Page 103 - The Art and Science of Analog Circuit Design

One Trip Down the 1C Development Road

test engineer tests, the market sets the price, and life is a breeze. This
simple approach rarely develops an 1C that is really new; and the compa-
nies that work this way rarely make any money selling ICs,
Most successful 1C designers I know are very good circuit designers
and enjoy circuit design more than anything else at work. But it is not
just their circuit design skills that make these designers successful; it is
also their realization that all the steps in the development of an 1C must
be done properly. These designers do not work to a rigid set of specifica-
tions. They learn and understand what the 1C specs mean to the customer
and how the 1C specs affect the system performance. Successful 1C de-
signers take the time to do whatever it takes to make the best 1C they can.
This is quite different from the custom 1C designer who sells design.
If you are selling design, it is a disadvantage to beat the customer's spec
by too much. If you do the job too well, the customer will not need a
new custom 1C very soon. But if you just meet the requirement, then in
only a year or so the customer will be back for more. This kind of design
reminds me of the famous Russian weight lifter who set many world
records. For many years he was able to break his own world record by
lifting only a fraction of a kilogram more than the last time. He received
a bonus every time he set a new world record; his job was setting rec-
ords. He would be out of a job if he did the best he could every time; so
he only did as much as was required.

Product Definition

Where do we get the ideas for new products? From our customers, of
course. It is not easy, however. Most customers will tell you what they
want, because they are not sure what they need. Also, they do not know
what the different 1C technologies are capable of and what trade-offs
must be made to improve various areas of performance. The way ques-
tions are asked often determines the answers. Never say, "Would you like
feature XYZ?" Instead say, "What would feature XYZ be worth to you?"
When an 1C manufacturer asks a customer, it is often like a grandpar-
ent asking a grandchild. The child wants all the things that it cannot get
from its parents and knows none of the restrictions that bind the others.
The only thing worse would be to have a total stranger do the question-
ing. That may sound unlikely, but there are companies that have hired
non-technical people to ask customers what new products they want. At
best, this only results in a very humorous presentation that wastes a lot
of people's time.
Talking to customers, applications engineers, and salespeople gives
the clues and ideas to a designer for what products will be successful. It
is important to pick a product based on the market it will serve. Do not
make a new 1C because the circuit design is fun or easy. Remember that
circuit design is only a small part of the development process. The days
of designing a new function that has no specific market should be long

86