530 likes | 704 Views
Outline. Project OverviewGoalsBackground InfoSpecificationsProgressComponentsFunctionLogicDiagramsSimulationWaveformsPhysical LayoutProcessLayout OptionsDiagramsLVS ConfirmationPadframe. EvaluationIssues / ImprovementsFinal ThoughtsLooking AheadReferencesSpecial Thanks. Projec
E N D
1. FFT COMPONENTS PROJECTFinal Design Review / December 12, 2005 TEAM MEMBERS: Scott Crawford (Leader)
Jon Oppelaar
2. Outline Project Overview
Goals
Background Info
Specifications
Progress
Components
Function
Logic
Diagrams
Simulation
Waveforms
Physical Layout
Process
Layout Options
Diagrams
LVS Confirmation
Padframe
Evaluation
Issues / Improvements
Final Thoughts
Looking Ahead
References
Special Thanks
3. Project Goals Support stage 4 of the fingerprint authenticator project
Create 16-bit complex multiplier and adder for use in FFT algorithm
Use VLSI technology to create an Application-Specific Integrated Chip (ASIC) to submit for fabrication
Document and present results
4. Background Info Multiplier - A 16 bit complex multiplier that is synchronous, triggered on a rising edge of the system clock.
Adder - A 16 bit adder
Butterfly – Radix-4, consists of multipliers, adders and a twiddle factor memory
Input/ Output Interface - FFT algorithm requires the entire sequence of input samples to be provided on its input port in order to initiate the operation. START and DONE signals should be available.
5. Project Specifications 100 MHz clock speed is desired
Algorithm radix-4
Arithmetic: 16 bit fixed point, 2’s complement
Pipelining of complex arithmetic operations if 100MHz not possible
Identical standards for all control and status signals
Clock input (CLK) and reset (RST)
Arithmetic operators contain an overflow status line (OVFL)
6. Project Strategy Make a tentative schedule of tasks to keep progress at a steady rate
Research topics and technical data needed in order to build components
For all components, follow these steps:
Sketch drafts of design as a gate level schematic
Design schematic on Mentor Graphics software (CAD)
Simulate and test component using SPICE based simulation
Correct and verify functionality (create block symbol if necessary)
Create physical layout of final transistor-level multiplier component
Correct all LVS and DRC-related issues and errors
Create padframe layout for MOSIS fabrication
Use Mach-TA to determine timing data for circuit
Submit layout to MOSIS
Confirm functionality of IC when received back from MOSIS
Create presentation for completed project
7. Collaboration Log 9/18/2005
Begin research and decide tasks that need to be completed.
9/25/2005
Figure out Boolean logic and begin creating gate level schematic.
10/11/2005
Continue designing gate level schematic for adder circuit. Write first pre-report for project leader.
10/14/05
Begin designing half-adder design in software
10/17/05
Finish half-adder design and begin full-adder design in software
10/24/2005
Finishing full-adder design in software and prepare for PDR presentation
10/31/2005
Create 8-bit adder circuit using single-bit full-adder blocks
11/2/2005
Simulate 8-bit adder design, confirm functionality
11/3/2005
Begin 4-bit multiplier design
11/7/2005
Implement 4-bit multiplier design
11/11/2005
Continue working on 4-bit multiplier
11/13/2005
Simulate 4-bit multiplier
8. Collaboration Log (2) 11/14/2005
Add 2s compliment functionality to multiplier
11/15/2005
Create 8-bit subtractor and simulate circuit
11/16/2005
Create complex multiplier in software and simulate circuit
Update documentation and create presentation for upcoming CDR
11/18/2005
Create buffers for multiplier to synchronize circuit
11/22/2005
Simulate multiplier circuit with buffers
Generate screen shots of simulation results for all components
11/28/2005
Begin physical layout of complex multiplier circuit
11/29/2005
Begin Debugging DRC and LVS errors from physical layout
12/4/2005
Begin padframe layout and auto-routing to pins
12/6/2005 – 12/9/2005
Manually route padframe to layout
Finish fixing DRC errors from padframe
12/9/2005 – 12/12/05
Gather information and create presentation for FDR
9. Task Assignments Tasks divided between implementation and documentation
Jon: Transistor and physical layouts in CAD, debugging, component integration
Scott: Scheduling, logs, reports, presentations, data collecting
Consistent collaboration and progress updating within team
Shared responsibilities for circuit design, simulation and verification
10. Current Project Progress
11. Project Components
12. Single-Bit Half-Adder Adds two bits together with a sum and carry output
Used to optimize partial product calculations in 4-bit multipliers
Building block for full-adder
13. Single-Bit Half-Adder SCHEMATIC
14. Single-Bit Half-Adder SIMULATION WAVEFORM
15. Single-Bit Full-Adder Adds three one-bit inputs and outputs a sum and a carry
Similar to half-adder, but has an input carry bit
Consists of two half-adders connected with an OR gate
Vital component in multiplier circuit
Cascaded full-adders used to create an 8-bit adder
16. Single-Bit Full-Adder
17. Single-Bit Full-Adder
18. 8-Bit Full-Adder Consists of eight cascaded, single-bit full adders
‘Carry’ bit is traversed through the circuit
Used in complex multiplier arithmetic calculation
19. 8-Bit Full-Adder Fig. A - Cascaded single-bit full-adders form an 8-bit full-adder
Fig. B - The 8-bit adder represented as a block symbol
20. 8-Bit Full-Adder Uses negative and positive numbers with a transition of the carry bit to show 2s compliment Input A: 00111000 (+56 in decimal)Input B: 11101011 (-21 in decimal)Expected SUM: 00100011 (+35 in decimal, CarryIN=0)
21. The waveforms show that the SUM is indeed 00100011 (+35) 8-Bit Full-Adder
22. Single-Bit Full-Subtractor Takes in three one-bit inputs: A, B, and “Borrow In”
Subtracts B from A, outputs the difference and a “Borrow Out” bit
Same circuitry as full-adder except for an extra inverter, since borrow and carry are inverse operations
Component is needed to create complex multiplier
23. Single-Bit Full-Subtractor
24. Single-Bit Full-Subtractor
25. 8-bit Full-Subtractor Identical design to 8-bit full-adder
Consists of eight cascaded, single-bit full subtractors
‘Borrow’ bit is traversed through the circuit
Used in complex multiplier arithmetic calculation
26. 8-bit Full-Subtractor Fig. A - Cascaded single-bit full-subtractors form an 8-bit full-subtractor
Fig. B - The 8-bit subtractor represented as a block symbol
27. 8-bit Full-Subtractor Uses negative and positive numbers with a transition of the carry bit to show 2s compliment and carrying functionalities
Input A: 00111000 (+56 in decimal)Input B: 11101011 (-21 in decimal)Expected DIFF: 00100011 (+77 in decimal, BorrowIN=0)
28. The waveforms show that the DIFF is indeed 01001101 (+77) 8-bit Full-Subtractor
29. 4-bit Multiplier Must work for 2s compliment numbers and be able to multiply two 4-bit inputs
Uses arrays of ‘AND’ gates to generate partial products and a sequence of adders to sum them together
To multiply two n-bit numbers, n˛ ‘AND’ gates required
4˛ = 16 ‘AND’ gates
n˛-n adder blocks required
4˛-4 = 12 adder blocks required
Product is length 2n
2*4 = 8-bit product
30. 4-bit Multiplier To convert into 2s compliment, all ‘AND’ gates containing sign bits are inverted
Final sign bit output also inverted
To optimize, half-adders can replace full-adders on right-most position in partial products (since no carry input exists)
31. 4-bit Multiplier
32. 4-bit Multiplier Used negative and positive numbers to show 2s compliment functionality
Input A: 1111 (-1 in decimal)Input B: 0111 (+7 in decimal)Expected PRODUCT: 11111001 (-7 in decimal)
33. 4-bit Multiplier The waveforms show that the PROD is indeed 11111001 (-7)
34. Complex Multiplier Capable of multiplying two 8-bit complex numbers
Splits outputs into real and imaginary products
Each output is 8-bits with each with an overflow bit
Makes use of previous components in design
35. Complex Multiplier
36. Complex Multiplier
37. Complex Multiplier Project specifications require multiplier to run 100 MHz or higher
Delay needs to be under 10ns
Max delay is measured as the time between the first input and the final output
38. Complex Multiplier Max delay = 4.816 ns
39. Complex Multiplier Use of buffers on inputs and outputs allows for multiplier to be synchronous on a clock pulse
Reduces noise in the data
User has more control of circuit
Ability to set speed of circuit using clock
Reset capability
40. Complex Multiplier Consist of eight D-latches connected to single reset and clock pins
Inverters placed on outputs to correct inverted QB output on latches
Input cannot be sent faster than the clock speed, otherwise errors will occur
41. Complex Multiplier
42. Complex Multiplier Given the following inputs:
A: 1001 (-7 in decimal)
B: 1010 (-6 in decimal)
C: 1001 (-7 in decimal)
D: 1010 (-6 in decimal)
The output should return:
(AC - BD) + (AD + BC) j
(49 – 36) + (42 + 42) j
13 (real) + 84 (imag.)
43. Complex Multiplier The real and imaginary outputs:
R: 00001101 (+13 in decimal)
I: 01010100 (+84 in decimal)
This is the correct answer for the given inputsAlso note how the data is cleaner after adding the buffers (no spikes or noise)
44. Physical Layout Consists of two main tasks
Converting transistor-level design to physical-level design
Placing and routing physical layout onto padframe
Using Mentor Graphics allows for greater compatibility
Each step requires rule checking and debugging
Choice of auto or manual (custom) layouts
We chose auto-layout for the physical-layer design and manual layout for routing the padframe
45. Physical Layout Auto layout lets software generate the physical layout
Manual layout requires the user to create the physical layout from scratch
Pros/Cons of auto-layout
Pros
Saves time – quickly created by Mentor Graphics
Software has more knowledge about design rules
Cons
Software does not use good space-saving techniques
Layout not as optimal as user can make it
46. Physical Layout
47. Physical Layout
48. Physical Layout
49. Physical Layout
50. Evaluation Time constraints caused the project to have limitations
Potential improvements in our design
More research on better full-adder designs, such as carry-select and carry-bypass
Better methods to calculate complex multiplication
Use D-flip-flops instead of D-latches
Simulations took long periods of time to complete
51. Evaluation Successful and rewarding project
Gained knowledge of full ASIC design process using VLSI
Good experience working as a team
Better understanding of the complexities of chip design
52. Looking Ahead… Complete layout testing in MACH-TA
Submit design to MOSIS for fabrication
Complete final design review report
Test chip (April 2006), verify functionality and record results
Prepare for project presentation in May 2006
53. References Logic Gates
http://isweb.redwoods.cc.ca.us/INSTRUCT/CalderwoodD/diglogic/index.htm
3-input XOR Logic
http://www.shef.ac.uk/physics/teaching/phy107/othergates.html
8-bit adder notes
http://www.ece.arizona.edu/~ece274/lectures/lecture9.pdf
Full and half-adders
http://www.phys.ualberta.ca/~gingrich/phys395/notes/node129.html
4-bit multiplier info
http://www-mount.ee.umn.edu/~sobelman/papers/kavitha_iscas04.pdf
2s compliment multiplier
http://tima-cmp.imag.fr/~guyot/Cours/Oparithm/english/Multip.htm
2s compliment
http://www.evergreen.edu/biophysics/technotes/program/2s_comp.htm
UNH ECE 715/815
http://ece.unh.edu/courses/ece715
ASIC Design Kit (ADK) Documentation and Tutorials
http://www.mentor.com/partners/hep/AsicDesignKit/Docpage.html
54. Special Thanks PROF. ANDRZEJ RUCINSKI – for helping us through the design process
FRANK HLUDIK – for his endless help in the design automation lab
EVERYONE IN ECE 715 – for their open cooperation with project issues
MENTOR GRAPHICS – for use of their Electronic Design Automation (EDA) software
MOSIS – for funding our chip fabrication