June 26, 2020 | Denise Rael
Moving intelligence to the edge can lead to systems with better real-time performance, better power efficiency and enhanced security. But more intelligence requires more code,...
Engineering, like life, is all about compromises.
You might want that new sports car, but how do you justify the cost? And does it have enough space to take the kids to football practice?
Similarly, in any electronics system design, you need to balance multiple factors, and make trade-offs. You can’t have everything.
SWaP (Size, Weight, and Power) is a term more familiar in the aerospace and defence worlds but is a good way to think about the compromises inherent in any design. How do you balance the desire for more compact products, with the need for lower power consumption in many applications?
Of course, size and weight are linked. And smaller components may use less power, so where’s the problem?
The issue is we can’t just consider the SWaP factors in isolation. The cost adds an extra dimension, so you’ll hear people talking about SWaP-C. And you need to think about performance and functionality.
Moore’s Law has enabled electronics to keep SWaP relatively constant, while adding new functions, increasing performance (often dramatically), and meeting expectations for year-on-year price drops. Alternatively, designers can prioritize one aspect – for example, making products ever-smaller or adding more features.
In the consumer world, just look at the smartphone: a Radio Shack advert from 1991 lists 15 gadgets, of which 13 are now replaced by your phone. Whole product categories have been consigned to history: who buys a digital camera or stand-alone sat nav nowadays?
In the industrial world, the change is maybe less spectacular, but no less significant. Microprocessors have become low-cost, low-power items, and ICs get ever more integrated.
For analog parts, Moore’s Law has enabled high levels of integration for mixed-signal Systems on Chip (SoCs). But the physics of the analog world has meant another innovation has also been needed, for example, new architectures to handle the reduction in signal-to-noise, as geometries have shrunk and supply voltages reduced.
For the Internet of Things (IoT), and specifically, the industrial IoT, how do the changes in SWaP, cost, performance, and functionality match customer demands?
The IIoT is a step change in expectations, in multiple ways.
But there will be a massive increase in the number of devices deployed (revenues for the IIoT are forecast up 17.7% in 2018, to nearly $36 billion), many of these are sensors and actuators, but with a need for local processing capabilities.
Many of these will be in portable applications, so SWaP is important – with power often coming from a battery or even energy harvesting. And with so many devices, unit cost will be more significant.
Are the familiar price/performance and SWaP improvements even sustainable for the IIoT?
If you’re designing with standard parts, you may feel there are just too many compromises. You’ll have to select the parts that come closest to meeting your needs, and simply accept where they fall short – or are bigger, costlier or more power-hungry than you’d like.
There is always more than one way to solve a problem, and an increasingly popular alternative is to use custom ICs – which can provide exactly the functionality needed in one, highly-integrated mixed signal ASIC. This helps keep power consumption down and reduce system size and weight.
In terms of cost, there’s inevitably a one-off investment with a custom IC, balanced by a lower Bill of Materials (BoM). But the payback period is shorter than you might think – at S3 Semiconductors, we find it’s typically one to two years, even for fairly low volumes.
The more devices, the better the numbers work – so high-volume IIoT applications benefit particularly well from the economics of custom semiconductors.
The IIoT is an inflection point in terms of functionality needed, as well as creating tough requirements for cost, power consumption, size, and weight.
To meet these challenges, and to square the SWaP-C circle, it’s always worth considering custom ASICs at the start of a new project.