CMOS and the future of integrated radars
Dr. Gordon Moore ends his seminal 1965 paper (the one with the observation that Carver Mead would later term "Moore's law") with a pretty interesting prediction:
"The successful realization of such items as phased-array antennas, for example, using a multiplicity of integrated microwave power sources, could completely revolutionize radar."
Nearly 50 years later, ICs have been prevalent in radar for quite some time, although we've yet to see the level of integration enjoyed in e.g. radio transceivers. Though technically similar (the first two letters of the acronym “radar” stands for “radio”), true radar SoCs are still almost unheard of. The difference lies not so much in the technological limitations as in the respective applications and markets. These days most people carry several radios with them at all times (even though they call it a cell phone), we have several in our homes, and the MacBook Air I'm writing this on has at least three. Radar, on the other hand, is pretty far from occupying that same privileged place in people's minds. For most people, "radar" means one of four or five things: something to do with air travel, something the police has, that thing on your boat and some dangerous electronic warfare stuff. Also, cars have radars these days, although that’s rarely part of the sales pitch.
One might say that commercial radar simply hasn't found its raison d'être yet. None of the above applications contribute with volumes large enough to leverage the ecosystem of semiconductor mass production (with the exception of automotive, although with slow adoption rates), where radio has been thriving for years.
At EUMW earlier this month I had the pleasure of attending a keynote held by Dr. Ron Reedy, SOI proponent and founder of fabless-gone IDM-gone fabless Peregrine Semiconductor. Actually, his keynote followed a rather heated panel debate on the future of compound semiconductors, featuring some of the biggest players in the industry (Win, UMS, OMMIC, etc.), and Reedy, having no stakes in III-Vs, unflinchingly prophesized the victory of CMOS over all competing technologies using the most compelling argument of all: economics. The argument is a well-known one, you can't compete with economies of scale. Whatever market silicon chooses to enter, it will win. Reedy's outlook is as simple as it is chilling (for the III-V providers at least), wide bandgap technologies like GaN (on Si and SiC) will come to dominate the high-power market (except for really high power/high frequency, where TWTs will continue to dominate for some time yet), CMOS, bulk and SOI, will take everything else.
Comparing silicon to compounds might seem unfair: III-Vs will give you higher speed, lower noise, better linearity, higher power density, higher quality passives, wider temperature operating regions and less temperature dependency. Put that way, bulk CMOS is really a pretty lousy transistor, except for one redeeming quality: it is cheap. To be fair CMOS also has higher hole mobility and the availability of a native oxide, but none of that would be relevant if it wasn't for the low cost.
Circuit designers tend to find new and creative ways to solve problems, working around the limitations of CMOS. Engineers will continue to solve design challenges, utilizing device quantity in lieu of quality. This is what Dr. Moore once referred to as “circuit and device cleverness”.
We are already starting to see the first practical commercial radar applications, and I believe the potential for growth in the upcoming years is tremendous. We will see a huge range of low cost/power/size/range commercial radar sensors addressing new and existing sensing markets, complementing and to some extent replacing existing non-EM sensing technologies, and emerging as a key enabling sensor technology for the true Internet of Things. In 10 years, ubiquitous radar sensors are as big a part of our lives as low-power radios are today.
Not a bad time to be in CMOS radars.