Laboratory Exercise 4

 

Post Simulation Exercise to Determine Functionality and Actual Timing Parameters of a VLSI Full Custom Layout

 

Introduction:

 

The objective of this laboratory exercise is to lead the student through a post simulation process associated with a full-custom VLSI layout.  Mach-PA timing and power analyzer

will be used to simulate the design created in exercises 2 and 3.  MACH-PA is a SPICE like simulator that is used to perform pre-layout and post-layout timing simulations.  This simulator is able to handle analog, digital, and mixed signal circuits.  The design input to MACH-PA is a SPICE netlist, usually with extracted parasitic capacitances.  A technology file describing the fabrication parameters is developed during the MACH-PA calibration process.  The calibration process has been completed for the AMI05 process and the calibration file is loaded when launching the simulator.  The circuit netlist is compiled into a binary file by MACH-PA.  Commands to operate the simulator can be entered in the User Interface Window or placed in a command file (do file) and executed by selecting the Run Commands icon  and selecting the file to run.  A test vector file with input signals to force (high or low) and the desired responses (output signals) can also be created.  The Run Test Vectors command applies the input signals to the circuit and checks the simulator's output signals against the desired output values in the test vector file.  Errors are reported when desired outputs do not match simulated outputs.

 

The Mach TA simulation algorithm has three main components:

 

Design partitioning - Mach-PA divides the design into several partitions, allowing the simulation algorithm to evaluate only those partitions that are changing at any instant instead of evaluating the entire design. The result is high-performance without compromising accuracy.

 

Time-step control - Mach-PA uses a combination of time-driven and event-driven methods for time-step control.

 

Integration - Mach-PA employs an integration method on the linearized device models and employs non-linear convergence techniques to attain convergence. It solves the linearized models using a proprietary high-performance graph theory based matrix solution algorithm.

 

Procedure:

 

 

Hand in:

 

ECE Home

EE715 Home