Digital Systems Engineering

Download Report

Transcript Digital Systems Engineering

CSE464
Digital Systems Engineering
L0: Logistics and Introduction
David M. Zar
Computer Science and Engineering
Washington University
[email protected]
(Based on Original Work of Fred Rosenberger)
Logistics EE464, Spring 2009
Lectures:
Textbook:
Grading:
MW 2:30-4:00 in Urbauer 116
Dally and Poulton, Digital Systems Engineering
Approximate weighting for grade determination
25% Homework (exams based on homework)
35% First midterm exam
45% Final exam
Collaboration: Academic integrity will be taken seriously. You
may collaborate on homework with other students, use solutions
from last year, or get help from anyone but you are to state who
you worked with or got help from, and give an estimate of
contribution from other sources to what you submit. This is just
acknowledgement of source of material, and recognition of the
work contributed by others, it has no effect on your grade.
Exceptions to this rule will be specified in the assignments. Tests
and projects are to be entirely your own work.
Intro-‹#› - DMZ - 3/27/2016
More Logistics
Exams:
Closed book. One handwritten (no photocopies) sheet (both
sides) allowed on first exam, two sheets final exam.
Homework:
Usually due in class (2:30 p.m.) on assigned date.
Instructor:
David M. Zar
Bryan 307C
GPS: N38 38.979' W90 18.360' Elevation: 550'
Office Hours: by appointment.
[email protected] (best contact method)
http://www.cse.wustl.edu/~dzar
Attendance:
Class attendance is important, material will
covered in class that is not in text. You will
wish to get copies of notes from classmate if
you miss class.
Intro-‹#› - DMZ - 3/27/2016
Discussion Topics
 Class
attendance
 Class participation (please!)
 Dally (Dally, not Daily) lecture notes
 Textbook cost, errata, etc.
 Homework length and style
 EE314: Engineering Electromagnetics 1:
Fundamentals
 Goal: Extremely practical backed up by theory
and analysis
Intro-‹#› - DMZ - 3/27/2016
Interesting Links and Sources
– Links
 http://www.signalintegrity.com/
 http://www.nesa.com/
 http://www.sigrity.com/
 http://www.ultracad.com/ (careful here)
 http://www.automata.com/ (careful here, some calculations are wrong)
 http://groups.yahoo.com/group/si-list (this is the signal integrity
mailing list archive; lots here, good and bad)
– Books
– High-Speed Digital Design, A Handbook of Black Magic, Howard W. Johnson
and Martin Graham, Prentice Hall, ISBN 0-13-395724-1, 1993.
– High-Speed Signal Propagation, Advanced Black Magic, Howard W. Johnson
and Martin Graham, ISBN 0-13-084408-X, 2003
– Brooks, Bogatin, Ritchey, Granberg, ...
– Transmission Lines with Pulse Excitation, Georges Metzger and Jean-Paul
Vabre, Academic Press, New York, NY, 1969. (Bergeron Diagrams)
Intro-‹#› - DMZ - 3/27/2016
Today’s Assignment
 Reading
» Complete before Wednesday, Jan 14 class
– Preface, Chapter 1
» Complete before Wednesday, Jan 21 class
– Chapter 2
» Complete before Monday, Jan 26 class
– Chapter 3, Sections 3.1 through 3.3.3
Intro-‹#› - DMZ - 3/27/2016
Digital Systems Engineering

From Dally
» noise management
– keeping signals clean
» signaling
– moving bits from here to there
» timing
– how we know when a new bit is here
» power distribution
– DC voltage with AC current
Rules of Thumb (e.g.: C/inch, L/inch)
Analytical/Calculation
Simulation (HSPICE)
Measurement
Tools (don’t be a “tool driver”):
Intro-‹#› - DMZ - 3/27/2016
» Signal integrity
– High-Speed signals
– low speed signals
–
reset
–
…
– All Signals
» Signaling (electrical representation
of signals)
» Timing/Clocking
» Power distribution
» Cooling/Packaging as part of
above
Wires
 Advanced
components: Wires
 Care and feeding of wires
 Wires can be expensive, even if it’s a scam
» http://www.monstercable.com
» http://www.jpslabs.com/aluminata.shtml (they claim
“Pricing- If you have to ask.”)
– "The shear mass of the particle shield alone proves that JPS has the
transfer of noise taken care of- Nothing gets through this cable's
shield and into the conductors beneath- NOTHING... "
Intro-‹#› - DMZ - 3/27/2016
Why is Digital Systems Engineering/Design Harder
Now Than Previously?


Wires are not ideal (never were, but valid approximation sometimes) !!!
Gross Simplification here:
Long wires or High-Speed Design (delay ~> 0.1 Clock Period)
»
»
»
»
»
»
»





1968 1 MHz; >40 ft
1978 10 Mhz; >4 ft
1988 100 Mhz; >5 inches
1998 1 GHz; >0.5 inch
2008 10 GHz??? >0.05 inch
2018 ?????
Overall size of system? approximately constant
Ad hoc methods that worked in the past now fail
To paraphrase Roy Jewell, President of TMA: “The rules of physics don’t
change for high-speed design, they are just more strictly enforced”.
The fact that long wires are harder to deal with does not imply short ones
are easy, nor that they can be ignored.
The fact that high-speed signals are harder to deal with does not imply that
slow ones are easy, nor that they can be ignored (e.g. Reset).
EMI: Even harder, we will not deal with this in CSE464.
Intro-‹#› - DMZ - 3/27/2016
Outrageous Statements
 Propagation
delay to closer device is longer
 Slower is better (and faster)
 A capacitor is an inductor unless you want inductor
» Converse for inductor (or resistor)
 Short
wire is worse than long one (e.g. probe)
 Negative characteristic impedance?
 Square corners on PC traces are bad?
 Vias on PC traces are bad?
 Resistor networks: Bad?
 We could not use perfect logic (Midas touch)!
 Tune your absurdity detector!!!
Intro-‹#› - DMZ - 3/27/2016
Thoughts to Remember (Models)
A
model is an artifice to make you think you
understand a problem better than you actually do.
 All models are wrong, but some models are useful.
 Make everything as simple as possible, but no
simpler (A. Einstein).
 In theory there is no difference between practice and
theory, but in practice there is!
 An approximate answer to the right question is
worth a good deal more than the exact answer to an
approximate problem. John Tukey (FFT Fame)
 Models are a really dangerous (and necessary) tool
 Example: ground, ideal ground, logic ground, safety
ground, …
Intro-‹#› - DMZ - 3/27/2016
Ground is Fiction
 At
low frequency and low accuracy ground is a
convenient model
 Be very careful
 What is “ideal ground” (see si-list)?: “What you
draw with chalk on a blackboard!”
Intro-‹#› - DMZ - 3/27/2016
Bonus: Safety Ground
Intro-‹#› - DMZ - 3/27/2016
What is Wrong Here?
Intro-‹#› - DMZ - 3/27/2016
What is Wrong Here?
Lots!!!
Try to avoid surprises!!!
What worked last time may not work
this time!
Intro-‹#› - DMZ - 3/27/2016
Surprises
 Invalid
Assumptions
 Individual effects don’t add linearly
 Consider effects one at a time, not the sum
» Coupling from multiple sources (lines)
» PS noise
» Reflection noise
» Component tolerances
» Temperature
» Process
» PC board noise
» Package noise
» Connectors
» Vias
Intro-‹#› - DMZ - 3/27/2016
More Surprises
 Multiple
backward xtalk coupling, increased V
 Unaccounted for parasitics
» Inductance
» Resistance
» Capacitance
» ESR, ESL
 Nonlinearities
» Series termination with Capacitance load
» Driver resistance when switching (1/4 wave)
Intro-‹#› - DMZ - 3/27/2016
Still More
 Transmission
line
 Manufacturer data sheets
 Stubs
 Split load and standing wave
 Layout rules/communication/slip-up
 Tolerances
 T-line traces, return currents
 R and C functions of frequency
Intro-‹#› - DMZ - 3/27/2016
Last of Surprises?
 System
cost
Vs. Design time
Vs. Manufacturing time
Vs. Reliability
 Metastability
Intro-‹#› - DMZ - 3/27/2016
Digital Systems Engineering
 Designing
systems that work by design, not by trial
and error, with reasonable cost (dollars, time, effort,
…). Using appropriate tools (analysis, simulation,
measurement) to insure correct operation. Avoiding
surprises.
 Question: which is better?
» Guess/Estimate?
» Analysis (e.g. equations and calculation)?
» Simulation (e.g. HSPICE)?
» Measurement (e.g. oscilloscope, TDR)?
Intro-‹#› - DMZ - 3/27/2016