TL;DR
This paper presents a fully integrated analog front-end (AFE) for Short-Wave Infrared (SWIR) detectors, achieving a high frame rate of 100 Hz and excellent linearity (99.59%). The design features a compact pixel structure, advanced noise reduction techniques, and a novel two-step single-slope ADC, pushing the boundaries for high-performance SWIR imaging.
What Problem Does This Paper Address?
High-performance SWIR imaging systems are crucial for applications ranging from military reconnaissance to environmental monitoring. The quality of these systems hinges on the performance of their detectors, which are composed of the detector array, the analog front-end (AFE), and digital signal processing. The AFE, in particular, is a bottleneck, needing to efficiently convert photocurrent into a digital signal with high linearity, low noise, and fast readout speeds. Existing solutions often struggle to balance these competing requirements, especially for large-format arrays where pixel size, noise, and frame rate become critical. This paper aims to address these challenges by proposing an integrated AFE that optimizes pixel design, noise suppression, and analog-to-digital conversion for improved SWIR imaging performance.
Methods
The proposed AFE integrates a readout integrated circuit (ROIC) and a 12-bit, 100 kHz two-step single-slope analog-to-digital converter (TS-SS ADC).
- ROIC: Employs a Direct Injection (DI) pixel structure with a small 10 µm × 10 µm size. To enhance the Full Well Capacity (FWC) and dynamic range within this compact area, the integration and holding capacitors are shared. Noise is further reduced at the column level using a passive correlated double-sampling (CDS) circuit.
- TS-SS ADC: This architecture is chosen for its simplicity and small area. Key innovations include a four-input comparator to mitigate linearity issues caused by charge redistribution during coarse quantization and a current-steering digital-to-analog converter (DAC) to compensate for switch-induced non-idealities, thereby improving differential nonlinearity (DNL) and integral nonlinearity (INL). The ADC also incorporates a 7-bit fine quantization stage with a redundancy window to eliminate dead zones.
- Implementation: The entire AFE is implemented using SMIC 180 nm 1P6M technology.
Datasets
Not discussed in the paper. The paper focuses on circuit design and post-simulation results.
Key Insights and Ideas
- Shared Capacitors for FWC: Sharing integration and holding capacitors within the pixel is a clever way to maximize FWC in a limited pixel area, a common constraint in high-density imagers.
- Passive CDS for Noise Reduction: Utilizing a passive CDS circuit at the column level is an effective strategy for suppressing fixed-pattern noise (FPN) and reset noise without significant area or power overhead.
- Four-Input Comparator for Linearity: The novel four-input comparator design directly addresses a known issue in two-step ADCs where charge redistribution during coarse quantization can degrade linearity. This innovation is crucial for achieving high overall linearity.
- Current-Steering DAC for Non-Linearity Compensation: Employing a current-steering DAC to counteract charge injection from switches is a sophisticated method to improve the DNL and INL of the ADC, leading to more accurate digital representations of the analog signal.
- Coarse/Fine ADC with Redundancy: The combination of a coarse and fine quantization stage, with a 7-bit fine stage providing redundancy, effectively eliminates quantization dead zones, ensuring a more complete and accurate conversion across the entire input range.
Key Results
The post-simulation results demonstrate impressive performance metrics:
- Frame Rate: 100 Hz, enabling high-speed imaging.
- Full Well Capacity (FWC): 2.8 Me⁻, indicating good charge handling capability.
- Linearity: 99.59%, signifying excellent signal fidelity.
- Equivalent Output Noise: 243 µV, demonstrating effective noise suppression.
- Dynamic Range: 73.8 dB, suitable for capturing scenes with varying illumination.
- ADC Performance:
- SNDR: 68.38 dB
- ENOB: 11.06 bits
- Sampling Rate: 100 kHz
Limitations and Future Work
- Limitations:
- The paper presents post-simulation results, not measurements from a fabricated chip. Real-world performance may differ.
- While the design aims for compactness, the overall area of the AFE is 3089 µm × 1895 µm, which might still be a consideration for extremely large arrays.
- The output swing of the ROIC (1.204 V) is noted as slightly smaller than some previous designs, though this is balanced by improved noise performance.
- Future Work:
- The authors suggest that techniques like order-statistics/optimal-arrangement (OA) of DAC elements and split-DAC calibration could be integrated in future revisions to further reduce INL/DNL.
- The paper highlights the potential for further improvements in pixel density and continuous enhancement of imaging detector performance.
Summary updated