reduce our dependency on Taiwan. The transition off Taiwan is already in motion. TSMC is building a plant in Japan and other chip manufacturing is being ramped up in South Korea and India.”

“The Taiwan vulnerability has been there a long time and now people are waking up to it. This was not a result of COVID. Some time back, five-year plans were put in place to address this. The impact of COVID just threw a lot more light on the problem,” Ritchey explains. “The impetus to bring ICs back onshore is because of what is happening in Taiwan.”

“The reason we went offshore for IC production was not cheaper labor,” he notes.  “The regulatory rules in places like Taiwan were such that you didn’t have to worry about pollution resulting from the fabrication process. In essence, we exported our polluting industries to other countries that didn’t have the same regulations in place. But, now, even China has gotten to the point where they are rigid about pollution because they realize they are only hurting themselves if they don’t control pollution. Malaysia has done that for some time and Singapore always has.”

Ritchey explains, “Wafer fab is not labor intense. Nobody touches a wafer from the minute it goes into the front and comes out the back. You can’t have humans in the process because they contaminate it. About the only cost savings resulting from having wafer fab in Taiwan might be the energy. Wafer fabs consume a huge amount of it.”

Moving IC manufacturing back to the States will alleviate many of the issues associated with having the U.S. (and other countries) being dependent on production in Taiwan. Ritchey comments, “What’s interesting about Intel is that they are going into contract manufacturing like TSMC. They’re not building those plants just to make more processors for their own needs. They are looking to become the fab facility for outside customers.”

The key challenge these days lies in the gap that exists between where we are now relative to our dependence on Taiwan and when on-shore facilities will be up and running in production mode. Ritchey states, “It takes two years to bring it a facility up to production once it is under construction. We’re still a fair amount away from solving the problem. This means that for the next two years or so, the component lead time is going to be pretty long and until we get enough capacity off Taiwan, we’re going to sit on the edge of our chair worrying about what happens if China starts messing around with Taiwan.”

An Example in Gauss Stack

As mentioned earlier in this post, changing from one component to another can require modifications to the PCB layout and/or stackup design. One of the key drivers behind this can be solder joint reliability. In the following example, Gauss Stack is used to simulate the solder joint reliability of two different (but comparable) components on the same stackup. These components have the same footprint and the same size and number of solder balls in their BGA pattern. However, they are packaged on different materials and have different solder joint heights.

Figure 2. Component 2 – Cycles to Failue

As one can see from the above example, Component 1 shows a median life to failure of ~2872 thermal cycles from 0 °C to 80 °C, while Component 2 shows a median life to failure of only ~1343 thermal cycles from 0 °C to 80 °C. Component 1 shows more than double the fatigue life of Component 2 on the same PCB. Let’s say your reliability target were 2000 cycles to failure for this particular thermal cycle. In this case, if you had to move from Component 1 to Component 2 due to a supply chain problem, you may need to look at changing your stackup, by, for example, switching to another material, or, potentially, switching to a different solder paste. Were you to keep the same PCB stackup, but switch your solder from SAC 305 to Innolot®, you could see the following result:

Figure 3. Component 2 – Cycles to Failure with Innolot solder.

This change of solder material increases the median life to failure to ~3557 thermal cycles from 0 °C to 80 °C. Such a change can still introduce additional costs, due to modifications of process due to the change of solder, as well as the cost of the solder, itself, which can be other factors in determining if changing components is viable or if it is better to stick to the original component and deal with the delays. 

Smart Designing

When shortages hit an industry, the smart companies are those that emphasize not only their current design efforts and maybe the next iteration beyond but also focus on their efforts downstream for successive product generations and what features and functionalities need to be included in them to proactively address customer needs and, hopefully, give them a leg up on the competition. This is certainly true for the companies who are designing PCBs for next-generation products. With IC availability being something of an unknown for the next couple of years, optimizing the PCB design process becomes an even more critical factor during the overall electronic product design process.

In this environment, getting the product designed right the first time and emphasizing current product development cycles as well as future ones is best done by proactively addressing those elements that link the comprehensive design-to-manufacturing-to operability cycles. The computational prototyping capabilities that are featured in Gauss Stack enable product developers to model and predict the impact of specific features on the design process that have not been previously available to product development. We are able to address a wide variety of elements such as glass stop and resin starvation; warpage prediction; dielectric stress and survivability; plated through reliability, microvia reliability (including stacked and staggered microvias), solder joint reliability and differential skew prediction. Being able to address any one of the foregoing elements is a significant benefit. Being able to address all of them not only reduces overall time-to-market and costs factors but also yields a PCB design that is optimized across all the disciplines that contribute to the successful creation of the final product.

Summary

If there is one message that can be learned from the current chip shortage as well as previous, similar events, is that the electronics industry continues to remain something of an unknown entity. Those companies that will successfully weather the current crisis and those downstream will be the ones that successfully embrace a product development process that takes into account critical market and business factors as well as technical ones.