What’s the story behind Milliwave: the genesis of the company and what are you aiming to do?
We started working on consumer mmWave designs in the early 2000s. Back then, we focused on wireless video at SiBEAM, and other markets emerged, like 11ad, radar, backhaul, gesture, etc.
While starting Milliwave in 2016, we wanted to continue to pursue and help the mass market mmWave realization, but we did not want to focus on a single application. At that time, many new markets for mmWave were emerging, such as 5G (NR), 77 GHz automotive radar, as well as 60 GHz for wireless, VR and contactless connection. This drove most of our business originally.
Your tag line is “the most trusted in mmWave silicon.” First, describe the markets you are targeting and the range of silicon blocks you can develop for those applications. Secondly, what makes Milliwave “most trusted”?
As consumer mmWave pioneers at SiBEAM, we had to clear the path for many challenges that this type of design brought. Tools were hard to find, test setups non-existent, process node PDKs had to be developed from scratch. On the productization front, regulatory test setups, enclosure design impact on antenna radiation and production reliability were among the many questions we had to answer.
We did address a lot of those questions and challenges and, ultimately, went to mass production and market with multiple products like TVs, laptops and cell phones integrating mmWave CMOS devices. At that time, the commercial success did not realize, but the amount of knowledge and expertise we accumulated in that decade is found to be precious from our clients.
Today, Milliwave is centered around our core competency of mmWave CMOS design, test and productization. This includes the design of circuit blocks such as PAs, LNAs, mixers and power detectors. In addition to circuit design, we also offer services in device modeling, test and measurement, packaging and antennas and regulatory strategy.
When you’re approaching a new design, how do you choose which RF silicon process to use (e.g., RF CMOS, SiGe, SOI)?
In many instances, our clients have already selected a process based on a strategic relationship or early evaluation that they had done. But on several occasions, we were involved in alternate process selection or migration investigations. At the end, it is often a question of compromise: the best performance from the pure radio standpoint may not be the one viable from a cost and availability point of view, especially when it comes to consumer applications.
Sometimes it even requires breaking up a single design across multiple processes to get the best solutions. Those who claim there is a single answer to that problem most likely don’t understand it.
Obviously, you’re bullish about silicon. Do you concede any circuit functions to GaAs or GaN?
When comparing different technologies, there is always a trade-off among several requirements: performance, cost, form factor and thermal dissipation.
We believe that silicon has great potential to address many mmWave markets. It clearly possesses the benefit of high integration, scalability and low cost, and the raw performance of silicon devices at mmWave frequencies has improved a lot over the past five years.
However, it will never be the best in terms of PA linearity and efficiency. Therefore, front-end modules in other technologies continue to exist. Silicon will not completely replace GaAs and GaN.
Over the past few years, millimeter wave applications have captured the industry’s imagination and considerable R&D investment. From your perspective, how successful is the commercialization of silicon semiconductor technology for mmWave applications? How about the commercialization of the end applications?
Certainly, silicon is a great vehicle for mass market application because of its cost and production volume. In our opinion, commercial success (or lack thereof) of silicon for mmWave applications is not just a matter of technology capability — also the integration of the strengths and weaknesses of mmWave properties into a compelling use case.
We have been searching for that killer app for over a decade. Now we are seeing a lot of investment being made in mmWave applications like 5G (NR) and mmWave radar. Those applications are pushing the convergence of technology capabilities and market push, which are making mass adoption possible.
Although it’s not an IC, your MilliBox product complements your IC development. What is the MilliBox, and what prompted you to develop it?
Milliwave was started as a small expert shop for mmWave consulting, and MilliBox was just an afterthought. While consulting, we saw a gap in the market for over-the-air antenna radiation pattern measurement setups. You had either a multi-seven figure all built-in solution, or you would go to a construction material big store and build your solution from scratch. There was nothing modular, nothing flexible and nothing affordable.
Over the years, we built many of our own test chambers tailored to our mmWave needs. We saw our consulting clients having the same issue, and they asked us to bridge this gap. We came back with a concept that encompasses years of trial and error on this subject. We started showing it around and saw great enthusiasm in the market right away. Several times, customers ended up buying our demo unit on the spot.
We knew we had designed something which checked the mark on many must-have features. Now MilliBox is its own product line within Milliwave, adjacent to our consulting services.
Describe the capabilities of the MilliBox system and how it’s used.
MilliBox is a family of modular, lab bench sized anechoic chambers and positioners for mmWave antenna radiation pattern measurements.
The chambers come in three standard sizes going from 80 cm to 2 m far-field capabilities. The absorber material we use is specific to mmWave and covers a range from 18 to 95 GHz with about 50 dB absorption. The frame and body are made of PVC, PLA and birch plywood. Unlike others, we try to avoid using metal as much as we can in our designs, because our experience shows that limiting stray reflection is much more important than interference shielding in mmWave measurements.
Our positioners, or gimbals, are capable of 360 degrees in both H and V directions down to sub-1 degree step size. This gives a full spherical plot of over 64,000 capture points. We have several sizes of gimbals to accommodate different sizes of devices under test, and the gimbals are controlled over USB with Python or MATLAB code which we provide in source. The positioner also comes with a crosshair laser pointer to achieve perfect alignment.
Tell us about your backgrounds and what led you to start Milliwave.
Dr. Doan started working on mmWave CMOS circuits and device modeling at UC Berkeley in 2002, for his PhD thesis. In 2005, he founded SiBEAM and was the VP of Engineering.
Jeanmarc Laurent has over 20 years of productization experience in the consumer electronics space and joined SiBEAM in 2008.
SiBEAM was the first company to commercialize the use of CMOS for 60 GHz consumer applications, and after more than a decade together developing this groundbreaking technology, we started Milliwave Silicon Solutions in 2016.
Through consulting, design contracts and now the introduction of the MilliBox low-cost mmWave anechoic chamber, we hope to see silicon achieve its full potential in enabling mmWave mass market adoption.