CMOS Camera System-on-a-Chip ECE Senior Project CMOS Camera System-on-a-Chip FIRST PAGE
TEAM MEMBERS Anil Kumar Angana Sheth Saurabh Desai George Moran Jason Moffa Takashi Ishihara ADVISOR: Dr. Brita Olson SECOND PAGE
CMOS Camera System-on-a-Chip A normal camera uses film to save the image that hits its lens. We are designing a chip that saves the image digitally by sensing the amount of light or image that each one of its pixel sees. The image consists of 128 by 128 pixels: there are total 16384 pixels. ANIL INTRO 1
PROJECT DIAGRAM 128 x128 ANIL INTRO 2
Design Goals ANGANA INTRO 1
Establishing a Chip Design Infrastructure Learning VLSI Design Learning VLSI Design Tools and Chip Design Process Establishing a Chip Design Flow at CPP Configuring tools ANGANA INTRO 2
Progress Calculations/Analysis: Component Development: Floor Planning Conversion Gain Parasitics /Loading Noise Component Development: Design Optimization, Simulation with loading, Port to new environment Decoder primitive Anti-blooming Circuitry Decoder Bus Driver Row driver: Row RST & SEL Driver Anti-blooming Circuitry Amplifier Bias Circuitry Pixel Analog Signal Chain Sample & Hold Circuit ANGANA INTRO 3
Progress (cont.) Chip Level Design: In progress: Pixel Array schematic 7-bit decoder Analog Signal Chain with Correlated Double Sampling ANGANA INTRO 4
Accomplished Tasks Floor Planning 7 Bit Decoder Anti Blooming Circuitry
FLOOR PLANNING Pitch = 150um 16 pads 16 pads 3mm 136*12um 3mm The floor plan estimates the area of major units in the chip and defines their relative placements. The floor plan is essential to determine whether a proposed design will fit in the chip area budgeted and to estimate wiring lengths and wiring congestion, so an initial floor plan should be prepared as soon as the logic is loosely defined. Pitch: It is a distance between two pads. Pads: They are wired to the pins on the chip package. They are for the I/O connection on the chip 128*128 Imager 3mm 16 pads 136*12um 3mm 4 AMI 0.5um Tiny Chips Packaging : 150um pad pitch Pixels : 12um pitch
APS Architecture Row Decoder: Selects row for readout. Column Decoder: V E S Row Decoder: Selects row for readout. Column Decoder: Controls readout of Pixels In given row. 128*128 PIXEL ARRAY Row[i] 27=128 128 A decoder is a Combinational circuit, when enabled, produces one of 2n minterms or maxterms at the output based on the input Combinations. READOUT CIRCUITS 128 COLUMN DECODER 27=128
Decoder Implementation Row-add<0:6> Row-add<0:6> Selects row 0 1 Selects row 1 A 7 input nand gate based implementation results in a compact and regular realization reducing development time and cost. The chip that we are designing has 128*128 pixels so the decoder consists of 128 of 7 input nand gates.
Decoder Design Requirements Master Clock = 20MHz Trise/Fall requirements are relaxed. In our design Trise/Fall times are 50% of clock. Trise/Fall = ½(1/20MHz) = 25ns
Schematic of 7-input nand gate PMOS Operational W/L = (W min/L min to 7W min/L min) NMOS Operational W/L = W min/7L min Both NMOS & PMOS are minimum sized transistor (W=1.05um, L=0.7um) W/L eff PMOS : 1.05/0.7 W/L eff NMOS : 1/7(1.05/0.7)
Simulation for fall time when W = 1.05um
Simulation for rise time when W = 1.05um
Rise / Fall Time Table Rise Time Fall Time W = 1.05um 1.46ns 2.709ns L = 0.7um Better Trise/Fall times results due to reduced W of PMOS transistor. Reduced W of transistor results in Reduction of power consumption Reduction in substrate noise Reduction in input capacitance – Reduces chip area, power, & noise.
Anti Blooming Circuitry Reduces charge buildup in pixels due to excessive illumination Prevents flow of excess charge into neighboring pixels. It does this by redirecting the excess current into the anti-blooming drain when the photodiode is too full. Without anti blooming circuitry imaging artifacts will happen.
Anti Blooming Circuit Operation Vdd RST = 5V RST_LO = 1V VTH = 0.7V RST 5v RST_LO = 1V Integration Normal Imaging Condition: When RST signal is applied FD is around 2.8V and after integration of light it becomes 1.8V. (Vgs < VTH) = 1 – 1.8 < 0.7 - Transistor Off Bright Light: Due to bright light FD decreases to 0.3V (Vgs >= VTH) = 1 – 0.3 >= 0.7 – Transistor on and excess carriers removed.
Anti Blooming Circuitry Final stage of row driver Anti blooming circuitry
7 BIT INPUT SIGNAL DECODER DRIVER ANIL SLIDE 1
Schematics Driver (4x8x) Driving a load of 64, CMOS N and P transistors. ANIL SLIDE 2
Simulations Rise Time ~ 6% Fall time ~ 5% Approximate time is 1 nano second to drive signal. ANIL SLIDE 3
VALUES GIVEN W = 1.05uM L = 0.7uM VDD = 3.3v
SCHEMATIC V5 V4 V0 VDD
SCHEMATIC RESULT
SYMBOL FOR THE AND GATE ROW [i] RST[i] RST
Accomplished Tasks 1. Determine Resistive and Capacitive Parasitic Loading (Caused by dimension of the pixel Array) 2. Row Driver Design -Determine best combination for Drivers -Rise and Fall Times -Reset Driver -Select Driver 3. Simulations using PSPICE 4. Implementation of Design using Cadence (LINUX) GEORGE SLIDE 1
Chip Schematic GEORGE SLIDE 2
Overview of Row Driver SEL SEL[i] ROW[i] GEORGE SLIDE 3
Parasitic Loading Determination of load resistor: Determination of Capacitive load: GEORGE SLIDE 4
The Row Driver GEORGE SLIDE 5
The Row Driver (cont.) GEORGE SLIDE 6
Simulations Summarized Table 1: Rise and Fall Times for Inverter driving 128 transistors (using 102,400,402,600ns; Vdd=3.3V; Cload= 41.4fF) Table 2: Rise and Fall times for inverters driving inv_8x GEORGE SLIDE 7
Schematics Two Drivers (4x_8x) Driving 128 CMOSN transistors GEORGE SLIDE 8
Simulations Rise/Fall Times GEORGE SLIDE 9
Row Select Driver Design Three Drivers (1x_4x_8x) Driving 128 CMOSN transistors GEORGE SLIDE 10
Row Select Driver Simulation with Loading Rise/Fall Times GEORGE SLIDE 11
Row Reset Driver Design Introduction of two different power supply levels GEORGE SLIDE 12
Simulations Rise/Fall Times GEORGE SLIDE 13
General Chip Layout
Pixel & Timing Diagrams
Current Mirror Circuit To Bias Pixel SF “Large” gate widths and lengths used Good threshold matching gm reduced Current Equation
Determining Initial Condition For Sample & Hold Capacitor Pixel Sample & Hold Capacitor Bias Optimize RST signal due to Body Effect
The Body Effect Source ≠Body Modified threshold voltage:
Initial Condition for Sample & Hold Capacitor
Determining Time to Discharge Sample & Hold Capacitor Within 0 Determining Time to Discharge Sample & Hold Capacitor Within 0.1% Accuracy VFD initial condition: V = 1.8V
Transient Response of Pixel Discharge
Determining time to charge sample & hold capacitor Within 0 Determining time to charge sample & hold capacitor Within 0.1% Accuracy
Transient Response of Pixel Charge
Accomplishments Designing output driver Determining the conversion gain Designing the pixel array
Output Driver Pixel Array ADC Analog Sig. Chain Parasitic Capacitance
Output Driver Operation CL = 8pF Vout Vin DV = 1V Slewing (25%) Settling (75%) Step function Ttot = 100% = .5msec Slewing: small sig. Analysis not appropriate Settling: transistors in saturation
Schematic
Settling Vin DV = 1V Amplifier transfer function: Output: Settling accuracy Vin DV = 1V Step function:
Tfall output Driver (35uA)
Trise output Driver (35uA)
Slewing and Settling
Conversion Gain Cm Cm(1-A) C(total) = Cfd + Cm(1-A) V = Q/C Miller Effect Cm Cm(1-A) C(total) = Cfd + Cm(1-A) V = Q/C C.G.= Vout/Number of Electrons = q/C(total)
Conversion Gain Gate Source (3) W 4λ 4λ W (2) (1) Source (1): Cj(W*4λ) (2): Csw(W+2*4λ) (3): Cswg*W Cfd = Cj(W*4λ)+ Csw(W+2*4λ)+Cswg*W =1.19fF
Conversion Gain Vout 0.8 0.9 0.9 C(total) C(total) = Cfd+Cm(1-A) = 1.19fF Input: C.G.= q/C(total) = 134uV/electron Output: C.G.= (0.8)(0.9)(0.9)(134u) = 87.1uV/electron
CHIP PROCESS ANIL CONCLUSION 1
Remaining Tasks Component development Chip Level Schematic Development Optimization of Analog Signal chain with CDS Chip Level Schematic Development Completion of Decoder Pad Ring Assembly of final chip schematic “Full” Chip Simulations Layout ANGANA INTRO 5
Acknowledgments ANGANA CONCLUSION 1