ON Semi Unveils SiPM Array for Automotive LiDAR Applications

Article By : Maurizio Di Paolo Emilio

ON Semiconductor has unveiled a new RDM Series silicon photomultiplier (SiPM) array that extends the sensor capabilities to its broad portfolio of smart sensing solutions.

At embedded world 2021, ON Semiconductor has unveiled a new RDM Series silicon photomultiplier (SiPM) array that extends the sensor capabilities to its broad portfolio of smart sensing solutions. The RDM-0112A20-QFN array is automotive qualified to meet the growing demand for automotive LiDAR applications.

During the media briefing, Joseph Notaro, vice president Worldwide Automotive Strategy and Business Development at ON Semiconductor, reminded that LiDAR (Light Detection and Ranging) is a sensing mode that can provide a high-resolution depth map and create a 3D mapping of the surrounding environment.

Standard imaging techniques can transform a 3D scene into a 2D image. A depth map eliminates any ambiguity in an image, helping to prioritize by knowing exactly what each object is. More importantly, within the scene, it helps determine which object is closer, which is farther away.

“When we’re designing a LiDAR system, we have three basic choices to make: Depth-sensing technique, wavelength and sensor technology,” said Notaro.

LiDAR can be used to improve safety and driver assistance systems (ADAS), aiding functions such as lane keeping and traffic jam assistance. The SiPM photomultiplier is a low-light sensor that is used in security and high-volume nuclear medicine applications but is now finding its way into automotive.

LiDAR techniques

Notaro pointed out three depth detection methodologies, each with its pros and cons (Figure 1). The dToF (direct Time-of-Flight) method has the ability to be used for short long intervals with high accuracy and can detect multiple echoes. It may indeed be able to measure various targets. The indirect methodology (iToF), on the other hand, is more suitable for short ranges but is much more complex in terms of computing power. FMCW LiDAR combines optical communication hardware and radar signal processing methods to provide accurate information on object distance and velocity (figure 1).

Depth-Sensing Methods - LiDAR
Figure 1: Depth-sensing methods (Source: ON Semi)

The applications of LiDAR systems are growing in all sectors, including robotics and industrial proximity sensing, where accuracy levels in the order of millimeters must be guaranteed.

They typically use dToF technology, which involves measuring the time it takes for a light pulse, usually with a wavelength in the near-infrared (NIR) region, to reach an object and return to its source. “dToF used with 905nm light sources is proven to achieve long range LiDAR, and is enabling mass adoption due to lower system cost and existing ecosystem,” said Notaro.

Cost is still its main driver, and the ability to use silicon-based CMOS processes helps scale up large volumes. In addition, near-infrared is clearly the wavelength of choice for autonomous driving applications, and it requires more laser power, which can lead to thermal issues. In addition, water absorption effects need to be considered, thus maintaining high sensing efficiency and vehicle safety during bad weather conditions.

SiPM technology

SiPM technology has already gained market share in recent years due to its large-scale depth potential. In terms of signal-to-noise ratio, they provide a high value for long-distance measurement in bright sunlight. Additional advantages, including lower bias voltage and sensitivity to temperature changes, make them strongly competitive with conventional avalanche photodiodes (APDs). SiPMs are manufactured in a high-volume CMOS process, allowing for the low detector cost and thus enabling large-scale LiDAR solutions.

The RDM-0112A20-QFN array is a monolithic 1×12 array of SiPM pixels based on ON Semi’s RDM process. As Notaro pointed out, the device enables high near-infrared (NIR) light sensitivity to achieve 18.5% photon detection efficiency (PDE) at 905 nanometers (nm). PDE determines the efficiency with which incident photons are detected (Figure 2).

Array Schematic Showing Pixel Connections
Figure 2: Array schematic showing pixel connections (Source: ON Semi)

The high internal gain of the SiPM allows sensitivity down to the single-photon level, a feature that in combination with the high PDE, enables the detection of weaker return signals. This translates into the ability to span greater distances even with poorly reflective targets.

Geiger mode operation and microcell structure
Figure 3: Geiger mode operation and microcell structure leads to high internal gain (>1E6) and sensitivity to single photons. This means that low threshold can be set to detect faintest signals, and correlation techniques can be used to overcome solar noise and to increase range further. (Source: ON Semiconductor)

“The newly released RDM-0112A20 array in a QFN package is an automotive-qualified sensor that supports the low cost, infrared and direct time of flight LIDAR systems, enabling ranges up to 250 meters. It is a one-by-12 pixel long monolithic optical array,” said Notaro.

He added, “The device operates in Geiger mode with a very high internal gain. This leads to very single-photon sensitivity. In Figure 3, for example, we already have a signal with a single photon. So that means that with the same laser power, it can really detect weaker signals allowing you to extend the range of the system.”

A photodiode operating in the Geiger mode is an avalanche photodiode biased at a reverse voltage higher than the breakdown voltage. Thus, there is a strong electric field in the emptied zone which transfers to the single carrier a very large amount of kinetic energy that is sufficient to trigger the ionization process by impact and the avalanche multiplication of the carriers.

Subscribe to Newsletter

Leave a comment