Research Achievements and Highlights

LiDAR systems serve as a critical technology in advanced driver assistance systems (ADAS), autonomous vehicles, robotics mobility, and industrial automation, with their performance largely dependent on the Signal-to-Noise Ratio (SNR). This research presents an in-depth analysis and comparison of different photodetectors (SiPM, SPAD arrays, and APD sensors) under varying temperature conditions. Through SNR analysis calculations, numerical simulations, and experimental validation, we explore system design optimization to enhance LiDAR detection performance, particularly in long-range applications. 

The study conducted detailed analysis on two typical Time-of-Flight (ToF) LiDAR systems – short-range (30 meters) and long-range (250 meters) – with particular focus on SNR performance variations among different photodetectors at operating temperatures of +25°C and 105°C.

Research Team

This research was conducted by a team from onsemi’s Advanced Sensing Division, with Andrii Nagai as the principal author. The paper compares the performance of three photodetectors – SPAD, SiPM, and APD – in Time-of-Flight (ToF) LiDAR applications, examining key performance factors including microcell count, dark count rate, dead time, optical system design, and environmental temperature effects.

Research Background and Key Findings

Impact of Photodetector Selection on Optimal Optical Aperture in LiDAR Systems

In LiDAR systems, the choice of photodetector (SPAD, SiPM, or APD) significantly influences the optimal optical aperture. This relationship stems from variations in light sensitivity, signal-to-noise ratio, and ambient light immunity among different detectors.

Figure 2 clearly demonstrates these effects:

• SPAD Arrays: LiDAR systems based on SPAD achieve maximum SNR with small-aperture lenses. This is attributed to SPAD’s exceptional sensitivity, enabling effective echo signal detection even at low photon counts. Thus, smaller apertures sufficiently collect enough photons for high SNR while effectively suppressing ambient light interference.
• APD: Conversely, APD-based systems require large-aperture lenses (Dlens > 25 mm) to achieve high SNR. This requirement stems from APD’s lower internal gain and relative light sensitivity, necessitating larger apertures to collect sufficient photons to overcome electronic noise.
• SiPM: SiPM-based systems exhibit characteristics intermediate between SPAD and APD. With sensitivity levels between SPAD and APD, their optimal optical aperture similarly falls between these two extremes.

Further explanation of how different detectors influence optimal optical aperture:

• Sensitivity: SPAD’s superior sensitivity enables the use of small-aperture lenses, while APD’s lower sensitivity necessitates larger apertures.
• Ambient Light Immunity: Small-aperture lenses more effectively suppress ambient light interference, particularly crucial for highly sensitive SPAD devices. While APD exhibits lower sensitivity, its excellent linearity and high photon sensitivity provide robust ambient light immunity, allowing for larger aperture usage.

SNR Performance of SPAD Arrays vs SiPM in Low-Photon Environments

In low-photon environments, SPAD arrays demonstrate superior SNR performance compared to SiPM. Here’s a detailed analysis:

• Lower Dark Count Rate: SPAD arrays typically exhibit lower dark count rates than SiPM. Dark counts, which are noise signals generated without photon incidence, degrade SNR. This effect is particularly significant in low-photon environments. Consequently, the lower DCR of SPAD arrays contributes to superior SNR performance in low-light conditions.
• Reduced Microcell Count: SPAD arrays feature fewer microcells compared to SiPM. While this reduction in microcell count limits dynamic range, in low-photon environments, fewer microcells result in lower aggregate dark counts, thereby enhancing SNR.

Figure 1 validation: Figure 1 demonstrates that SPAD arrays achieve notably higher SNR compared to SiPM under identical echo laser power (PS) and background power (PB) conditions. This confirms superior SNR performance of SPAD arrays in low-photon environments. The graph illustrates the relationship between SNR and echo laser power (PS) and background power (PB) at 25°C for SPAD (including 2×2 and 7×7 micropixels), SiPM, and APD.

The red solid lines and orange dashed lines represent SNR = 10 conditions at 25°C and 105°C respectively. Notably, SPAD devices demonstrate highest echo laser power sensitivity, surpassing SiPM by one order of magnitude and APD by two orders of magnitude, attributed to their smaller effective area and lower dark count rate. White and black lines indicate expected PS and PB values for short-range and long-range LiDAR systems, calculated based on Table 1 specifications.

Analysis of Figure 1 reveals minimal temperature impact on SPAD device SNR, while APD devices show maximum temperature sensitivity, consistent with the previously discussed temperature stability characteristics of SPAD and APD.

Maximum SNR Degradation in APD Under Temperature Variations

• Dark Current Effects: The study indicates that SiPM and SPAD dark count rates (DCR) double approximately every 8°C temperature increase. However, APD dark current (ID) exhibits significantly higher temperature sensitivity, increasing by a factor of 1.1 for each 1°C rise.
• SNR Formula Comparison:
  • For SPA and SiPM, electronic noise (Nelec.) in their SNR formula (Eq. 2) directly correlates with DCR.
  • For APD, dark current (ID) in SNR formula (Eq. 4) directly impacts the noise term in the denominator.

Figure 2 validation: Figure 2 illustrates the relationship between photodetector SNR and lens aperture. At identical lens apertures, APD exhibits the most significant SNR variation between 25°C and 105°C, substantially exceeding SPAD and SiPM variations. This demonstrates APD’s high temperature sensitivity. Due to APD’s temperature-sensitive dark current, its SNR experiences substantial degradation at elevated temperatures, exceeding degradation observed in SPAD and SiPM.

Solid and dashed lines represent SNR values at 25°C and 105°C respectively. For SPAD-based LiDAR systems, maximum SNR is achieved with small-aperture lenses, while APD-based systems require large-aperture lenses (Dlens > 25 mm) for optimal SNR. SiPM system SNR falls between SPAD and APD performance. Additionally, Figure 2 confirms APD systems’ maximum temperature sensitivity, further validating APD’s inferior temperature stability.

Experimental Methods and Validation

Validation Methodology

This study employs Signal-to-Noise Ratio (SNR) as the primary performance metric, validated through three distinct methods:

• Analytical Calculations: Derivation of SNR analytical expressions based on photodetector operating principles and system parameters.
• Monte Carlo Simulation: Development of numerical models for photon transport and detection processes to simulate signal and noise characteristics of different photodetectors.
• Experimental Measurements: SNR performance measurements of various photodetectors using the onsemi Gen1 LiDAR demonstration system under laboratory conditions.

Figure 3: Comparison of SiPM SNR simulation, calculation, and experimental results.

Figure 3 presents a comparison of SNR simulated values, calculated values, and experimental measurements for SiPM under varying overvoltage (∆V) and target reflectivity (η) conditions. The graph displays SNR values obtained through three different methods: analytical calculations (Eq. 2), Monte Carlo simulation, and experimental measurements. Experimental data was collected using the onsemi Gen1 LiDAR demonstration system, measuring targets with 12% and 100% reflectivity. The consistency observed among all three methods validates the accuracy of both the analytical calculations and Monte Carlo simulation approaches.

Research Results and Conclusions

Key Research Findings

• Sensitivity: SPAD devices demonstrate superior echo laser power sensitivity, exceeding SiPM by one order of magnitude and APD by two orders of magnitude, attributed to their smaller effective area, lower dark count rate (DCR), and rapid recovery time.
• Temperature Stability: SPAD devices exhibit superior SNR stability at elevated temperatures (105°C) due to lower dark count rates. APD devices show maximum temperature sensitivity due to significant dark current (ID) increase with temperature.
• Dynamic Range: SPAD arrays exhibit limited dynamic range, necessitating controlled background light power on channels. APD devices demonstrate superior ambient light immunity due to excellent linearity and high photon sensitivity.
• Optical System Optimization: Optimal performance of SiPM or SPAD requires optical system design that suppresses ambient background light, utilizing small-aperture lenses (Dlens) and restricted field of view (FoV).
• Microcell Count Impact: Reduced microcell count leads to lower dark count rates, enhancing SNR in low-photon conditions while decreasing ambient light immunity.

Device Performance Characterization

The study details key parameters of SPAD, SiPM, and APD, including:

• Photon Detection Efficiency (PDE): SPAD achieves 30% PDE. PDE serves as a crucial metric for photon-to-electron conversion efficiency, closely related to quantum efficiency (QE).
• Quantum Efficiency (QE): SiPM demonstrates 55% QE. QE quantifies photon-to-electron conversion efficiency, determining photodetector sensitivity.
• Dark Count Rate (DCR): SPAD and SiPM exhibit DCR values of 25 KHz/mm² and 150 KHz/mm² respectively. DCR represents noise signal generation rate without photon incidence, impacting SNR.
• Dead Time: SPAD and SiPM feature τdead values of 6 ns and 14 ns respectively. τdead represents recovery time required between successive photon detection events.
• Gain: SPAD and SiPM demonstrate gains of 1E5 and 10013 respectively, representing electron multiplication per incident photon.

Conclusion

This comprehensive analysis of SPAD, SiPM, and APD photodetectors in ToF LiDAR systems provides valuable insights for LiDAR system design and optimization.

Results indicate that SPAD devices excel in sensitivity, temperature stability, and SNR performance, despite limited dynamic range. SiPM offers moderate performance characteristics, while APD demonstrates superior linearity and ambient light interference immunity. Researchers can select optimal photodetectors based on specific application requirements, maximizing performance through optimized optical system design.

Enlitech’s SG-A is recommended for measuring all parameters mentioned above.

Original Publication Information

Published in 2024 International SPAD Sensor Workshop
Images and source material:
https://imagesensors.org/Past%20Workshops/2024%20ISSW/Papers/R05.5.pdf