life, by the way. I’d never used or seen one. So I didn’t know the first thing
about them. I’d never taken a course in floppy controllers, I’d never seen a
floppy controller, I didn’t know what they did. But I knew on a cassette tape, I
generated signals of certain timing patterns and, when they came back from the
cassette tape, I analyzed them to figure out what were the ones and what were
the zeros. The microprocessor did the timing, because the timing was loose; it
wasn’t in fractions of a microsecond. I just wrote programs that waited a certain
amount of time and saw when the signal went from high to low or low to high,
and made decisions right in the microprocessor of our Apple II. But I couldn’t
do that on the floppy disk. So I looked at Shugart’s design to figure out how it
worked. And I figured out, oh, you put some data here and some signals here
and you set a clock bit at a certain speed every 4 microseconds, and you shipped
in some new data. I went through chip after chip after chip on theirs, and I said,
“If I take all these out, it’s just as easy for me to run the wires straight over to the
Steve Wozniak 53
head that’s writing onto the disk. And the signal coming back from it, I just run
a wire over to my controller and I just do all the timing here and I don’t need all
their complicated interface to work.” So I took 20 chips off their board; I
bypassed 20 of their chips.
Steve Jobs really liked this because, when it came negotiation time, he said,
“That’s a good reason to sell it to us at a lower price. We don’t need your controller
board. All we need is a little bit of it. So you can sell it to us cheaper than
you are selling it to other people.” It was a good deal for Shugart, a good deal
for Apple.
I thought I could write some data onto a floppy disk and interpret what was
coming back as ones and zeros. Here’s the problem: you got a whole big track of
data and there’s thousands and thousands of ones and zeros and then the track
repeats. The head goes around and around. You have to know where and when
data starts and stops. And that was an issue I’d never done in my life. I came up
with an approach of writing a certain kind of data, a certain pattern—AA D5 AA
55—some pattern like that. I just wrote it for a long enough sequence at the
start of every section of data, and it was something that would somehow get my
circuits into sync so they knew when a one and a zero started a byte, instead of
was in the middle of a byte. It just automatically caused it to just sort of slip into