Pang: How did the labor divide up within this group?
Sachs: I was the electromechanical person and the systems person. Dean was the project lead. Jim was the mechanical engineer. We were the main part of the team. David Kelley had more of the user interface, figuring out what would this look like and feel like. Rickson Sun provided some electrical and optical consulting.
One of the first things we decided was to attack the failure points. The Xerox laboratory instrument was an over-constrained problem. It was a bunch of ball bearings, held by ball bearings, rolling on a slippery table, so the slightest amount of dirt would jam up the whole thing and it would stop working. It was made of machined aluminum blocks screwed together, and you'd have to completely disassemble it to fix it; and chances were, you'd introduce more dirt fixing it than you could remove in fixing it.
So we said, "We've got to unconstrain it and have the ball more loosely held. It can't be a tightly constrained ball." So we created a mechanism that had a ball that could roll on a table, and we needed to have an x/y encoding scheme to translate the motion of the mouse into cursor movement-- which is what Xerox did too. But we let the ball float, so that dust wouldn't hurt it, and soft rollers, so dirt could roll through them and not get pinched.
The second constraint was they used a mechanical system of a wheel rubbing against metal wiper fingers, which would break and were electrically noisy. At the University of Michigan, I had done a project in an electrical engineering class making an optical encoder, which used the characteristic of quadrature, where you have two optical devices looking at a slotted disk or strip, and could determine motion and position from this setup. I decided we could make a rotary encoder, and have optical components looking at wheels with slots in them, and do it without any mechanical parts rotating up against one another.
(May 1980 prototype)
So the very first prototype had a floating ball, it had optical encoders, we built the circuit, and we were able to roll it around on the table, and the oscilloscope would show signals on the screen. We said, "We're done! It only took two months, and we're done!" Of course, there was a lot more to be done.
It turns out one of the next principles-- that resulted in the patent-- was to have a spring-loaded roller holding the ball against the x/y encoders. If you look at any mouse today, you'll see a spring-loaded roller holding the ball against the rollers, and that was something fundamental that we patented that nobody had ever thought of before. I believe it was Rickson Sun who said that component should be spring-loaded, and that it would unconstrain the mechanical system.
We had solved a number of problems, but we had created something that required such precision it probably couldn't be mass-produced. Our team member Jim Yurchenco said, one way to solve that using known technology was injection molded plastic. An injection molded part can be made repeatedly, with sub-thousandths of an inch tolerance, with respect to itself. So if we made a single piece of plastic, and located all the components within that piece of plastic, then you wouldn't have any alignment problems with the components over here stuck to the components over there.
(Drawing of ribcage)
And so we designed-- and this is also a fundamental part of the patent-- a single unit which holds the x/y encoders and rollers and bearings. We eliminated the real ball bearings and used plastic bearings. It also made the mouse easily assembled by untrained people: you didn't even need to screw anything together, everything would snap together.
(Ribcage drawing from patent)
That turned out to be the linchpin. Through optical encoders, through a spring-loaded third roller, and through a unified cage to hold all these parts, we came up with something that made a mouse mass-producible, reliable, and inexpensive, and patented it. And hundreds of millions of them have been made.
Pang: It sounds as if you were working to a level of precision in the manufacturing that went beyond what computer manufacturers normally had to worry about. Ordinary computer cases didn't have to have this kind of absolute precision and level of accuracy.
Sachs: We were at the intersection of technologies that weren't commonly combined. Precision electronics had been made, and if you needed it to be extra reliable you could have military spec electronics, which were expensive; and you could have inexpensive electronics that didn't have tight tolerances. On the mechanical side, you could have very tight tolerances mechanically in a laboratory instrument, and it would be very expensive; but if it was inexpensive it was sloppy. So we needed to combine all of these, and be inexpensive yet have the performance of high mechanical and electrical tolerance-- which was not anything that you could buy on the market.
The mouse may have been one of the first devices had that had this unusual combination of being a mass-produced, low-cost product, but delivering high electrical and mechanical tolerance. We came up with high electrical and mechanical tolerance by essentially canceling out all the variables. It was all digital, and it was all a single piece. That made it unusual.