Brrr! Transmitter failure in the Arctic chill

4/21/2011 9:30 AM EDT

Fairbanks, Alaska has a very extreme climate. I've been in 90 degrees in the summer and -64 degrees in the winter. The Alaskan environment offers unique engineering challenges and Professional Engineers, in addition to passing the PE exam, must take a graduate level Arctic Engineering course or write a research paper on Arctic engineering.

In 1974 I was an electrical engineer freshly graduated from the University of Alaska. I worked on a NASA project for the ATS-6 experimental satellite. The project included broadcasting television to villages around Alaska for health, education, and news. The studio was a normal 1970s TV studio. It was at the University of Alaska Fairbanks in the basement of one building. The transmit and receive antennas were ten foot S-band dishes. They sat like Mickey Mouse ears on the top of another, taller building. On Tuesday nights we broadcast a show about Alaska news from Fairbanks. This particular night we turned on the satellite system to broadcast the show and ... nothing. Was the receiver or the transmitter the problem? We had an audio link on another satellite and our operator asked if anyone could see the signal - no one could.

The antennas could be adjusted to compensate for the orbital variations but our remote control adjustments had no effect. It was a cold, cold night so we thought the motor drive might be having problems; lubricants can get pretty stiff at forty below. With another engineer, I raced to the antennas to see if that was the problem. We might be able to adjust the antenna manually. When the winter cold settles in at Fairbanks, the air usually gets very still. But that night there was a very cold wind. When we came out on the roof the temperature change due to the wind shattered my friend's glasses. A tough Alaskan, he insisted he was ok and we went on. (Later on he was checked out and really was fine.)

The motor was working so that wasn't the problem. We checked the outdoor unit of the transmitter and it sure looked dead. There was a heatsink that felt as cold as the rest of the metal up there. I thought the unit had failed and we would be off the air until we could repair it or get a replacement. So much for our program that night.

The next day we couldn't find anything wrong. We were able to operate the system for another broadcast during the day. It dawned on us that the extreme cold that night must have made the transmitter unit stop working. We hauled it into the shop and had a look. The designers had anticipated cold weather and had provided heaters to keep the transistors at an operational temperature.

Unfortunately they didn't anticipate that much cold. Power transistors were attached to the heatsink to dissipate heat during normal operation, of course. But the heat sources were just power resistors bolted to the same heatsink and controlled by a temperature sensor. The wind pulled the heat off the heatsink faster than the resistors could add it so the transistors just stopped working.

We had to fix the problem soon. Since we were halfway through the one-year project, it didn't seem likely that any manufacturing changes would happen in time. My solution was to build an insulated box to cover the transmitter to give it a little extra help during the cold arctic winter. We had no more problems that winter. When it stayed above -20 I simply removed the insulated box.

Author Mark Espe currently writes software for secure communication and has been involved with hardware and software for broadcasting, telecommunication, and computer systems.