I worked for SMC Pneumatics Corp. in Tustin, CA and Tskuba, Japan from October of 1995 to August of 1997, developing specialized chillers and heat exchangers for the semiconductor industry. I was also the liaison between the engineers in Japan and various customers in Texas, with a heavy emphasis on solving problems caused by cultural differences.
Here is a description of one of the products I designed: a semiconductor manufacturer had a system that contained a plate that was electrically heated to 400 degrees C., then rapidly cooled to 100 degrees C. by running water through internal passages in the plate. As the water flashed into steam, the plate quickly cooled down to the boiling point of water. In order to avoid buildup of minerals inside the passages, ultra-pure water was used for the cooling. The problem was that this used too much of the expensive ultra-pure water. My job was to design a system that would conserve the ultra-pure water.
I sketched out a general design consisting of a tank of ultra-pure water that would be used again and again to cool the plate, a heat exchanger inside the tank that dumped the heat to the plant's recirculating (but not ultra-pure) cooling water system, an overflow to the drain, and a valve-controlled ultra-pure water supply to the tank. A resistivity meter and a level gauge controlled the flow of ultra-pure water into the tank; if the tank got too low or the water became too conductive, additional ultra-pure water would be added, and the old water would flow into the drain. This reduced the amount of ultra-pure water used by a factor of over one hundred.
Once the basic design was approved by the customer, I started on the detailed design. The engineering department in Japan also started a design, giving the project to one of their best (which in Japan means oldest) mechanical engineers. There were some interesting corporate politics involved; the engineers in Japan did not want to lose design work to the US, while the board of directors were actively hiring engineers and staffing design centers in San Francisco and in Los Angeles. If this had been a US company, the board of directors would have won, but in Japanese companies top management is weaker, and there was a real question as to which faction would prevail. At the same time that I was in fierce competition with the engineers in Japan over who would design heat exchangers, I had a very good relationship with the same engineers as the US technical representative for their Carnot cycle chiller line. This is typical of Japanese culture; both the competition and the cooperation were considered to be good for the company.
It is interesting to compare the differences in the two completed designs. My design was built into an off-the shelf thick-wall 55 Gallon polypropylene drum, with all of the controls and fluid/electrical connections through the lid, spaced so that the piping to the customer's equipment was all straight runs. The design from Japan was a welded stainless-steel cube with the controls on the front and the fluid/electrical connections on the rear, necessitating several bends in the pipes. Both were based on my original design - it would have been interesting to see what they would have came up with by themselves. The most striking difference was in the electronics. The Japanese design consisted of a commercial resistivity meter and rows of mechanical relays. My design was based on a Parallax Corp. Basic Stamp with a back-lit LCD screen and a touch keypad, with the Basic Stamp measuring the resistivity of the water. The difference was dramatic; the Japanese design took up almost two cubic feet for the internal electronics, while my electronics were the size of a postage stamp. My design also cost half as much to build (one tenth if you only consider the electronics), and I finished the design with a staff of three in far less time than the Japanese team could with a staff of twelve.
One interesting technical challenge was measuring the resistivity of the water. 100% pure water has a bulk resistivity of around 18 Megohms per cubic centimeter. This resistivity is also strongly affected by the temperature of the water. A Basic stamp has the ability to measure resistance by measuring the rise-time of a digital pulse through an RC circuit, but the range it measures is usually only a few Kilohms, not the Megohms I needed to measure. I managed to get the range up to 250 Kilohms by careful capacitor selection, but the impedance of the Basic Stamp pins wouldn't allow me to go any higher than that. I solved this problem by designing a parallel plate sensor with an area of 200 square centimeters and a distance between plates of one centimeter. This converted a bulk resistivity of 18 Megohms per cubic centimeter to a resistance of 90 Kilohms - well within the capability of the Basic Stamp. I used the same Basic Stamp to measure a thermistor to detect the water temperature and wrote a calibration table in software to compensate for temperature and sensor nonlinearity. The Basic Stamp has limited memory, so I wrote the table with 11 temperature values from 0 degrees C to 100 degrees C in 10 degree steps and 21 resistivity values from 0 ohms to 20 Megohms in 1 Megohm steps. From these values, I extrapolated 1 degree and 0.1 Megohm steps, which were displayed on the LCD. The same sensor also served as a water level detector; when the tank level got too low, there was air between the plates instead of water and the open sensor reading would trigger the addition of more ultra-pure water. I wrote code so that if the water stayed on for too long without filling the tank it would give up and display an error. I used the same scheme so that if the water stayed on for too long without increasing the resistivity to an acceptable level, it would give up and display an error.
Another interesting challenge occurred when the prototype started acting strangely. Over the course of several hours, the bulk resistivity of the ultra-pure water as measured by the Basic Stamp would rise above the theoretical maximum of 18 Megohms per cubic centimeter, while the commercially available resistivity meter would show no change. Eventually the Basic Stamp software would erroneously conclude that the water level was below the plates and would add ultra-pure water. As soon as a small amount of additional ultra-pure water was added, the reading would immediately go back to normal. I found that stirring the water made the problem go away - it only happened when the water was very still. It turns out that the ions in the ultra-pure water were moving away from one plate and collecting on the other plate. This is where the flexibility of the Basic Stamp really came in handy; I programmed it to swap which plate was grounded and which plate was exited at each reading. At this point, the sensing system tracked the commercial meter perfectly.
I left SMC Pneumatics when the faction that wanted all design work to be done in Japan prevailed over the board of directors. The customer in Texas was quite unhappy with the corporate decision to only offer him the stainless steel version at twice the price, and made his displeasure known by canceling the project if he couldn't get the lower-cost polypropylene version. I offered my resignation to the board of directors in order to save them further embarrassment.