Author: Ole-Fredrik Morken, Executive Vice President of Supply Chain, Nordic Semiconductor
Wireless connectivity underpins everything, including the Internet of Things (IoT) and artificial intelligence (AI). But advanced wireless devices demand an equally advanced supply chain, one that may be forced to use AI itself.
Compare a modern Bluetooth chip today to one from just over 10 years ago and the two would have little in common. The old chip was designed to add Bluetooth wireless connectivity to a product and be low power enough to support battery operation.
Today a leading-edge Bluetooth device can feature upwards of five microprocessors, multiple types of memory, and includes a range of extra electronic functions built-in. That means fewer components to source and integrate, but a vastly more complex device.
In fact, in Bluetooth, there are no chips anymore, there are Systems-on-Chip (SoCs). It’s like going from a device that was once the size of a village that has now grown into a full-blown city with all its added complexities.
It’s the same story in every major wireless IoT connectivity chip market. From cellular (which now has a low-power IoT-targeted version called cellular IoT) to Wi-Fi (also with a low-power IoT version). And these chips touch almost every market from consumer (e.g. smart home) to medical (e.g. wearable monitors) to retail (e.g. electronic retail labels) to industrial (smart lighting) to smart agriculture.
These wireless chip ‘cities’ are disruptively enabling applications that can do 10x more with 10x less. Applications that are helping to create a safer, greener, healthier, happier, more productive, and efficient world. But they are also creating a much more complex supply chain that is harder to track and trace. This article will outline a few of the key areas it pays to be aware of.
AI will improve process control
Nordic Semiconductor’s chip fabrication process – from semiconductor wafer to finished package – produces petabytes of data (one ‘peta’ is a thousand million million). There is potential within this big data to use AI to provide invaluable insights into the chip production processes.
This includes improvements in how changes in critical quality parameters are trapped if they happen during production. And spotting complex, multi-dimensional correlations that will lead to the source of a problem before it becomes a problem. On the current wireless chip complexity trajectory, AI is almost certain to allow for radical improvements in process control, ramp cycle times, and generally enhanced product quality.
One area of wireless chip production where complexity can really bite is testing. As chips get more complex it becomes harder to test and verify that they are functioning properly before shipping to customers. There are so many variables in play, each of which can interoperate and affect each other in very subtle ways.
There will come a day in the not-too-distant future where AI becomes deeply embedded into the test and verification process as well. Certain performance parameters can only be checked in very specific and specialised ways to simulate real-world operations. The more verifications made, the more it’s possible to guard against the worst and most expensive failure of all: field failure.
In the semiconductor business this may require companies to own the testers operating within the facilities of their chip manufacturing partners.
Nordic Semiconductor has long adopted this practice which proved vital during the post-COVID-19 global semiconductor supply chip shortage. By maintaining complete control of its fleet of test machines it was able to overcome this critical capacity bottleneck at the end of the manufacturing line, allowing chips to ship virtually straight to a customers’ product manufacturers as they were made.
Verify your suppliers are as green as they say
While it’s standard practice for a chip manufacturer to request test reports and materials declarations from all of its suppliers, this should be backed up with periodic checks to make sure what suppliers say is so, is actually so. Semiconductors are becoming so complex that it’s not impossible for something to creep into the supply chain that shouldn’t be there.
The way to guard against this is to go beyond the minimum regulations and require your chip suppliers to periodically check chip samples using an independent test house. This can then be used to prove to your own end customers, now or at any time in the future, that you have made sure what you say is completely accurate.
Walk your green talk
When it comes to ESG (Environmental, Social and Governance), there are many industry-standard minimum good-practice regulations (e.g. around the use of harmful substances like lead) for what a semiconductor chip supplier should do to minimise its environmental impact.
But climate change and the environment are too important to work to minimum standards. You have to do your bit.
One thing my company has pioneered with its two biggest component assembly and test partners – both world leaders – is to convert to component reels made from recycled plastic to reduce plastic waste. (Reels are the big old-fashioned, cinema film-looking circular reels that hold components so that they can be fed into high-speed placement machines used to assemble printed circuit boards). A good semiconductor supplier should be working very hard to minimise its environmental impact.
No risks on quality
A robust supply chain that is properly managed and monitored will be reflected in a minimum cost level of a component. If you’re sourcing a cheap component, then you have to question if any supply chain shortcuts were taken? And can you really afford to take such a risk? While a higher cost doesn’t always mean a better product, in the semiconductor chip industry it can be reassuring because it can mean critical compromises haven’t been made.
This will insure you and your company against major problems down the line. Seek competitive pricing of course, but don’t gamble with quality, field reliability, and environmental credentials.
