TL;DR
This paper introduces a novel, low-cost testing methodology for LiDAR receiver integrated circuits (ICs) that effectively eliminates the "walk error" issue. By incorporating a high-speed buffer and a variable delay line before the time-to-digital converter (TDC), the system generates START and STOP pulses with nearly identical shapes, enabling more accurate and simpler measurements without complex compensation algorithms.
What Problem Does This Paper Address?
LiDAR (Light Detection and Ranging) systems are increasingly vital for applications ranging from augmented reality to autonomous vehicles and indoor monitoring. A core component of these systems is the receiver IC, which detects reflected light pulses to determine distance. A common challenge in testing these receiver ICs is the "walk error." This error arises because the electrical START pulse (sent from the transmitter) and the received STOP pulse (reflected light) have different rising edge slopes. This difference, exacerbated by signal processing within the receiver, leads to inaccuracies in measuring the time-of-flight, and consequently, the distance. Traditional methods to mitigate this error often involve costly laser drivers or complex compensation algorithms that add to the IC's complexity and cost. This paper aims to provide a simpler, more cost-effective solution for testing LiDAR receiver ICs by addressing the walk error at the measurement methodology level.
Methods
The proposed solution centers around a custom-designed CMOS optoelectronic receiver IC (CORIC) that incorporates a unique measurement methodology. The key innovations are:
- On-chip APD: To avoid signal distortion and packaging costs associated with off-chip avalanche photodiodes (APDs), an on-chip APD is integrated.
- Analog Front-End (AFE): A robust AFE, comprising a dual-feedback folded-cascode differential transimpedance amplifier (DFD-TIA), an active single-to-differential converter (ASD), a cross-coupled inverter-based post-amplifier (CI-PA), and a two-stage differential amplifier with negative impedance compensation (TDA-NIC), is used to process the weak optical signals.
- High-Speed Digital Input Buffer (HS-DIB): This buffer converts the differential analog signals from the AFE into full-swing digital signals, crucial for accurate timing estimations by the TDC.
- Variable Delay Line: This is the core of the walk error mitigation. Instead of relying on the original START pulse and the processed STOP pulse, the START pulse is intentionally delayed using a series of inverter chains with controllable MOS capacitors. This allows the START and STOP pulses to be generated with very similar pulse shapes, effectively eliminating the slope difference that causes walk error. The delay can be adjusted by varying the number of delay cells.
- Modified 2D Vernier TDC: A resettable T-latch based Vernier TDC is employed. This architecture is designed to handle narrow input pulses and avoid ambiguity from pulse overlapping, further contributing to accurate time interval measurement.
Datasets
Not discussed in the paper. The paper focuses on the design and simulation/measurement of a custom IC.
Key Insights and Ideas
- Walk Error Mitigation at the Measurement Level: The most significant insight is that the walk error, a fundamental issue in LiDAR receiver testing, can be effectively addressed by manipulating the timing and shape of the START pulse during the testing process itself, rather than solely relying on complex compensation within the receiver IC.
- Cost Reduction through Integrated Design: By integrating the APD on-chip and developing a simpler testing methodology, the overall cost of LiDAR receiver IC development and testing can be significantly reduced.
- Simultaneous Pulse Shaping: The variable delay line's ability to shape the START pulse to closely match the STOP pulse is a clever way to equalize their characteristics, thereby neutralizing the walk error.
- Flexibility in Delay Control: The use of a 4-bit thermometer-to-binary decoder to control the delay line offers fine-grained control over the time interval, allowing for testing across a range of distances.
Key Results
The paper demonstrates the successful implementation of the proposed CORIC in a 180-nm CMOS process.
- Elimination of Walk Error: Post-layout simulations and chip measurements show that the START and STOP pulses generated by the proposed methodology have nearly identical shapes, effectively eliminating walk errors.
- Accurate Time Interval Measurement: The variable delay line allows for precise control of the time interval between the START and STOP pulses, with tested delays ranging from 1 ns to 11 ns.
- Low Power Dissipation: The chip core exhibits a power dissipation of 20.9 mW from a single 1.8-V supply.
- Compact Chip Area: The chip core occupies a small area of 841 × 268 µm².
- Performance Comparison: Table 1 shows that the proposed CORIC offers competitive performance in terms of distance measurement and power consumption compared to prior art, while achieving its primary goal of simplified testing. For instance, it achieves a minimum distance of 0.3m and a maximum distance of 3.0m with a power dissipation of 20.9 mW.
Limitations and Future Work
- Limited Detection Range: The current design is optimized for short-range applications, with a demonstrated maximum distance of 3.0 meters. While the methodology itself is sound, extending the detection range would require further optimization of the AFE and potentially the laser driver.
- Future Work: The authors suggest that the design space of the delay cells can be expanded to enable the detection of reflected pulses within longer ranges. This implies that by increasing the maximum delay achievable by the variable delay line, the system could be adapted for longer-distance LiDAR applications.
Summary updated