Surviving the Semiconductor Shortage on SoMs

By Brandon Lewis

Editor-in-Chief

Embedded Computing Design

By Taryn Engmark

Associate Editor

Embedded Computing Design

June 23, 2023

Story

“We had a customer that was looking to get COMs for their products,” says Dan Millar, CMO of MistyWest, a design and engineering firm headquartered in Vancouver, British Columbia. “If I remember right, the order was delayed three times then ultimately just outright canceled. 

“It just got strung along and then no product,” he continues. “That was pretty traumatic for them.”

Such was life at the height of the semiconductor shortage when the world got a taste of life without new chips for the first time in more than 60 years. The shortage drags on to this day for many types of semiconductors, but it was those early days of multi-year lead times from many major semiconductor manufacturers that presented an opportunity for those with the creativity to capitalize.

“Everything was 52 weeks,” Millar says. “I can't remember what part it was listed as a 99-week lead time, which I think is just because they didn't have three spaces. It was effectively infinite.

“It was one of those situations where we came to think that we should really be better prepared for possibilities like this,” he recalls. “We want to make sure we provide the best ultimate outcomes for our clients that we can.”

As a design engineering firm that performs contract technology development work for a range of clients,  MistyWest has the in-house expertise required to develop system on modules (SoMs) and has done so in countless client designs. The advantage of a SoM (also referred to as a computer-on-module (COM)) in a climate like the semiconductor shortage is that users don’t have to source the individual components needed to prototype, build, and test their designs. They’re all already on the board.

So MistyWest set out on a new project – and new business model, for that matter – that would eventually become the MistySOM (Figure 1). The line of solar- or battery-power-class modules and an open-source carrier board are built around Renesas RZ microprocessors to target edge computing applications like AI-enabled computer vision and human-machine interfaces (HMIs) where cost, thermal dissipation, and access to mains power are concerns.

Figure 1. MistySOM is a line of cost- and power-efficient modules based on Renesas RZ microprocessors that slot into an open-source carrier board for system interface and expansion. (Source: MistyWest)

Whether used as a development kit or production module, the fact that MistySOM boils down the bill of materials to a single supplier is an obvious win for customers at a time when components are scarce. But MistyWest’s existing relationships with SoC and other semiconductor suppliers also gave them an inside track in identifying reliable chip suppliers to troubleshoot supply chain issues for their new product.

“It was our senior electrical engineer who was aware of the Renesas RZ/G2L and RZ/V2L,” Millar explains.

Good Judgment from Experience

“At the worst of it, our lead times were about 26 weeks, which nowadays is considered pretty bad but at the height of the issues we saw a lot of customers finding 52 weeks and even greater than that looking for suppliers,” says Brian Witzen, Principal Business Development Manager for the RZ/V and RZ/A product lines at Renesas. 

The obvious question: Why did Renesas parts have such reasonable timelines when seemingly nothing else did? 

The unfortunate answer: The company was prepared because of previous incidents like the 2011 Tōhoku earthquake and subsequent tsunami and a 2021 fire at their Naka semiconductor manufacturing plant.

This resulted in Renesas’ active involvement in its supply chain.

“In the beginning you started to see the components that make up semiconductors’ lead times starting to go out as well,” Witzen explains. “The raw materials to manufacture substrates started getting pushed out, and that led to the substrate vendors starting to push out their delivery as well. We saw points, especially in the highlight of the pandemic, where these substrate manufacturers were quoting lead times up to two years away. And that leads all the way down into the semiconductor devices.

“This was not the first time they'd seen it happen, whether from natural disasters or other things,” he continues. “And we still do keep a supply of some of that material warehoused, ready to go. So we not always waiting for, say, a wafer to be developed and shipped to us. We may actually have a stock of wafers waiting, and we fill it that source as we deplete it.”

But as we all learned at the height of the shortage, the semiconductor supply chain is a long and complex one, and simply having additional materials is not enough to stave off a global crisis. Again learning from its past, Renesas has secured its “own fabs, assembly sites, and testing, so if something was to go wrong in one location we can move that development to a different testing center.”

“Even for the fabs and devices that we outsource to other fabs, trying to make sure that we have more than one source for everything from the fab to the assembly to the testing the devices,” Witzen says. “And if times got really rough, we could actually split it up and try to get product from more than one place. The MPUs that MistyWest used on their development, those are mostly outsourced. But even at that point, we're trying to do dual outsourcing to make sure we're not stuck with just one vendor.

“That's one of the things that we really saw as a big driving factor during this time,” he goes on. “So by the time we heard from MistyWest, they were actually one of many customers coming to us because of lead time issues.”

Component Selection: How Engineers Take Back Power

The 1.2 GHz Renesas RZ/G2L and V2L proved a good fit for the MistySOM’s target applications as both support video input and output, but the V2L integrates and AI inferencing accelerator that provides scalability for use cases that require extra performance. But, just as important, the two devices are also pin-compatible (Figure 2).  

Figure 2. Block diagrams of the Renesas RZ/G2L and V2L microprocessors that headline MistyWest’s MistySOM show product line similarities that lend to pin compatibility. (Source: Renesas Electronics)

Pin compatibility not only means different MistySOM variants can be swapped onto the same carrier board for improved design flexibility, but that MistyWest can source one MPU over the other in a pinch and still deliver a viable product.

“Our first point of contact was the manager for the RZ/G2L, which is very similar to the RZ/V2L as in they're pin-compatible, they have almost the same structure,” Millar says. ““This enables us having a pin-compatible SOM between the G2L and the V2L to have two products developed very quickly.”

Pin compatibility within the Renesas RZ product line extends beyond the G2L and V2L to include other devices, and also crosses architectural boundaries. Witzen notes that the portfolio also includes a RISC-V based device that’s pin-compatible with the Arm-based RZ/G2L, which opens up completely new doors in terms of design flexibility.

“So a customer or a SOM vendor can actually start with one and move to the other pretty easily, back and forth,” Witzen adds.

Back in Stock

The MistySOM is currently shipping to alpha customers, and the carrier board design files are now available on Github. Millar acknowledges that while the SOM and carrier are deployable, “the real intent is … you can test it out and then start stripping away components you don’t need when you start moving towards a real product.”

Quite the contrast from the height of the shortage when engineering teams were tirelessly searching for any component they could find.

Of course, the value exchange isn’t one-way, and it’s not just between MistyWest and its clients. Renesas also benefits from partnerships with SoM partners and engineering firms like MistyWest as they help drive devices like the RZ/GL2 and VL2 into system designs.

“One reason we look for SoM partners is because even though as a semi vendor we try to make it as easy as possible develop with our devices … a lot of customers just don't have the resources or don't want to put the resources on it to develop their own product,” Witzen acknowledges. 

“The MPU to the memory to some of the really high-speed peripherals is the hardest part to do,” he says. “And when we can have a SoM vendor like MistyWest take on that work and put the hardest part onto a small SoM board, it relieves a lot of effort off the customer. They can basically just take that SOM board from MistyWest and develop the peripheral set, which is the slower-end part of the design. Then they can just drop this board on and they're good to go.”

So, in retrospect, thank you, chip shortage. Your chokehold on electronic devices wound up increasing design flexibility for us all.

“The path there has been challenging, undoubtedly. We are doing something new for ourselves but compare that to the alternative of just sitting and waiting to see if we have supply chains get resolved,” says Millar. “But taking that step forward to be the best partner we can for our customers has, I believe, helped us a lot mature as an organization and helped us build trust with the people that we work with and the people we seek to serve.”

For more information on the MistySoM, visit www.mistywest.com/mistysom or find design files on Github.

For more on Renesas Electronics’ RZ/G2L and V2L microprocessors, visit www.renesas.com/us/en/products/microcontrollers-microprocessors/rz-mpus.

 

Brandon is responsible for guiding content strategy, editorial direction, and community engagement across the Embedded Computing Design ecosystem. A 10-year veteran of the electronics media industry, he enjoys covering topics ranging from development kits to cybersecurity and tech business models. Brandon received a BA in English Literature from Arizona State University, where he graduated cum laude. He can be reached at [email protected].

More from Brandon