If you want your car to automatically brake for collisions, beep when you’re about to back into something, turn on its wipers when it rains and deploy its airbags during a collision, you’re going to need a lot of sensors. And when those sensors start being used for phone accelerometers, automated driving systems and aerial inertial navigation systems, you’re going to stretch your capacity pretty thin.
That’s why Bosch, the world’s largest automotive supplier, brought me to see its under-construction wafer fabrication facility in Dresden, Germany. Despite currently producing 4 million sensors per day, Bosch is expecting to hit capacity as automakers require more radars, more cameras, more ultrasonic sensors and more chips in general. In 2018, for instance, the average car sold had nine Bosch chips on board already.
[Full Disclosure: Bosch flew me to Germany for the Frankfurt Auto Show, after which they took me to see existing production in Reutlingen and construction on the Dresden facility. They covered my hotels, travel and most of my meals.]
But it’s not just the automotive sensor business that’s booming. It turns out that, when you develop your own automotive-grade sensors, there are applications far outside of the world of cars.
Take, for instance, airbag sensors. Early mass-type mechanical sensors used magnets, gold-plated sensing plates and large sealed housings. There were multiple failure points and the technology could only measure if a crash exceeded a certain deployment threshold, but not give a precise acceleration figure.
So Bosch developed a micromechanical acceleration sensor. Using “The Bosch Process,” a procedure now used worldwide to etch complex structures onto silicon wafers, Bosch was able to create cheap and accurate Micro-Electro-Mechanical Systems. These MEMS were 50 times smaller than conventional sensors and used 100 times less power.
These early MEMS consisted of a microscopic movable mass suspended between fixed conductive plates. When the mass is accelerated, it’s proximity to each plate changes, which alters the capacitance between the plates and the mass. That can be measured to precisely determine acceleration in two dimensions. Here’s a good visualization:
Bosch soon developed a three-dimensional MEMS sensor, which was originally used for rollover sensing in cars. But accurate three-dimensional acceleration measurements are useful in a lot of ways. For instance, turn your phone sideways. The screen flips.
More than likely, it’s doing that based on information from a Bosch sensor. People at Bosch generally say that their MEMS are in “over 50 percent” of phones, but the figure on the company’s site is more precise and impressive: three out of four phones contain a Bosch accelerometer. And even if Bosch doesn’t make the accelerometer, it’s made with a process Bosch invented.
The process, deep-reactive ion etching, changed the MEMS industry forever. Essentially, it allows for fine etching of deep, precise structures into hard materials like silicon.
“We all know we can use a knife to cut cheese, but if you’re trying to cut rock it’s a lot harder,” Chang Liu, MEMS expert and founder of sensor tech company Sensic, said. “Bosch invented this process to cut rock like cheese, so that’s a game-changer.”
While accelerometers and electromechanical sensors are possible without deep-reactive ion etching, it allows complex MEMS to be produced at a scale and cost that enabled large-scale proliferation in the smartphone market. Though it was developed to produce cheaper airbag sensors, it enabled everything from smartphone augmented reality apps to the Nintendo Wii.
The scale of automotive sensor production led to Bosch’s involvement in the smartphone industry, but it’s the high safety and durability standards of the auto industry that’s got Bosch eyeing a new market.
A 15-person team is currently developing Bosch’s first product aimed at a controversial market: air taxis. While some write off air taxis—or, as you may know them, flying cars—a lot of people are investing in what they see as a huge long-term market opportunity.
“All the market surveys kind of expect [air taxis] to be a pretty large market, but the issue is there are a lot of hurdles to overcome,” Graham Warwick, who manages Aviation Week’s technology coverage, said.
Those studies predict the potential market for air taxis to be phenomenal if they can be made safe, quiet, affordable and legal under aviation regulations. One study by NASA, for instance, predicts a market that generates $500 billion in revenue in a “best-case unconstrained scenario.” The U.S. airline business, for reference, is a $150 billion per year industry. It’s easy to see, then, why Bosch wants to get in early.
The company’s first venture into the aviation sector is a small sensor box. One of the many problems with making a working, affordable, compact air taxi is that flying machines that carry human beings typically contain Inertial Guidance Systems. Doubly important in any sort of autonomous application, INS allows an aircraft to know its position and heading even when GPS and other external guidance systems are offline.
Commercial aircraft contain INS, but there are three big problems: they’re huge, expensive and consume a lot of power. A typical unit is a box a few feet in every direction and costs around $80,000. If you’re trying to make a small, electric air taxi, a huge INS sucking power isn’t something you can really work with.
To solve this problem, the team at Bosch has come up with a prototype sensor box using MEMS and other sensors primarily from the automotive business unit. The sensor box meets or exceeds current aviation standards for accuracy, redundancy, drift and durability while fitting in your hand and weighing about 300 grams. The target price for the sensor box is $10,000, assuming there’s enough demand for air taxi production.
Long term, the company sees its expertise in powertrain control technology, MEMS, automated operation and safety systems as huge advantages to supply different technology to the burgeoning sector. And while it’s not common for automotive-grade parts to make it to the aviation sector, Warwick says the rapid progress in automotive sensor tech and the democratization of air travel may meet at a perfect juncture.
“We have to recognize that autonomous cars are driving automotive [parts] to similar levels of safety that aerospace is already at,” Warwick said. In fact, some aerospace processes are already making their way into automotive software development.
“Automotive manufacturers are using exactly the same software certification and software writing rules that were developed for the aerospace industry,” he said. If automotive suppliers and companies can prove to regulators that these sensors, engine management systems and navigation systems meet aviation standards, they’ll have access to that market.
To be clear, though, the company has long had a “bet on all the horses” strategy. When you’re that big, you can’t afford to miss the next big thing. It’s rare and notable, however, for Bosch to talk publicly about a new market it’s exploring this early on.
So, to recap, nine Bosch semiconductors exist in the average car, one in most smartphones, many in dishwashers and other consumer products around the globe and some are even supposed to power flying cars. It’s not hard to see why 4 million per day isn’t enough.
Already—since most supplier deals require two production sources—that 4 million figure only accounts for half of the Bosch-branded sensors produced every day. The rest are built under license by partners, but it’s still not enough for the growing business sector.
So Bosch dropped 1 billion Euros—its largest single investment ever—on a new wafer fabrication facility in Dresden. The Dresden facility will double Bosch’s sensor output, but current predictions suggest that still won’t be enough. As luck would have it, there’s a parking lot directly next to the Dresden fab facility that’s perfectly sized to mirror the fabrication building and double the output of the entire compound. Fancy that.
And since scale is everything, Dresden is a 300mm wafer fab facility, up from the 150mm and 200 mm wafer fabrication centers in Reutlingen. Essentially, the larger the wafer of silicon, the more semiconductors you can fit on one wafer. Since entire wafers are processed at once, bigger wafers allow you to produce more semiconductors and shrink down the structures on the semiconductors.
That means Dresden can produce more sensors at a lower price to power the onslaught of automated driving features and technological nonsense on modern cars, freeing up Bosch’s other facilities to provide the accelerometers that power your phone’s crappy Augmented Reality games. All of this, from a patent originally intended to make airbag sensors smaller.