Moving chips from industrial to industrial IoT

IHS has put out its 1Q2016 Application Market Forecast predicting the highest growth rate segments for semiconductors over the next five years – and what was once old is new yet again. There it is, in the top right corner: industrial, projected to outpace even the automotive sector.

What should that mean for new chip designs? Embedded types have seen these cycles before, where the buzzworthy mobile and consumer sectors burn white hot for a while then burn out for a cycle, while the decidedly unglamorous sectors like aerospace and defense, automotive, medical, and industrial chug along. This time, there’s a difference. Every time someone says the Internet of Things, an angel gets its wings.

The IoT isn’t really an application segment, but more of a set of aligned technologies for connecting devices, users, and data on a broad scale. Embedded devices have been getting a lot more intelligent, even wirelessly connected, for years now. Pessimists say this IoT stuff isn’t anything new, just more marketing hype for the same things we’ve been doing.

Optimists are moving ahead on the IoT at warp speed. I was in a Twitter chat recently where someone stated the hardware doesn’t matter, just grab a smartphone chip and get going with IoT software. If you’re a fan of old cartoons like I am, you’ve likely seen more than one episode of Tom & Jerry where Tom runs around a corner at full speed, steps on the tines of a yard rake, and gets blasted right in the face as the handle flips up quickly. I’m afraid that is the experience awaiting some design teams as they try to take chips made for “industrial” (or worse yet, mobile) and use them, unmodified, for the “industrial IoT”.

Makers. Out to change the world with a few lines of IoT code on an inexpensive board. I must admit, it makes an interesting reality show and fun at trade events, it gets more people (especially STEMmers beginning at a young age) involved in IoT device design, and it is a great way to get consumer IoT stuff imagined and prototyped quickly. As the above chart indicates, and I’ve been seeing in my consulting practice, the actual money will be on the industrial IoT side for this next cycle.

Industrial IoT designs have two overriding concerns that must be addressed in hardware: security, and reliability. I sat with the NXP folks recently, and they are talking about the disaster-in-progress that is IoT security right now – and these guys are experts on financial and embedded chip security. They are starting to ask about security profiles, parameterizing devices (and their underlying chips) by their level of security-aware features and sorting them by increasing levels of robustness. More robust industrial IoT designs must have reprogrammable, tamper-proof security keys; some of the biggest breaches in the industry so far have been due to fixed keys that once hacked render IoT devices untrustworthy.

Reliability sounds easy, until one adds the high-temperature requirements of the industrial landscape. Flash memory, available in ample volumes for mobile and consumer IoT devices, becomes a much bigger problem in industrial settings. Data retention goes away rapidly at high temperature. Case in point: just a couple years ago, TI introduced a “harsh environment” 4MB flash part (yes, 4MB) with an SPI interface, designed for 1000 hours of data retention at 210 degrees C. That’s about 999.9 hours longer than a typical maker module would last in a high-temp application such as deep well drilling.

These two requirements are part of the reason I’ve been saying for some time we need chips specifically designed for the IoT. Fortunately, we have a piece of critical technology: one-time programmable (OTP) memory. Sidense’s 1T-Fuse OTP technology can be configured to emulate multi-time programmable memory, essentially using several banks of OTP with a controller, allowing security keys to be reprogrammed as needed over the lifetime of an industrial IoT device. 1T-Fuse is also extremely difficult to reverse engineer and offers outstanding endurance, and high-temp qualification at 175 degrees C is underway, making it a viable and more cost effective alternative to flash in industrial IoT and automotive applications.

Sidense has more on their IoT strategy on their website. By putting smaller amounts of highly secure and extremely reliable OTP memory exactly where it is needed on an industrial IoT chip, designs can be differentiated and enhanced at low power and low cost.

What do you think of the IHS chart, and their prediction that industrial is the near-term semiconductor growth opportunity? What chip-level security and reliability problems have you seen crop up in industrial IoT applications? Have you seen OTP memory used to solve them?

Also see: Designing Low-Power IoT Systems