My first non-MIT job was with a start-up company called Micronetic Systems Inc, which we sometimes casually called "Micronetics". They built a mini-computer controlled production machine which used a laser beam to trim resistors on thick-film or thin-film hybrid microcircuits. I was hired to take charge of the software used to control the machine. I was very interested in getting some experience at a start-up company.
I've already posted a previous entry about an event that occurred at Micronetics, entry #0031, called Madhu and the death ray. In that entry, I went into some technical details on what the Micronetics machine did, and how its laser worked, in a section at the end called "Stuff for you techies".
Small companies tend to go from crisis to crisis, and solving these problems is exhilarating, in addition to being very hard work. Since by definition the staff is small, employees are called upon to do all sorts of work. So although my job description called for me to be in charge of software development, I also installed machines, went on service calls, and sometimes sales calls (the latter for technical support, since machines sometimes had to be modified for particular customers).
One of our early sales was to Motorola, to be installed in a manufacturing plant in Fort Lauderdale, Florida. These expensive machines were not exactly off-the-shelf items - each one was custom built for a particular buyer. We had some delivery problems while constructing the Motorola machine, and we politely asked our customer if we might deliver it a bit later than the originally promised date. He was furious at us for suggesting a delay, and insisted that the machine absolutely had to be delivered on time.
So we assembled all the major modules (the computer, the laser, the power supply, the RF supply for the laser's Q-switch, etc.), and put them together into what looked like a completed machine, even though in fact several of its parts did not work, due to missing internal components. Thus an apparently complete machine did arrive on the promised day. Although our shipper verified the arrival, we didn't hear anything from the customer about scheduling an installation date and operations course. And a week later, and then two weeks later, we still heard nothing. There was one thing we did know. The customer wasn't silent because the machine was working. We knew that it was not capable of operating.
Finally, we called to ask what was going on. At that point, the customer told us that the machine was sitting in a warehouse, still in its (very big) box, Note 1 because the manufacturing building it was to go into had not yet been completed. The customer had known that perfectly well when we had asked about delaying our shipment to him. But he had insisted on maintaining the original delivery date based on his belief in some perverted business principal - "When you're given a delivery date, NEVER allow the vendor to delay". In fact, he could have allowed us some extra time without any inconvenience to himself at all. This would not be the last thing this customer would do to make me think he was a bit of an idiot.
Eventually, a date was set up, and I was included in a small team sent down to Fort Lauderdale to "install" the machine. Note 2 By then we had all the parts, so upon arrival we set about finishing and testing all the modules, and getting the machine up and running. As far as the customer knew, we required a very long installation process. In fact, if the machine had really been complete when we had shipped it, we could have set it up in a day (align the laser and go).
In the midst of the so-called "installation", I was urgently summoned from Florida back to Boston, to give a sales demo to a potential customer. RCA wanted to use our machine to trim resistors on a hybrid ceramic circuit card to be used as part of a television tuner. Ordinarily, I would not be needed on a sales demo, but RCA had a particular problem. The ceramic card whose resistors were being trimmed was substantially larger than the average thick-film circuit card, meaning the resistors being trimmed were far apart ("far" in this context meaning perhaps five centimeters, about two inches). We positioned our laser beam by moving two mirrors mounted on an X-Y table, driving the two stages using stepping motors (sometimes called "stepper motors"). This type of motor moves a fixed distance for every pulse you send it, so that you always know its exact position. But the large size of this particular circuit caused us to take too long to re-position the laser beam between resistor trimming operations - we took more time to process the circuit than our competitor's machine.
I was asked if I could somehow make the machine reposition faster without any hardware changes, just by changing the software. Sure, I replied. We were driving the stepping motors with a train of pulses at a constant rate, the fastest rate at which the motors could start running from a standing start without missing a step. But if I wrote software to start the motors more slowly, and then increased the pulse rate gradually, I imagined I could quickly ramp the motors up to a substantially higher speed.
It was a Friday, and the customer was scheduled to fly to Boston the following Monday, in order to observe the machine in operation, and measure the time it took to trim one of their circuits. And we knew our existing machine was not fast enough to get the sale. Could we use an un-tested technique to speed up the machine, and do it in one weekend?
Hey, that's what I do for a living - except, I was in Florida. No problem. I re-wrote our laser beam control program on the flight from Fort Lauderdale to Boston, entered it into the assembler upon arrival, debugged it, tested it, and tweaked it over the weekend, and helped with the sales demo on Monday. We got the sale, and I returned to Florida that evening to help complete our "installation".
Of course, not all crises have a happy outcome. The weekend after the RCA demo, the company tried for a repeat performance with another potential customer. This time, however, our only demo machine refused to cooperate; as soon as we fixed one part, another would break down. The prospective buyers, who had flown several people thousands of miles to see the demonstration, were getting quite impatient hearing our sales staff regale them with grandiose promises. Finally our expensive laser Q-switch broke, and it was the only one we had.
"What are we going to do now, Larry?" the president moaned to me in the basement of the building, kicking a hole in the wall for emphasis (the building had a flooding problem, and the wallboard was a little moldy).
"Why don't we give up?", I said brightly.
The president turned bright red, and glared at me viciously; I could almost imagine smoke coming out of his ears. "We don't give up at Micronetics, Larry", he said. But of course, there was really nothing much else we could do. We lost that particular sale.
I discovered some things about Florida on the "installation" trip I described above. I arrived in Fort Lauderdale to find our machine had been brought in from the warehouse, and set up in one side of a single-floor manufacturing building that was still under construction - the other side was getting longer every day. Construction in Florida seemed to be a lot simpler than construction in the Boston area. Bulldozers just scraped away the dirt down to something that looked like coral, and the building was set down on top of rows of masonry blocks - I don't recall any poured concrete foundation. It all went up in a flash.
I came in one day to find a back-hoe digging a hole near the building, and I asked the operator what it was for. He told me that we were near what would become the main entrance to the building, and the hole was to be a decorative pool (I noticed some large decorative boulders had been set nearby as well). I asked him how the pool would be filled, and he looked at me as if I were crazy. He said, "Clearly, you're not from Florida - come back tomorrow." I did, and the next day, the hole was full of water. The ground water in southern Florida is not very far below the surface. Dig a hole, and it will fill up.
When I talked about the ground water with people in Florida, I was told what it takes to build an underground swimming pool there. After all, if you dig a hole for the pool, the hole quickly fills with water. So how can you then build the pool?
Here's how. Pipes are placed in the ground around the site of the pool, and connected to pumps, which are used to pump out the ground water. This allows a dry hole to be dug, in the shape of the pool-to-be. Steel "rebars" ("reinforcing bars") are assembled, and the concrete walls of the pool are sprayed over them. The concrete is then smoothed and allowed to harden. All the while, the pumps need to keep running to keep the ground water at bay. In fact, more than one pump and a backup generator are generally used, in case the main pump fails or the power goes out. The pumps cannot be allowed to stop.
Eventually, the concrete hardens, and the pool is done. But even at that point, the pumps cannot be shut off. If they are, the newly constructed pool will float up out of its hole in the ground, and it will crack into pieces. The concrete must be allowed some further drying time, and then it can be painted, and the paint can be allowed to dry. All the while, the pumps must keep running. Finally, when the paint has fully dried, the new pool can be filled with water. Only then can the pumps finally be turned off.
The pipes, by the way, are left in the ground. Whenever the pool needs to be re-painted, every few years, pumps are brought back and connected to the same pipes. Only then can the pool be drained for re-painting.
Florida is a strange state. Enjoy it while you can - the entire state south of Tallahassee is less than three meters (ten feet) above sea level. When global warming melts portions of our polar glaciers and causes the sea level to rise, much of the state will be lost to the Atlantic. I've always wondered when Florida real estate values will start to drop, as its ultimate fate comes more clearly into view. Bye-bye, Disney World. Of course, this will take a couple of hundred years. Fortunately, I'll be dead by then.
Note 1: I don't remember the exact size of the machine, but it was probably about 2 meters (6.5 feet) wide, and about a meter (39 inches) high and deep. It included a rotating table to carry the ceramic circuit card (often only about 2 to 3 cm. square) into the machine for trimming. The operator would place a circuit down on the table, pressing it against an L-shaped guide, where it would be held down by a vacuum. The operator would then push two buttons on the table, on either side of the turntable, to verify that his or her hands were out of the way when the table rotated the part into the machine, into position under the laser beam. The trimming operation could be seen through a window in the machine's cover, which had a yellow window that allowed visible light to pass, but completely blocked any stray infra-red light reflected out of the laser beam (wavelength 1064 nanometers).
In discussing the operation of the turntable with the firm's mechanical engineer, I discovered that machinery operators really do need to be protected from themselves. The electrical circuit to prevent the turntable from turning unless both safety buttons were pressed seemed to me to be needlessly complex. I pointed out that all you needed to do was to power the turntable through both buttons wired in series - it could then not move unless both buttons were depressed. Well, yes, came the answer, but then the operator would jam something into the buttons to just hold them down, defeating the safety feature.
Oops. OK, I countered, so wire the buttons in series, but include a circuit that verified that they were released and re-pressed before the next machine cycle. Not good enough, I was told, the workers would then jam one of the buttons down, requiring the other to be released and re-pressed, but leaving one hand free (to be caught in the turntable as it rotated). A separate circuit was needed on each button individually to verify that it had been released and re-pressed between operations.
The designers of all machine safety interlocks know this has to be done. To increase their speed (and hence their pay), almost all workers will try to defeat interlock systems, even if they are feeding a five-ton press that can mash one or both of their hands into pulp. [return to text]
Note 2: Sounds like a great trip, right? Down to Fort Lauderdale, Florida, and it was in the winter, when Boston was snowy and cold. The only problem was, I was working like a dog. Putting up employees on a business trip is very expensive, and the company wants you to get the job done and come home as soon as possible. I've had a few really interesting business trips in my career, but only a few. Most of them are pure drudgery, although for an engineer who usually stays at home, they do provide a break in the routine. For sales people who are on the road all the time, they can be hell.
Determined to get some sun during my visit to Fort Lauderdale, I woke up extra early one morning, and walked down to the beach for a swim. The others complained when I showed up a bit late for breakfast. [return to text]