Page 16 - EE Times Europe Magazine – June 2024

16 EE|Times EUROPE

AUTONOMOUS VEHICLES | SENSORS
Avalanche Photodiode Innovations

Target LiDAR Cost Challenges

By Robert Huntley

ithout sensors, an autono- “At 905 nm, you can use cheap
mous vehicle cannot perceive gallium arsenide lasers and low-
its surrounding environment cost silicon detectors.”
Wand potential hazards. The However, he cautioned, “At that
same goes for the plethora of advanced wavelength, you are quite close to
driver-assistance systems (ADAS), which are the wavelengths the eye absorbs
increasingly becoming the norm in today’s since your cornea can still focus
vehicles. Some ADAS functions, such as blind- light onto your retina, so you have
spot detection and adaptive cruise control, to use low-power lasers. But in so
are required to detect over different distances, doing, you don’t have the photon
leading automotive engineers to select suit- budget required to see out the
able sensor technology, such as ultrasonic, required 250 meters over a rea-
radar and LiDAR. Increasingly, through sensor sonable field of view using a single
fusion techniques, data from multiple detec- channel. You’ll see companies
tion methods are “fused” together to create using 256 lines of laser detectors
a more accurate, multi-dimensional image of instead to get around that. Yes,
the environment surrounding the vehicle. you can use low-cost components,
LiDAR is a popular method of detecting but you end up using a horrendous
objects, typically up to 200 meters away, by number of them to make up. So
emitting short pulses of infrared light from you have sacrificed cheap compo-
a laser and detecting reflections. The time of Avalanche photodiodes are operated with a reverse nents for system complexity.”
flight yields the distance between the vehicle voltage to place their operation just below breakdown White said that Phlux is focused
and the object. Initially, a LiDAR subsys- to achieve the maximum gain. (Source: Phlux) on the 1,550-nm wavelength with
tem was expensive, particularly those using its high-sensitivity Aura InGaAs
spinning mechanical and optical methods to photodiodes range. “At this wave-
control the field of view (FOV). However, with told EE Times Europe. “My co-founder and length, you can use large amounts of laser
its growing popularity and the development professor [and I] had an early breakthrough power without damaging the human eye. The
of MEMs-based FOV control, costs have been when we discovered how a particular alloy cornea doesn’t focus on the light, but also, the
reduced significantly. composition of antimony exhibited some goo in your eye absorbs that wavelength.”
As with any sensing application, the sen- unique electron properties of very low-noise
sor’s specifications dictate the performance behavior you could exploit for a detector. IMPROVED APD PERFORMANCE
characteristics of the complete subsystem. “There is a growing adoption of LiDAR for AT 1,550 NM
The receiving sensor in a LiDAR camera is automotive applications,” he said. “How- While using 1,550 nm offers an eye-safe
typically a photodiode, converting the reflect- ever, the cost of the laser is still a limiting wavelength, many APDs are not particularly
ing light into electrical energy. Avalanche challenge for mass LiDAR rollout. With our sensitive at that shortwave infrared wave-
photodiodes (APDs) are highly sensitive, and highly sensitive APD, subsystem designers length. “With our Aura Noiseless APDs, we
their operation uses an avalanche process can use a lower-cost laser to achieve the same can deliver a much more competitive position
that yields a multiplier factor, creating more performance figures as an expensive laser against the cheaper, silicon-based APDs,
electrical output from a given number of with a traditional InGaAs APD. That reduces rebalancing the design tradeoffs using low-
reflected photons. the overall bill of materials down to a point cost lasers,” White said. “The detectors in use
where it can be mass-produced.” today haven’t evolved much in the last
COST STILL A MAJOR CONSIDERATION EE Times Europe asked White if there are 25 years. The new applications, such as
To date, the fabrication of APDs has focused other limiting factors besides cost. “Reliability LiDAR, bring much higher performance
on using an indium gallium arsenide (InGaAs) is a massive concern for automotive, since the demands. What differentiates us from others
process operating with a 905-nm laser. cost of a recall, even if it’s only due to the fail- making detectors for 1,550 nm is that we add
However, Phlux Technology (Sheffield, U.K.) ure of a few units, can create really bad press,” an antimony alloy to the compound semicon-
claimed it has developed an InGaAs technol- he said. “So it is essential to have extremely ductor during manufacturing. With a
ogy capable of operating at 1,550 nm that is low failure rates, both naturally and ran- 12× better LiDAR image resolution for a given
12× more sensitive than traditional APDs and domly, throughout the product’s lifetime, and laser power and an APD gain of up to 120,
overcomes many of the challenges associated as we understand, it is one of the issues with we can discern the smallest signals above
with 905-nm lasers. EE Times Europe spoke today’s LiDAR systems.” the noise floor of a connected transimped-
with CEO and co-founder Ben White to learn ance amplifier and offer size, weight and
why the 1,550-nm wavelength offers better 905 NM VS. 1,550 NM bill-of-material cost reductions.” ■
performance for automotive applications. White told EE Times Europe that the LiDAR
“Phlux Technology started as a research industry currently appears divided into two Robert Huntley is a contributing writer for
project at the University of Sheffield,” White camps regarding the best wavelength to use. EE Times Europe.

JUNE 2024 | www.eetimes.eu