Can you design your way out of a chip shortage?
As an engineer in the early 1990s, I was faced with replacing an infrared transmit encoder. The receiver was still in production, but the manufacturer discontinued the transmitter with little notice.
One solution was to replace both the transmitter and receiver, which would require two new PCBs. The solution I chose was to emulate the transmit encoder with a Microchip PIC® microcontroller, requiring just one new PCB. I emulated the transmitter’s functionality in assembly code, and it was a successful workaround.
At that point, PIC MCUs were still relatively new to the UK market but, even then, they were price-competitive with discrete logic. Another option would have been to use a CPLD or FPGA.
Now, the current chip shortage is forcing engineers to think their way out of component supply issues.
As the Avnet Insights Report: Deconstructing the Chip Shortage shows, engineers expect the current situation to continue for 12 to 18 months. This puts pressure on OEMs but also provides valuable intelligence that may influence the strategy they put in place over that period.
To find out if emulation is still a viable engineering solution to replacing missing parts, Avnet spoke with Shakeel Peera and Alexis Alcott, two of Microchip’s experts. Peera is the associate vice president of Microchip’s FPGA business unit. Alcott is senior manager of product marketing in Microchip’s 16-bit MCU business unit.
Device emulation
Replacing multiple discrete integrated circuits using a single programmable device is common practice for CPLDs and FPGAs. But what part would you use if you wanted to emulate a standard device?
“It depends on your cost and performance requirements,” Peera said. “If your application requires low-level sequential processing with tolerance for a certain amount of latency or has many state-machine based control structures and needs to consume very low power, then an MCU might be the best choice.”
For more intensive applications, an FPGA may be the better option, as Peera said: “If you require a high level of integration, with massive acceleration based on power-efficient parallel processing and highly complex I/O, then FPGAs are the way to go.”
For low power applications, Peera recommends the PolarFire® FPGAs, Microchip’s most power-efficient compute platform. “They combine extremely power efficient compute with memory and processing structures that offer the industry’s lowest power along with high-speed I/Os.” Peera also noted that FPGA design-starts now outstrip ASIC and ASSP starts. “The supply chain crisis will accelerate this.”
When measured against an ASIC, using a programmable device can improve time-to-market and lower the non-recurring engineering costs. “Traditionally, designers would need to understand register transfer level (RTL) code or move existing code to an FPGA flow. Tools such as Microchip’s Libero SoC tool suite manage the synthesis and the place-and-route involved,” Peera said.
These tools can also work with C/C++ code and remove or reduce the need to understand RTL, as Peera explained: “We are now enabling this through abstraction tools like our SmartHLS and associated partitioning and modelling tools, that allow design entry with C/C++.”
This indicates that, if you can describe the functionality in C, you can emulate it with an MCU or FPGA. While lead times for MCUs and FPGAs are currently high, due to the shortage, their flexible nature suggests engineers should be less constrained to using a specific device.
Chip shortage mitigation
Another way to avoid supply chain issues is to accommodate multiple, alternative parts on the PCB. Is it feasible to do this for MCUs, too?
Alcott says yes, but it comes with conditions. “Designing a PCB to accommodate multiple MCU footprints is challenging and definitely more costly. However, it does add flexibility.”
The board could support MCUs from multiple vendors, or multiple MCUs from the same vendor. “This is feasible if the following things are taken into account,” Alcott said.
- Mechanical and assembly constraints, which includes the size and shape of the board along with the way the board is assembled
- Signal integrity for signals prone to noise, such as communications or analog traces
Other than to mitigate against supply issues, migrating to a different MCU might be necessary:
- to meet the product’s evolving requirements, which can demand increased memory, faster processing or additional I/O. An example would include moving from a PIC18 (8-bit) to PIC24F (16-bit).
- to deliver a lower cost. This can be achieved by reducing the number of components used and therefore the size of the PCB.
Microchip supports migration by making many of its MCUs pin-to-pin compatible within a family. This makes it easier to move to a device with more memory or enhanced peripherals, with the same footprint.
When software migration is necessary, Microchip’s MPLAB Code Configurator (MCC) can help preserve the design investment already made by supporting automatic code generation and I/O configuration. The tool auto-generates C code and helps with peripheral configuration. MCC Melody and MPLAB Harmony provide application libraries and peripheral drivers for the PIC, AVR® and SAM MCUs, as well as dsPIC Digital Signal Controllers (DSCs).
Future-proof your design
Using a programmable device to consolidate as much functionality as possible is an attractive design option under normal conditions. During a chip shortage, replacing a fixed-function ASSP with an FPGA or MCU is viable and feasible. Of course, this may explain why the Avnet Insights Report: Deconstructing the Chip Shortage indicates that both MCUs and FPGAs are high on the list of parts in short supply, according to the engineers who responded to the survey question.
Moving forward, preparing for shortages may become more imperative. Avnet’s FAEs are on hand to lend support.