In this Big Compute Podcast episode, Gabriel Broner hosts Mike Hollenbeck, founder and CTO at Optisys. Optisys uses HPC in the cloud and 3D printing to design customized antennas which are smaller, lighter and higher performing than traditional antennas.
Overview and Key Comments
Optisys designs customized antennas for each customer. They simulate them using HPC in the cloud, and 3D print them, making antennas that are one piece instead of being built from hundreds of parts, which makes them light, small enough to fit on the palm of your hand, and very high performance. Their story is interesting from an antenna design perspective and also as a startup capturing technology shifts to disrupt an established category.
Traditional Antenna Design
“Antennas have historically been built using the manufacturing tools that were available at the time. These were subtractive manufacturing tools where you start from a giant block and remove material to get the structure you want. Since you need to be able to reach every part from the outside, you break the problem into individual parts you can manufacture. You need to design hundreds of smaller parts, hold them together to achieve a particular antenna function, and it’s expensive to assemble.”
Changes in the Industry
“At the time we started Optisys, there was a confluence of three key technologies that had evolved significantly: Big Compute or High Performance Computing in the cloud, simulation capabilities, and Additive Manufacturing. In the past you had to write your own interface to use an HPC system at a company. Now there are companies like Rescale that provide a wonderful interface. Ansys software enables us to design really complex structures. Metal additive manufacturing has significantly changed for electromagnetic structures in the last five years. Additive manufacturing is the same as 3D Printing.”
The Beginning at Optisys
“The way most people were approaching additive manufacturing was to build what they had already designed. We instead looked at what we can build with additive manufacturing and design structures that had never been created before.”
“The biggest challenge was convincing people that the little antenna structures we built worked and were made out of metal. We took a really complex antenna and reduced it from 100 parts to a single part, from 15 inches to 2, and from 15 pounds to two ounces. We dropped that on the palm of people’s hands. Some people did not believe, others believed and were in shock. We had to prove this really can be done. It was an education process for the first year and a half. I view some of our earlier pieces as a work of art.”
“We work on antennas that are used by government, defense and commercial, for airplanes, ships, and remote terminals carried by a soldier.”
Working with customers
“We had to change the way companies work with antenna providers. People tend to ask for a part. We want to interact at the specification level as a partner to build the antenna that addresses their needs. We use ‘Lego building blocks’ we developed that we put together and then simulate the customer’s antenna. We bring to market ‘mass customization’ of antennas and we approach the cost of traditional mass production.”
“We are seeing excellent results. We are able to produce structures that are orders of magnitude smaller and lighter. With traditional processes, adding complexity adds cost, weight and lead time. For us adding complexity reduces cost, because we simulate, we print one piece, so we are able to produce higher performing antennas at lower cost.”
HPC in the Cloud
“We have a core belief not to reinvent the wheel. Our talent is additive manufacturing and RF design. We needed to use HPC as a simulation intensive company. If you want to buy hardware, you have to anticipate the largest problem you need to simulate, but you will use it 10% of the year. What HPC in the cloud allows us to do is to dynamically scale our resources to meet the demands of the day, focus on design, have cheap systems in house, and use Rescale for any heavy lifting or to free up our internal system. It’s a spectacular solution.”
“Often people learn how to do something and they keep doing it that way, even if it does not make sense. We should always think ‘Is there a new tool at my disposal that enables me to work differently?’”
“I am excited for what the future holds for HPC in the cloud. People used to need a gigantic system. We dynamically scale our hardware capability every day. I’m excited for the possibilities HPC can bring to other areas.”
Mike Hollenbeck is CTO and Founder at Optisys. He has 11 years of experience designing and analyzing antenna and RF components for commercial and military communication systems. At Optisys, he designs cutting edge RF and antenna components using additive manufacturing and provides technical direction for the company. Mr. Hollenbeck has led numerous projects, including design and simulation of complex RF components such as waveguide, antennas, arrays, and filters to critical success. He has worked on iterative designs to produce RF products in the 1-100 GHz (3-300 mm) ranges.
Gabriel Broner is VP & GM of HPC at Rescale. Prior to joining Rescale in July 2017, Gabriel spent 25 years in the industry as OS architect at Cray, GM at Microsoft, head of innovation at Ericsson, and VP & GM of HPC at SGI/HPE.