In the early days of commercial satellite operations, the limitations of the satellites meant that, even with a 30-meter dish, you needed to be able to push several kilowatts up to the bird to carry a decent volume of traffic. Our first Earth stations used a pair of 10kW travelling wave tube transmitters that were very expensive, and each occupied two huge cabinets. They were superb pieces of kit and, once commissioned, were very reliable.
The huge cabinets accommodated the TWT and its massive magnet assembly, the 18kV 3A collector power supply, and a complicated water cooling system and heat exchanger. Because parts of the water circuit were at very high potential, it was essential to maintain exceptional purity and hence extremely low conductivity. The system was carefully designed to achieve this and incorporated a water conductivity monitor. Once filled with top-grade deionized water, it would normally work very well. We tend to think that water and electricity at around 20kV don't mix, but the designers had the problem licked in this case.
However, by 1969, the advent of the new generation of satellites with de-spun antenna systems and improved solar cells (Intelsat 3 onwards) meant that these powerful transmitters were usually operated well under maximum power. A decision was made to design a smaller and cheaper version based on a 2kW klystron for future stations.
The klystron was also water cooled, so the unit incorporated most of the same cooling system but was repackaged to fit in one smaller cabinet. So it was with great interest that we installed the first pair of these new pocket-sized transmitters in the tower of the Earth station in East Africa in 1970. In contrast with previous stations, the transmitter floor looked empty, but the transmitters performed well straight out of the packing cases, and the station went on air as planned after the customary shakedown period.
After a few weeks, things began to go wrong with both transmitters. The water conductivity monitors began to show an inexorable and ever-more rapid rise in conductivity levels. Flushing and replacing with fresh deionized water was only a temporary cure, and eventually we could see that the water flowing through the glass flow meters was no longer clear -- it had a distinct grey-brown hue. Gradually, the coolant pressure started to rise, and the flow rate through the klystrons began to fall, which was a sure sign of blockage within the klystrons themselves.
With the station already carrying traffic, it was essential to maintain at least one transmitter on air at any one time. So the transmitter team sent off some of the grey-brown liquid to be analyzed by a local laboratory and devised a program of regular flushing to manage the situation until the designers back in the UK could figure out what was going on. We couldn't risk dismantling the cooling systems on site until we were sure we had the answer. Otherwise, we'd just have to go completely off air at huge commercial cost.
The analysis results made interesting reading. Tiny traces of copper, tin, and zinc were not too surprising, as the metal parts of the system were made of copper, brass, bronze, and stainless steel. The big surprise was the presence of plenty of iron and cadmium. Wherever the iron was coming from, it probably wasn't from stainless components, because we might then expect also to see nickel or chromium. No internal part of the coolant system was supposed to use mild steel, although we did make extensive use of cadmium-plated steel fasteners elsewhere. Clearly, there was a serious mistake somewhere if the coolant was coming into contact with such materials, but with no leaks and a supposedly well-tried design, it was a puzzle to see how it could have happened.