ASCII Version of "Ernies Story"
"Ernie's Story" was first published in Electronic Design Magazine, December 16, 1996" 

(Reprinted with Permission) 


Dr. Howard Johnson is president of Signal Consulting Inc., a high-technology consulting firm specializing in solving high-speed digital design problems. He regularly presents technical workshops for digital engineers, including courses for Oxford University and UC-Berkeley. 

He is the author of "High-Speed Digital Design: A Handbook of Black Magic" (Prentice-Hall, 1993) ISBN: 0-13-395724-1. He can be reached at (806) 556-0800; fax (206) 881-6149; e-mail <howiej@sigcon.com>. 

Signal Consulting The Art of High-Speed Digtial Design    http://www.sigcon.com

 
 
 Ernie fidgeted uncomfortably
on the big leather sofa, listening to 
the others drone on. His tie was 
tight. He wasn't often invited into 
the corner office, and he knew why. 
It's because I'm not like the others, 
and because I can't keep covering 
things up....
   Finally, Ernie couldn't stand it 
any longer. uot;Jim, you're the presi-
dent, and you've got to know," he 
blurted. "There's some sort of 
glitch on the main processor board. 
It might be a crosstalk effect or a 
noise problem. We just don't know.
We'll have to hold up shipments
until we get it sorted out." 
Jim turned white, then red. The 
veins began to stand out on his
forehead. He hesitated, and then 
stood up and thundered, "So that's
what you guys are trying to tell 
me?" He was used to problems, but 
this was absolutely unbelievable. 
"Let's see, gentlemen. We have 
145 manufacturing people, 35 tech-
nicians, and a warehouse staff wait-
ing for this product. Our sales team 
is trained, advertising is in the 
works, and I have four press inter-
views starting tomorrow. The comp-
any burn rate is about $100,000 
per day. If this problem takes a 
week, we lose five hundred grand. 
That's more than you're worth, 
Ernie. Why didn't you tell me about
this before!! How could we get this 
far and not know there was a prob-
lem???" 
   Ernie's lips trembled. He tried to 
make everyone understand. "Hey, 
look, I did everything I could. The 
simulator said everything was per-
fect. The timing was all checked 
out. How was I supposed to know 
we'd have crosstalk." How was I 
supposed to know that fast signals 
wouldn't go through that con-
nector?" 
"They don't even teach crosstalk
and ringing in college. Those are
the subjects we're supposed to
learn through experience. That's
just the way it works."
   It's cruel how our educational sys-
tem has failed us. Twenty years 
ago, all computer designers went 
through a common electrical engi-
neering curriculum. This included
basic analog circuits, transmission 
I lines, and linear systems theory. 
Trouble is, the computer hardware 
of that era was so slow that few 
people needed to know anything 
I about analog circuits to make their 
systems function. Take a look in the 
first edition of Texas Instruments 
logic catalog (if you can still get a 
hold of one). The typical LS-TTL 
logic gate had a rise/fall time of 
about 20 ns. That's very slow by to-
day's standard. It doesn't take an 
analog guru to plug together a
lot of LS-TTL logic. 
   As a result, colleges and univer-
sities felt they could safely drop the 
old analog curriculum requirements, 
in favor of newer, more modern 
computer science classes. And, 
there were even those who said that 
analog design was becoming ir-
relevant. 
   Not that I have anything against 
computer science. It's a wonderful 
discipline. Without it we wouldn't
have the processor sophistication
we now have, or have nearly as many
people trained to work with 
complex computer architectures. 
I can't blame the educators for 
their decisions. A modern univer-
sity has a lot on its plate. There are 
too many subjects to teach, and too 
few hours available to teach them.
 Today, modern processors are
clocked at over 100 MHz. Plain 
vanilla chip outputs sport sub-
nanosecond rise times. Analog ef-
fects are beginning to dominate 
system design. Digital designers 
without basic analog training
operate at a serious disadvantage. 
In reaction to this trend, some col-
leges and universities are reintro-
ducing analog material in their 
computer science curriculum. I par-
ticularly applaud Clayton Paul at 
the University of Kentucky and 
Henry Ott of Bell Labs for educat-
ing digital engineers about the 
tough new world of electromagnetic 
compatibility (EMC) standards. 
   Other courses of importance to 
digital engineers include basic cir-
cuit theory, ELM waves, and trans-
mission lines. A well-rounded dig-
ital engineer should understand
enough circuit theory to recognize 
the ways two adjacent digital sig-
nals will behave like a loosely cou-
pled, single-turn transformer. He 
or she should know enough E&M 
wave theory to understand why a 
ground plane is better for high-
speed boards than a loose mesh of
ground traces. All digital engineers 
should have a grasp of signal propa-
gation, superposition, and reflec-
tions on transmission lines. 
   A proper understanding of basic 
signal integrity and EMC is crucial 
to the continuing evolution of dig-
ital technology. Engineers with a 
solid background in signal integrity 
and EMC EE will greatly enhance their 
chances for career success.

Dr. Howard Johnson is president of 
Signal Consulting Inc., a high-tech-
nologyconsulting firm specializing 
in solving high-speed digital design 
problems. He regularly presents 
technical workshops for digital en-
gineers, including courses for 
Oxford University and UC-
Berkeley. He is the author of 
"High-Speed Digital Design: A 
Handbook of Black Magic" 
(Prentice-Hall, 1998). He can be 
reached at (806) 556-0800; 
fax (206) 881-6149; e-mail 
<howiej@sigcon.com>.

 

 

 
 



TOP