Welding brought with it new mysteries. Skunk Works engineers were at a loss to explain why A-12 wing panels that were spot-welded in the summer failed early, while those that were welded in the winter held together indefinitely. They traced the trouble to a characteristic of titanium: It is absolutely incompatible with chlorine. “We finally traced the problem to the Burbank water system, which had heavily chlorinated water in the summer to prevent algae growth, but not in winter,” Johnson related in a classified CIA journal article in 1982. “Changing to distilled water to wash the parts solved the problem.”
The A-12’s inlets, which include forward air bypass doors that automatically manipulate the air—and the famous spikes—were designed under Lockheed engineer (and later Skunk Works president) Ben Rich’s supervision and are the key to the airplane’s performance. The B-58 Hustler bomber used fixed spikes in front of its General Electric J-79s. Like his counterparts at Convair, Rich found that air hitting the unprotected front of a turbojet at high Mach numbers creates insurmountable pressure problems.
The spike’s purpose was to control the supersonic shock wave and, by working in combination with doors and openings to bleed away excess air, prevent supersonic airflow from entering the compressor intake. It accomplished this trick by moving back 26 inches into the throat of the inlet on a programmed schedule as the speed built up in a manner somewhat like the way the nozzle on a garden hose changes to adjust the water’s flow. At Mach 3.2, each inlet swallowed about 100,000 cubic feet of air a second.
When a spike failed to work perfectly, which happened repeatedly for many years, supersonic air pressure instantly built up and choked the compressors. They reacted without warning by violently spitting the shock wave back out. This event was the dreaded “unstart,” which, at Mach 2 or higher, would cause a thunderous bang that the crew heard and felt, while the airplane jerked so violently in the direction of the unstarted inlet that, according to Lockheed Martin’s Garfield Thomas, “it was like running into a brick wall.” He remembers one unstart yaw violent enough that the pilot “slammed his head so hard against the sill on the window [that it] cracked the helmet and knocked him semi-conscious.” A computer restarted the inlet, Thomas adds, and the pilot survived.
Blair L. Bozek, a reconnaissance systems officer with extensive experience with SR71s, says an unstart at Mach 3.2 “is like driving on the highway fast, taking an off ramp, hitting black ice, and having your car slam into the guard rail.” Justin Murphy, SR-71 program manager at the Skunk Works, remembers that “the way you knew which side the unstart was on was which side of your head hurt.” Bud Wheelon observed that Lockheed test pilots were “remarkable in their ability to delay normal human emotional reactions to chaos. In Area 51 [the flight test area at Groom Lake, Nevada], I’ve talked to them on the radio. They were calm as can be and very clinical, until they came down. And then they’d start weeping.”
The problem in the early days, says Wheelon, was that the hydraulic servo that moved the spike reacted too slowly to changes in Mach number. He told Johnson that hydraulic fluid can only move so fast; it has friction and inertia, and he began to advocate that the spikes be moved electrically.
“What you have to know about Kelly Johnson is that he’s everything everybody says he is,” Wheelon explains, still using the present tense more than six years after Johnson’s death at the age of 80. “He’s a wonderful guy. But he’s a stubborn son of a bitch to boot. And Kelly did not trust electronics.” Johnson also knew that electronic systems did not fare well in an engine that got as hot as the J-58 did.
“We’ve got $30 million invested in this servo,” Wheelon quotes Johnson as saying, “and we can’t just run a truck over it.”
“Every time we crack up an airplane we spend that much money. And we’re killing people,” Wheelon says. Of the initial group of 18 F- and A-12s, seven had been lost with at least two fatalities, though not all instances were attributable to the servo. Losses of SR-71s continued during the mid-1960s.
But Johnson was absolutely adamant: “I’m not going to have electronics in the airplane.”