Patents - issued and otherwise


W. A. Barrett, San Jose, CA
Email Barrett
I have six issued patents in my name, as the inventor. Three were issued while I was a staff member at Bell Labs in Murray Hill.
The oldest one, 3,067,408, was actually manufactured and installed into one of the first solid state digital telephone switching systems of AT&T, and used for (I think) about a decade before being replaced by a more modern variation of the thing. Its idea was actually very simple - carry a bit on a magnetized wire, switch it back and forth with a certain electric current's magnetic field. But also place a stronger permanent magnet near it so that it could be switched back. An array of these formed a semi-permanent memory, and it was used to hold an array of telephone numbers to be mapped into circuit closures. The permanent magnets were embedded on a sheet to be slid out to change their state, and that was done when a phone number was changed.

Two patents were issued for a magnetic analog-to-digital encoder, 3,187,324, and 3,191,168.
The idea here was to come up with a reasonably low cost, but very fast device to convert a variable current into its digital equivalent. "Very fast" in 1965 meant something that could convert 10 million levels per second into something like an 8 bit digit code, that is, convert at a 10 Mbyte rate. Today, it's done with tiny semiconductors on a chip running more than 200 times faster than that. To my knowledge, this contraption was never manufactured, but I got a prototype to work at that speed in the lab. And I was given a special award for its ingenuity at a solid-state circuits conference the next year.

The last three patents were issued for ideas developed at Lasa Industries. All three were implemented on the system under development. Unfortunately, the company collapsed without selling a single machine. It died for various process reasons, but not for lack of excellent software, mechanical, optical and electronic engineering.
All three depending on a unique machine that could move around a focussed laser beam, control its intensity, and sense light reflected from it under computer control. 4,978,841 describes a way of using the reflected light to move across a chip edge, then looking at the reflected signal. The chip could be mechanically moved and oriented, so the trick was to work out which way to orient it in order to obtain the sharpest transition from dark to light as the beam moved across an edge. 5,315,111 and 5,338,924 describe two different approaches to focusing the laser beam. This requires moving the chip up or down by submicron steps until the beam image is as tiny as possible. This was vital in order to achieve a controlled temperature at the focal point. Of these two, I personally liked the 111 patent, since it made use of a set of optical fringes that showed up through multiple reflections in one of the clear plates of the system. The patent contains a theoretical treatment of the fringes, which are known to physicists as Brewster's fringes.

Would-be Patents


A would-be patent is something that reads, smells and tastes like a patent, but isn't. The description is there, there are often claims, the engineering is pretty good, but I eventually realized that it had some fatal flaw, and just not worth paying for an expensive patent. Even a simple patent today will cost you at least $10,000 in attorney fees. And, after paying that, you may discover that your idea is covered by some other senior patent, or is considered a known process. Or, it would be almost impossible for a small inventor to put it into practice.
I've ground out several dozen of these things that have never reached the light. Judge for yourself whether you would be willing to invest the thousands - if not millions - of dollar to put one of them into practice.

A Supermarket Price Indicator System

.
This was an idea way ahead of its time - replace the little paper tags showing the price of something on the shelf of a supermarket with an active LED device that could be automatically controlled for specials or a price increase, etc. through the store software system.
See supermarket.
I decided against patenting it. Look at the year - 1991. It needed several more inventions and technological advances to have a prayer of working, and being made cheaply enough. It took almost 20 years to develop the cheap, tiny displays and low-power microprocessors to make this work. In fact, I did notice something like it installed in a store about five years ago. But my patent would have expired in 2009.

A Scoring System for Schools


I spent a lot of time writing up this patent, which includes some nice diagrams.
I was concerned about the fact that so many schools in low-income school districts could simply not afford to equip some classrooms, or even a lab, with computers for the students to use. Even when the manufacturer, like Apple, offered them free, and even installed a room full for a school, the school often discovered that maintaining them was expensive, and there were usually no teachers with the time and experience to keep them working.
A secondary concern was finding ways to help a school teacher with grading lots of papers and tests churned out by her pupils.
Maintenance is also a problem for public universities. I was asked at SJSU to maintain a lab of Unix systems, and I managed to do it for several years, through at least three complete model changes (HP, then IBM's AIX, then Sun workstations). It took a lot of work, plus a budge to replace stolen or damaged mice, keyboards, and monitors. A small school in a rural setting wouldn't have a chance.
So what to do? Well, this is a compromise. The students don't use a computer directly, but the school needs to provide a networked system for its teachers. A special digital scanner-printer is needed, to provide a way for a busy teacher to assign homework and have someone else grade it and mark it up.
So why didn't this make the cut? Several reasons - the price of small desktops kept falling, and now almost every kid has a cellphone. It also depended on some printer company (like HP) to make a special printer that can a header in its paper feed, then acquire what is to be printed from a computer server. No one makes a printer like that, but maybe should.

These are just two samples of would-be patents. I have several books full of such worked-out ideas. Back in the late 1970s I was interested in finding ways of automating slide projector shows, coordinating it with an audio track. Producing a movie meant a lot of painstaking hours on a manually operated film editor - why couldn't that be automated?
I was later frustrated by the huge parking garages at San Jose State, and the way everyone had to drive through carefully checking either side for an empty place, only to find out the thing was completely full. Such a waste of time and gas! Surely there had to be some way of telling a driver the thing was full, or where to reserve and find a parking spot if there was one. This is still an unsolved problem, although more garages now at least keep track of the number of unfilled spaces by counting the entry and exit of each car. Many have a large sign outside announcing that there are XX spaces for parking here. But it's still often hard to find one of those spaces.