Oral History – Hans Camenzind
Continued)
|
I read your article in the
IEEE Spectrum and noticed you provided a detailed description of the design
process you used for the 555. How
has the IC design process changed since then?
Incredible! The design process itself for the 555
took a year, from start to prototypes and small production quantities. Now, it was right the first time, but
that was just luck. I have gone
through so many projects where you have teams of people, two people each
with one looking at the circuit diagram and the other checking back and
forth maybe three teams checking in succession, and there were still
mistakes. So, in this case, I had
to do it all myself. I was a one man design house, and I got it right the
first time. I think that was luck – I may have spent extra time doing it
because it was my project. There
was no way at that time to make absolutely sure you had it right.
In the IEEE Article, you mentioned that there was a
design review at Signetics. Was that a contractual obligation?
No. I just thought it was a good idea. I’ve always had design reviews for all
my designs. Get feedback, bounce
off ideas with colleagues. But,
just how laborious the design was.
First, the breadboarding of the circuit. The one thing you want to do is make sure the circuit works
well in production, and in production you have parameter variation - high
gain, low gain for transistors, high resistor values, low resistor values,
all sorts of combinations.
|
|
Oral History – Hans Camenzind
(Continued)
|
So, in the breadboard, all
you can do is build the first one and it works. And then you start plugging in other components. You know, maybe ten different sets, at
least, to make sure you get the variations. You still can’t be sure you get all of the variations. This is what I did with the breadboard,
and this just takes months.
If you didn’t do this well,
there might be a bad outcome in production?
Matter of fact, that was the
rule, rather than the exception.
You see, there are so many lazy designers that build a breadboard,
or nowadays, do a simulation, and not worry about the variations. A circuit may work fine the first time
(they were lucky), and then in production they have problems, yield
loss. And that’s why you see so
many revision letters. I have circuits that go to “M” in
production, afterwards, for the first
two or three years, with revision after revision.
The second stage is where you
have a circuit diagram and you’re pretty confident it will work. Now you have to make a layout. The only choice I had was to buy a large
drafting board, and just draw it out. You know, you want to shift the
components around so you have a minimum total area. You may make ten
different drawings. Another few
months gone!
Go
To Camenzind Oral History,Page 8
|
|