Supply chain interruptions and staffing shortages make SOMs an essential design strategy

If you want to see the impact of interruptions to global supply chains and widespread labour shortages, you don’t have to look any further than your neighbourhood restaurant. If I walk to my favorite place down the street, the front door of the restaurant diagnoses those global issues with a clarity that puts complex macroeconomic dynamics into practical terms everyone can understand:

“The entire world is shorthanded right now, so please be patient with those of us who are working hard for you. And thank you for understanding that our menu is limited right now due to shortages of many ingredients.”

Those macroeconomic challenges are having a very practical impact on your local pub, and the impact is just as significant on the design engineering world. Despite some headlines claiming the chip and component shortage is over, those appear to be premature declarations of victory. Supply chain interruptions will likely continue to plague the electronics industry at least through the end of this year and perhaps beyond. Many industrial grade chips are manufactured on older process nodes (40nm, 65nm, and even larger), which do not have any new semiconductor capacity being built. The shortage of chips and other components has had a profound impact on product engineers, particularly those that work on wirelessly-connected products that use multiple kinds of chips for both product functionality and for wireless communications functionality.

The global labour shortage is also an enormous challenge for product engineering teams. Engineering support is hard to come by regardless of whether companies try to hire it or bring in contractors or engineering firms for extra arms and legs. The result is that many engineering teams are just as shorthanded as the restaurant down the street from my home. This is especially true for companies with smaller engineering departments that do not have the resources of larger companies to compete for talent or pay high billing rates to borrow it.

It would be great if engineering teams could respond to those twin challenges with a sign asking their business leaders to be patient, but no such luck. If anything, pressure has increased on engineering teams because of the economic pressure on companies emerging from the pandemic. Lost time from the pandemic combined with threats from competitors are compelling business leaders to push for faster product development timelines. Engineering teams are therefore in the uncomfortable position of being asked to do more with less and get products to market faster. This puts engineering teams between two rocks (supply chain and labour challenges) and a hard place (internal pressure to accelerate product development). All three of those dynamics are a fact of life for the foreseeable future, which is why engineering teams need a strategy that can navigate through all three. A SOM design is a strategy that will help many teams accomplish exactly that.

System on Module (SOM) is a design strategy that most product engineers are familiar with, but its use in the past has often been limited to low volume products or niche use cases make it a preferred approach. It should be considered for a much broader set of projects now because it gives engineers a successful strategy for accelerating product development while also sidestepping the chip and labor shortage. Rather than utilising a lengthy and expensive design cycle of having the processor, memory, and power controller chip down a mainboard along with a  wireless communications module, which all need to be sourced separately, a SOM combines all of that into a single integrated board that is a pre-designed, pre-certified modules that eliminates the need to do the complex engineering around chip integration, embedded wireless programming, antenna selection and more. As a rule of thumb, I would always estimate to engineers that a SOM design can take six months to over a year off of typical design timelines, but that positive impact can be much larger given the global challenges I discussed above. One company I am working with is estimating that a SOM strategy is allowing them to cut 18 months from their go-to-market timeline.

SOM can be particularly impactful for companies that develop products that are manufactured in smaller production runs than that of a mass-produced product (such as the one above that is slashing its timelines by 18 months). These smaller-volume products include a wide range of wirelessly-connected industrial sensor products, medical devices and other commercial IoT devices. Companies making these products often have the least power to negotiate with suppliers for chips – compared to larger companies making massive chip purchases. These companies also have far fewer resources to compete for engineering support, leaving them short of engineering expertise that is necessary for a chip-down, from-scratch design strategy. I should also mention that using a SOM approach gives design teams the ability to quickly incorporate advanced functionality and more stringent security into products without the complex work that would be required to accomplish those with a chip-down approach. Given the internal pressure on engineering teams to deliver products that are winners, this additional benefit of a SOM is not insignificant.

SOM design also has a number of benefits beyond allowing engineering teams to sidestep shortages caused by global supply chain issues. One key advantage is how a SOM design approach allows engineers to take advantage of pre-built security features in these modules. This is particularly important for teams designing medical devices, industrial sensors and other devices that must meet stringent regulations, roll out software security fixes, and incorporate standards for security like secure boot and FIPS. Building out these security features can require months for a chip-down design approach because of how much of the work is time-consumingly done from scratch. It is slow, expensive and risky. SOM design using pre-designed hardware and software solutions can deliver those security features out of the box, saving months of development time in the process.

Resource partitioning is another advantage of SOM design. Resource partitioning on the board gives designers the ability to build layers of protection and isolation within the overall design. The most impactful form of this is the ability to run a Linux OS and RTOS simultaneously on different parts of a multi-core heterogenous application processor. This allows the device’s most critical functions to run in real-time on the microcontroller without being interfered by user interruptible processing priorities like touchscreen displays.

Virtualisation within the device’s multi-core microprocessor is another major advantage because it allows different features to be fully supported by their own dedicated versions of Linux that are firewalled from one another. For example, connectivity can be isolated to its own Linux instance while display and user input are isolated to different Linux instance. This ensures that critical features do not have to compete against one another and are prioritised with their dedicated version of the embedded Linux OS. 

SOM design principles are unlikely to be completely new to readers of this article. What is new is shifting the way engineering teams think about SOMs. It is no longer just a niche design strategy that you consider using only in a few narrow scenarios. The global challenges we face are putting SOMs center stage as a must-have design strategy for a much broader set of IoT devices.