1st Edition
High Performance CMOS Range Imaging Device Technology and Systems Considerations
This work is dedicated to CMOS based imaging with the emphasis on the noise modeling, characterization and optimization in order to contribute to the design of high performance imagers in general and range imagers in particular. CMOS is known to be superior to CCD due to its flexibility in terms of integration capabilities, but typically has to be enhanced to compete at parameters as for instance noise, dynamic range or spectral response. Temporal noise is an important topic, since it is one of the most crucial parameters that ultimately limits the performance and cannot be corrected. This work gathers the widespread theory on noise and extends the theory by a non-rigorous but potentially computing efficient algorithm to estimate noise in time sampled systems.
This work contributed to two generations of LDPD based ToF range image sensors and proposed a new approach to implement the MSI PM ToF principle. This was verified to yield a significantly faster charge transfer, better linearity, dark current and matching performance. A non-linear and time-variant model is provided that takes into account undesired phenomena such as finite charge transfer speed and a parasitic sensitivity to light when the shutters should remain OFF, to allow for investigations of largesignal characteristics, sensitivity and precision. It was demonstrated that the model converges to a standard photodetector model and properly resembles the measurements. Finally the impact of these undesired phenomena on the range measurement performance is demonstrated.
1 Introduction
2 State of the art range imaging
2.1 Triangulation
2.2 Interferometry
2.3 Time-of-flight
2.3.1 Direct time-of-flight
2.3.2 Continuous wave method
2.3.3 Pulsed wave method
2.4 Comparison of optical range imaging methods
3 Temporal noise
3.1 Introduction to noise analysis
3.1.1 Basic probabilistic concepts for the analysis of uncertainties
3.1.2 Stochastic processes
3.1.3 Propagation of noise in linear time-invariant circuits
3.2 Noise analysis in non-linear and time-variant systems
3.2.1 Transformation of probability density functions
3.2.2 Employing z-transform for noise analysis
3.2.3 LPTV methods
3.2.4 Propagation of noise in non-linear time-variant systems
3.2.5 Noise in the time domain
3.2.6 A sequential method using a switching time-frequency domain
3.3 Fundamental noise processes in electronic devices
3.3.1 Thermal noise
3.3.2 Shot noise and photon noise
3.3.3 Remarks on thermal noise
3.3.4 Generation-recombination noise
3.3.5 Random telegraph signal noise – burst noise
3.3.6 Flicker noise
3.4 Noise processes under time-varying bias
3.5 Impedance field method
4 Noise performance of devices available in the 0.35μm CMOS process
4.1 Transistor noise basics
4.1.1 Bipolar transistor noise model
4.1.2 Field-effect transistor noise modeling
4.2 Noise performance of standard MOS Field-Effect Transistors
4.3 Noise performance of available bipolar devices
5 Noise in active pixel sensors
5.1 Photodetector principle
5.2 Photodetector noise and reduction techniques
5.2.1 Dark noise
5.2.2 Photon noise
5.2.3 Reset noise
5.2.4 Thermal, flicker and RTS noise
5.3 Correlated double sampling
5.4 Novel JFET readout structure for CMOS APS
6 On the design of PM-ToF range imagers
6.1 Basic concept and constraints
6.2 Physical limitations due to photon induced shot noise
6.3 Design objectives and considerations
6.3.1 Design objectives
6.3.2 Photodetector selection
6.3.3 Sensor system architecture
6.3.4 Fabrication technology
6.4 Detector design and evaluation
6.4.1 Readout circuitry
6.4.2 ToF-LDPD design
6.4.3 Evaluation of the first generation LDPD based PM-ToF imager
6.5 Speed considerations for ldpd based TOF image sensors
6.5.1 Design Considerations for charge transfer speed improvement
6.5.2 Evaluation of the second generation LDPD based PM-ToF imager
6.6 Matching considerations
6.6.1 Alternative ToF-LDPD concept
6.6.2 Evaluation of the third generation LDPD based PM-ToF imager
6.7 Impact of finite charge transfer speed and parasitic light sensitivity on PM-TOF
6.7.1 Concept of the generalized MSI ToF model
6.7.2 Verification
6.7.3 Fitting and comparison of the ToF-LDPD designs
6.7.4 Impact on precision
7 Conclusions
Appendix A Derivation of the autocorrelation formula of shot noise
Appendix B Measurement setups
B.1 Noise measurement setup
B.2 Setup to measure according to the emulated TOF principle
Appendix C Photon transfer method
Nomenclature
Abbreviations
Bibliography
Index
Biography
Andreas Süss received his BSc from the University of Applied Sciences Düsseldorf in 2008 and a PhD degree from the University of Duisburg-Essen in 2014. From 2007 until 2014 he was affiliated to the Fraunhofer Institute IMS where he was working mainly on high-speed, low-noise imagers for e.g. ToF applications. From 2014 until 2015 he had a scholarship from the KU Leuven and worked as a postdoctoral researcher on global shutter imaging at the MICAS department in collaboration with IMEC, Leuven. As of 2015 he is hired as an R&D engineer in the IMEC imaging division, where he is currently responsible for the pixel development for global shutter and high-speed applications. His research interests include modeling, temporal noise, optimization, compressed sensing and depth imaging.
