Lined up on the ground floor were the transmitters,
six pairs of 18 inch diameter klystron tubes, manufactured by
Litton Corp., mounted in oil-filled transformers and extending
up through the main floor some 10 feet. Each klystron tube would
provide an average 2.5 megawatts of radar energy on the same frequency
band as 2-meter amateur radio. Each pair of these klystron tubes
was joined by waveguide nearly a yard wide and 18 inches deep,
big enough for a small adult to crawl through. These pairs of
transmitter tubes were routed to waveguide switches which routed
the radar energy out of the building to the scanners.
The waveguide
switches allowed any of the transmitter pairs to be switched into
either of the adjacent scanners and allowed transmitters to be
switched in or out as maintenance requirements or equipment failures
dictated. Massive water-cooled dummy loads balanced the transmitter
energy between the switches.
Each pair of klystron tubes was fed from
an electrical charge built up on immense capacitors which lined
the walls of a capacitor vault, an enclosed room about 18 by 30
feet in dimension. The capacitor vaults lay next to the transmitters
against the walls of the main floor, several vaults on each side.
The individual capacitors were cylindrical, about three feet tall
and a foot in diameter. The capacitor vault held dozens of these
capacitors which were connected together to make up the pulse
forming network of the transmitter. Occasionally, a capacitor
would fail, exploding in the vault, blowing the insulating tar-like
substance all over the inside of the vault, sounding just like
dynamite and occasionally causing a small fire.
The first explosion
I heard was frightening, but after hearing several and not having
responsibility for the transmitters, these explosions soon became
just part of the sounds of the site. Never having had to clean
up one of the capacitor vaults, I can nonetheless imagine it to
be a terrible job that is time-consuming, stinking, and perhaps
dangerous.
The middle transmitter building, Building
Two, held the main control room for the radar system and the MIPS
computer room. Special clearance was required to enter both these
rooms. From the central control room, signals travelled to the
other buildings commanding switching transmitters to scanners,
bringing repaired receivers online, and executing other control
tasks. Technicians had to receive permission from the controllers
here in order to take a component offline for service.
The MIPS received signals from the three
transmitter buildings, absorbing information about targets seen
penetrating the detection radar energy beams. The MIPS consisted
of a pair of IBM 7094 mainframe computers. Data from radar targets
was analyzed and calculations made resulting in a prediction of
whether a radar target could be a threat to North America. The
MIPS output was sent to NORAD Headquarters at Cheyenne Mountain,
near Boulder, Colorado. There, on a display panel representing
a map of North America, would be displayed an ellipse which represented
the probable impact area of a missile launched in the Soviet Union.
Happily, with one exception, the only targets displayed on the
display board were those of countless simulations run to test
the effectiveness of the system.
The single exception occurred shortly after the BMEWS at Thule went into operation. In October, 1960, the moon rose over the horizon directly in line with one of J Site's detection radar beams. The engineers who designed the BMEWS system had apparently not considered that the ultra-high powered radar beams would reach the moon and in about 2 seconds, return to the super-sensitive BMEWS receivers. The resulting returns swamped the MIPS with return information, sending thousands of threat warnings to Cheyenne Mountain. While the angles, speeds, and doppler information did not fit the model algorithms of a real threat, the sheer vastness of the return information overwhelmed the system. The U.S. did not react to the point that we were brought to the brink of war, but the doors to Cheyenne Mountain were closed and locked for several hours while analysts tried to determine the cause of the fiasco. Once it was understood what caused the problem, a solution was quick to come. A modification to the radar receivers, called a "Moon Gater" for its ability to block, or gate, moon returns by shifting receiver frequency every one-and-a-half seconds, was designed by RCA engineers and installed on all the BMEWS receivers. When moonrise was forecast in one of the BMEWS sectors, the Gater was turned on. Every second and a half, the receiver frequency shifted, and the returns from the moon were ignored. The frequency shift caused the receivers to run somewhat detuned, and lights in the DRAM room routinely turned yellow.