The Magic of Voyant

I never intended to get a degree in electrical engineering. I did not even know what that was when I applied to college. I had intended to get a degree in mechanical engineering like my father. The application process at one particular college had a waterfall model on department selection. I picked the department at the top of the waterfall, so I could have more choices, with the intent of changing majors later.

I struggled in the introductory courses. My advisor suggested I change majors. I had a complete lack of natural talent for the material. Despite all this, I fell in love with electrical engineering (EE). EE is as close to magic as we can find in the modern world. EEs use arcane devices that almost no one understands (not even other EEs) , devices that are often invisible to the naked eye. EEs manipulate invisible forces for various effects, and these forces are not limited by distance or time.

A mechanical engineer can build a robot. An aerospace engineer can build a rocket. Physical, clunky, noisy things that you can see and touch and smell.

It takes the magic of electrical engineering to make that robot walk, sing or dance, to give it the illusion of having a soul. It takes EE magic to let you see what that robot sees, even if that robot is on another planet.

Even though I have not been a practicing electrical engineer for decades I still dabble and read. I thought I knew what magic was. Then I started working with Voyant.

I realized that my level of magic was, in comparison to what happens at Voyant, the kind that you see performed at fourth-grade birthday parties. Not the serious Gandalf vs Balrog kind. Not the kind that changes the world.

Steve and Chris, the founders of Voyant, are arcanists at a completely different level. They not only conceived of a complete imaging LiDAR system on a single semiconductor chip, but have the skills to make it a reality.

Engineering at Voyant’s level is truly magic. Optical components used in current LiDARs are items you buy out of catalogs, little lenses or wedges or mirrors measured in centimeters that get bolted onto a table or inside a shoebox.

Optical components in Voyant’s devices are a few microns in size or smaller.  For comparison, an average human hair is 75 microns in diameter. These lenses, splitters, and mixers are connected by the tens of thousands via microscopic optical waveguides into our chip-scale LiDAR solutions.

Transistors in the processor chips in your computer or phone are much smaller, 1/10 or even 1/100 the size of Voyant’s smallest optical components.  Electrons traveling in metal, for the most part, don’t care about the size or shape of their controlling blocks of doped silicon. You are probably familiar with the square and rectangular city-scapes found on semiconductor chips.   Squares and lines work for electrons, and are much easier to design with.

Light waves are more finicky.  Voyant’s designs require careful, artistic, organic shapes.  Lines and rectangles simply won’t do for photons.

Here are some snapshots of parts of our chip designs. The wave guides are the magenta lines, surrounded by white.

Voyant’s chips are filled with these gorgeous shapes.

The image below is the output end of a delay line used for signal mixing.  The “L” of magenta waveguides is the corner of a giant spiral.

The surface area of this entire structure is less than 9 square millimeters, but the waveguide packed inside is 0.6 meters long.  Two feet of optical wiring packed inside a surface a tiny fraction of your pinky fingernail!  Light travels through this particular chip-based optical wiring slower than in air, so this millimeter-scale  spiral is the equivalent of optical elements on a lab bench spaced 5 feet apart.
The thickness of this optical wiring is 3 microns, which by Voyant’s standards is huge, but still only half the width of a human blood cell.

I could go on describing these magical structures for a long time and still not get to the topic of how Voyant uses these structures to create a LiDAR system.  More magic there of a signal processing variety. Signal processing for active sensing is a topic dear to my heart. Perhaps in another blog post.

I am extremely humbled to work with people who have the skills and audacity to dream these things up, and more importantly, produce them at commercial scale.