An illustration of the Titan IIID rocket used to launch the KH-9. (larger version) (credit: © Giuseppe de Chiara)

The flight of the Big Bird (part 2)

The origins, development, and operations of the KH-9 HEXAGON reconnaissance satellite

<< page 1: returning to Earth

Pushing iron

NRO Director Al Flax proposed a new competition for the spacecraft element. This led to complaints, but he said that it would not delay the first launch by more than a few weeks.

On May 25, 1966 Flax authorized the creation of a source selection board for the “satellite basic assembly” with Frank Buzard in charge. Buzard had been involved with the CORONA program from its earliest days when he was responsible for overseeing the Thor rocket and launches for the early CORONA missions. (See “Has anybody seen our satellite?”, The Space Review, April 20, 2009.)

Buzard arranged to have the HEXAGON office physically collocated with the CORONA office to share staff. They expected CORONA to cease in the near future and most of the CORONA people would then move to the new program, but because the HEXAGON development dragged on, eventually they had new people come in who worked on HEXAGON but had never worked on CORONA.

Apparently Lockheed protested the renewed space vehicle competition. According to NRO historian Robert Perry, Lockheed officials thought that they had “twice won competitions” and nobody else could really compete with them. Flax pointed out that conditions had changed and that was why he was re-competing the contract.

It is easy to understand why Lockheed officials might have been annoyed. By this time they had more experience with providing space vehicles for military payloads than any other company. Lockheed Missiles and Space Division had started work on the Agena during the mid-1950s as part of the WS-117L program, when it was unofficially called the Hustler. The Agena had proven to be a rugged and reliable vehicle, and in addition to its many military missions, NASA had selected it as a rendezvous target for a number of Gemini missions.

The Agena was used for multiple purposes. In addition to serving as a second stage aboard Thor and Atlas launches to boost the payload into orbit, the Agena was often used to provide stability and control, as well as power—either in the form of batteries or solar panels—to the payload during its operational phase. Some payloads presented demanding requirements upon the Agena, which had to point reconnaissance cameras with high precision and minimal vibration at targets hundreds of kilometers away. The Agena also had an engine restart capability, which NASA used during two Gemini missions to boost the manned spacecraft into much higher orbits. It was a versatile spacecraft, but limited in size at a time when satellites were getting larger. In particular, it had a five-foot (1.5-meter) diameter that was dictated by the size of earlier rockets. But as the larger Titan II missile with its ten-foot (three-meter) diameter became available for space missions, the thinner Agena could not take advantage of this wider diameter to carry more fuel.

In 1966–67, Lockheed engineers proposed an improved version of the Agena to be used atop the Titan III booster. They designated it the Agena E. This upgraded version would retain the five-foot diameter of its predecessor. It would incorporate a lengthened forward equipment rack and an improved attitude control system. But the major change would be the Agena E’s propulsion system. It would have a new multi-start rocket engine and an “integral” secondary propulsion system that used the same propellants as those burned by the main engine. At the time, the Agena D used small solid-propellant Drag Makeup Units (DMUs) to periodically boost its orbit after atmospheric drag slowed the satellite. Other vernier or secondary propulsion systems could also be fitted to the Agena. Although very little information on the Agena E is available, presumably the new integral propulsion system would take the place of the DMUs or secondary propulsion systems, providing maneuvering capability during the operational phase of the mission, but not during the launch phase.

The Agena E was probably a multi-purpose vehicle, intended to serve as both an upper stage and a support spacecraft for payloads in orbit. It is unclear if it was initially proposed for any new search satellite program (i.e. including the S-2 and MATCHBOX) or also for another new reconnaissance satellite then entering service using the existing Agena D. But in 1967 the Agena E was apparently deemed insufficient for handling the beast that HEXAGON had become.

Lockheed soon came up with a proposal for a new satellite vehicle. This vehicle was named the Satellite Control Section, or SCS. The SCS was 10 feet (3 meters) in diameter and 9.25 feet (2.8 meters) long. This allowed it to take maximum advantage of the Titan rocket’s diameter and also made it twice as wide as the Agena D, but less than half as long, providing more room for the payload.

Although the Air Force has declassified its existence, specific information on the SCS remains somewhat sketchy. The SCS utilized the canceled Agena E’s integral secondary propulsion system, which used hydrazine monopropellant for fuel. The SCS could provide attitude control, propulsion, and utilities for the overall spacecraft, but did not have to provide any propulsion during the launch phase because the Titan with its two large strap-on boosters had enough power to place the entire vehicle in orbit. Thus, unlike its predecessor, the SCS was not a true upper stage. With its larger diameter and hence fuel supply, and no requirement for providing launch propulsion, the SCS could dramatically increase the time that the satellite could stay in orbit compared to other satellites using the older Agena D. Atmospheric drag would slowly pull the HEXAGON down, and the SCS’s engine would periodically fire to reboost it to the proper orbit. Although the Agena D could provide support for a reconnaissance payload for 60 to 90 days in orbit, the SCS increased that to over 100 days and eventually to 275 days.

The SCS was a major departure from previous American reconnaissance satellites in the way it kept pointed in the right direction. Once a satellite separates from its rocket and enters orbit, it will tumble unless it has equipment for keeping it stable. The CORONA and GAMBIT satellites all used some variation of a cold gas stabilization system. They carried tanks of nitrogen and spurted this out of small nozzles whenever the satellite started to move too far from the direction it was supposed to point. But once the gas was used up, the satellite could no longer point itself.

Lockheed decided to use gyroscopes for the SCS. Just as a spinning top will stay upright, a spacecraft equipped with spinning gyroscopes will remain pointed in the right direction. The gyros are spun up by small electric motors and they can remain going as long as the satellite has power. Furthermore, the spacecraft can actually use the gyroscopes to rotate around its center of gravity if necessary.

The Agena D used a different version of hydrazine as its propellant, combined with an oxidizer. Hydrazine itself does not require an oxidizer, but this lowers its performance and it is usually combined with nitrogen tetroxide to produce a higher specific impulse and hence greater thrust. Why only hydrazine and no oxidizer was selected for the SCS is unknown, but it may have been because it made the engine simpler and more reliable.

The SCS provided utilities to the reconnaissance payload, such as stabilization and power. The SCS was equipped with two large extendable solar arrays that unfurled at a diagonal angle from the rear of the craft. However, it is unclear if the SCS itself was equipped with star trackers for providing high-accuracy pointing for the spacecraft, or if the star trackers were actually mounted on the payload.

In the mid 1980s, Lockheed’s Space Systems Division produced a report outlining the capabilities of a vehicle it designated the “SSB Satellite Support Bus” that was clearly derived from the SCS. A satellite “bus” is a standardized structure upon which the payload and other subsystems are attached. Lockheed apparently produced this declassified report for the Strategic Defense Initiative Organization, which was then studying the possibility of operating anti-missile defenses in low earth orbit. Presumably, the SSB could have been used to support a large chemical laser or other system for destroying ballistic missiles in flight.

The report stated that the SSB “consists of an Equipment Section and an Orbit Adjust Module/Reaction Control Module Section.” It was “made of semi monocoque construction with a corrugated aluminum external skin.” It weighed approximately 3,600 pounds (1,633 kilograms) excluding propellant. The Orbit Adjust Module/Reaction Control Module Section had 12 cargo bays which could be used for additional equipment or expendables. The equipment section consisted of 12 equally sized bays, each capable of supporting up to 550 pounds (249 kilograms) of equipment on individual trays. Each of the bays could be accessed through an external door.

At the center of the vehicle was a 76-inch (1.9-meter) diameter hydrazine propellant tank. The orbit adjust system consisted of a hydrazine monopropellant motor with a specific impulse of 230 seconds and a thrust level of 145 to 230 pounds (645 to 1,023 newtons). The motor had an 85-to-1 expansion nozzle. The vehicle carried 7,300 pounds (3,311 kilograms) of hydrazine. The Reaction Control System consisted of 16 hydrazine motors in four modules utilizing the same propellant supply as the orbit adjust system. In addition, up to 1000 pounds more of hydrazine could be carried and by the mid-1980s an operational mission apparently had already used this capability.

The SSB was not equipped with star sensors as standard equipment, although they could be carried. The SSB was equipped with two solar arrays with 11 panels each, for a total of 172 square feet (16 square meters) of active area. These provided an average of 600 watts of power, of which the SSB required 270 watts, leaving the remainder for the payload. The peak power output was two kilowatts when the panels were directly facing the sun. The SSB, obviously derived from the KH-9’s Satellite Control Section, was a massive piece of equipment in its own right; if it had been launched on its own it would have been one of the heaviest low-Earth orbiting satellites ever flown.

Titan’s flames

When the Air Force started the WS-117L satellite reconnaissance program in the mid-1950s, they planned on using the Atlas ICBM as a launch vehicle. The Samos satellites had all flown on Atlases. When the KH-7 GAMBIT came along it too flew on an Atlas. However, the CORONA satellites were relatively small and they launched atop the Thor, which was cheap. Even when reconnaissance satellite programs are declassified, one of the few things that remains classified is the cost. But when CORONA was declassified the CIA apparently accidentally let slip how incredibly cheap it was—around $10 million apiece. Another source indicated that each CORONA mission cost around $20 million in the late 1960s, apparently also including the Thor rocket cost.

But even as early as 1963 many of those involved in the effort to develop the new search satellite system realized that it was likely to be big and would therefore require a more powerful rocket. It is unclear when they first started talking about the possibility of launching on a Titan III rocket, but the Titan III was mentioned as a possible reconnaissance satellite launcher by the Purcell Panel in summer 1963. The Manned Orbiting Laboratory, which was first started in late 1963 and then formally approved in 1965, after it acquired a reconnaissance mission, was always going to use a powerful variant of the Titan designated the IIIM. The new search system wouldn’t require anything that big, but they were going to need some upgraded variant of the Titan III.

By the time HEXAGON was approved the Air Force officers responsible for the launch vehicle development must have known that they were going to use a Titan III and the only question was which variant. They initially considered using a Titan IIID booster with two or three-segment strap-on solid boosters. Each of the boosters was a massive 120 inches (three meters), the largest operational solid rocket motor then under development. But apparently some people in the program advocated an alternative five-segment strap-on, which would have been closer to the Titan IIIM then under development for MOL. The launch vehicle managers also wanted an improved engine for the Titan and initial procurement of ten Titan IIIDs. The Titan IIID would become the workhorse launch vehicle for the HEXAGON, but was replaced late in the program by the more powerful Titan 34D.

Adding a mapping camera

The HEXAGON’s two big cameras could photograph vast swaths of territory. But they were not explicitly designed for producing accurate maps. Mapping cameras have different requirements than reconnaissance cameras. Not only do they need to photograph relatively large amounts of terrain—so that mappers don’t have to tape together dozens or even hundreds of photographs to produce a decent map—but they have to provide a high degree of fidelity, meaning that one square inch of the photograph should show the same amount of territory as any other square inch.

By this time it was common for photographic reconnaissance satellites to carry mapping cameras. CORONA had acquired a rudimentary mapping capability that was improved over time and the GAMBIT apparently also used the same camera as developed for CORONA. Because HEXAGON was going to replace the CORONA and operate for many years to come, it also needed a mapping capability. Preliminary studies of mapping, charting and geodesy systems for HEXAGON were not completed until March 1967 and apparently a configuration was not decided until May 1967.

Whatever the system was that the NRO selected for HEXAGON, it had two drawbacks. Apparently it could not meet a requirement for 1:50,000 scale maps, and it was also going to be expensive. A committee recommended deleting plans to incorporate a mapping capability into the first several vehicles, possibly until a better mapping camera could be developed. The CIA argued that HEXAGON should not carry mapping equipment at all.

However, the Army, which had primary responsibility for producing maps for use by the US military, continued to insist that a mapping camera was necessary. DNRO Flax apparently argued that the mapping camera could not be carried on an initial launch, but the EXCOM agreed that it should be used, perhaps for a block 2 vehicle. No formal contract for a mapping camera was signed until November 1968.

The first four HEXAGON missions did not carry a mapping camera, but it was included for the fifth launch. In 2002, when the intelligence community declassified imagery from this mapping camera, they revealed the mission numbers, which also included the number of the satellite recovery vehicle that carried the film. For example, a HEXAGON mapping camera image of the Dolon Airfield taken on May 28, 1974, came from mission 1208-5. The mission numbers for the mapping imagery all had the suffix five at the end, indicating that a fifth satellite recovery vehicle had been added to the spacecraft.

In the 1980s NASA announced plans to fly two space shuttle missions carrying a Large Format Camera manufactured by Itek. This was a film camera to be used for producing 1:50,000 scale maps. The shuttle Challenger carried the camera during mission STS-41G in October 1984. It is possible that this camera was adapted from the mapping camera carried aboard the KH-9, which was then scheduled for retirement. Arzhan Surazakov and Vladimir Aizen, writing in the Journal of Photogrammetry and Remote Sensing in May 2010, noted a number of similarities between the film produced by the two cameras, although there were also a number of differences. But there is no firm evidence to support the conclusion that the camera carried on the shuttle was directly derived from the KH-9 mapping camera.

Turmoil, cancellation, and resurrection

Final contracts for the satellite vehicle were not signed until December 1967. There were substantial cost increases, and the problem of additional weight, possibly due to the mapping camera. Some kind of redesign of the camera and satellite systems was required in December, and this apparently affected the planned launch schedule. In addition, there were still unresolved questions about the film path, the kind and quantity of test equipment, and the scope of the camera system testing to be performed once it was passed on for final integration in the spacecraft, which was to occur at Lockheed’s Sunnyvale, California, facility.

Recovery vehicle development was apparently approved in early 1968, with a few other subsystem approvals following in the next few months.

Although HEXAGON was fully underway by 1968, several other proposed unmanned camera systems “were beginning to demand attention,” according to historian Robert Perry. For example, a high-resolution readout system with near-real-time capability was also under evaluation. These other proposed systems threatened the HEXAGON program with cutback or even cancellation.

The biggest problem was apparently the Manned Orbiting Laboratory. MOL in some ways is more enigmatic than the HEXAGON. Although it started out as an experimental laboratory to determine if military astronauts could accomplish useful missions in orbit, by 1965 it evolved into a manned reconnaissance platform with a very powerful camera. MOL was apparently supposed to fly along in its orbit and if a military astronaut saw something interesting in its powerful camera, he would press the shutter button and take a picture. MOL’s KH-10 DORIAN camera was reputed to be capable of spotting objects on the ground as small as four inches (ten centimeters) on a side.

DORIAN and HEXAGON had very different capabilities and requirements. HEXAGON would produce massive amounts of imagery that was good enough to determine technical details of the targets it spotted. DORIAN would produce very high-resolution images of very small areas, perhaps only a few kilometers wide. Important targets could evade MOL’s view.

But the one thing that both systems had in common was that both were big and therefore expensive. Both used variants of the Titan III, with MOL’s Titan IIIM requiring substantial development work. According to one source, Air Force officers attempted to persuade members of Congress that HEXAGON should be canceled and MOL continued. But ultimately this was a decision for the President, not the Congress.

According to Perry, the Bureau of the Budget “revived an earlier suggestion that the combination of GAMBIT and an improved CORONA [presumably some variant of what was generally known as the CORONA J-4 proposal] would satisfy the requirement” and possibly be cheaper than the HEXAGON. But nothing had changed since the Drell Committee had started its evaluations several years earlier. The modified CORONA could not do much better than about 4.5 feet (1.4 meter) resolution and according to Perry “all those agencies were agreed that search resolutions better than three feet (0.9 meters) were essential to verification of arms limitation agreements.” What had started as a CIA conclusion that any new search system needed both broad area coverage and high resolution had now become accepted by everyone else involved.

The Director of the Bureau of the Budget Robert Mayo took budget bureau arguments to President Nixon who on April 9, 1969 ordered that HEXAGON be canceled. Nixon also approved continuing the Manned Orbiting Laboratory program to completion.

According to Jeffrey Richelson, in his 1990 book America’s Secret Eyes in Space, Roland Inlow, who was chairman of the Committee on Imagery Requirements and Exploitation (COMIREX), which established the target lists for American reconnaissance satellites, spoke to James Schlesinger, assistant director of the Bureau of the Budget with responsibility for national security programs. Inlow told him about the importance of HEXAGON for arms control verification. Presumably MOL, with its relatively small viewing area, had less usefulness for arms control purposes. In addition, an independent recommendation of the Land Panel, the senior advisory committee to the president on satellite reconnaissance issues, went to Nixon on May 6, 1969. Land and his group favored canceling MOL.

According to one source who had been involved in training MOL astronauts how to become effective photo-interpreters, at some point Vice President Humphrey was given a briefing on both the KH-9 and the MOL at a secure briefing room in Washington, which he remembered was at the Washington Navy Yard and could have been at the National Photographic Interpretation Center located on the edge of the Navy Yard. The Director of Central Intelligence, Richard Helms, sat next to the Vice President and during the MOL briefing he wrote something on a piece of paper that he slid over to Humphrey, who looked at it without commenting. After the briefing ended, this source waited until everyone had left the room and then he retrieved the piece of note paper. On it, Helms had written “Why four inches?” Four inches was the resolution of the MOL’s large DORIAN camera. Presumably, MOL’s advocates lacked a convincing argument for why such a powerful, but limited reconnaissance system was necessary.

According to NRO historian Perry, Nixon “reversed his earlier verdict” on HEXAGON and instead ordered cancellation of the MOL.

Shuffling schedules

The NRO leadership had ordered enough CORONA systems to cover a gap in coverage if HEXAGON development lagged. But they also started an insurance policy known as HIGHER BOY. It involved adapting the high-resolution KH-8 GAMBIT to fly in a higher orbit. HIGHER BOY did not prove necessary, and the sole example was placed in storage. It was later retrieved from storage and flown in 1982, although it apparently was not successful.

According to Perry, in late 1968 program managers conducted a review of engineering work undertaken up to that time. The review discovered a still-classified problem requiring prompt attention.

In June 1969, John McLucas who replaced Al Flax as DNRO in April 1969, asked his deputy, F. Robert Naka, to undertake an assessment of the launch schedule, in part to determine if they should order additional CORONA vehicles to cover any delays. Naka concluded that there was reasonable probability “that at least one of the first three [HEXAGON] missions would be successful.” Given those odds, he suggested that the twelve CORONAs programmed for launch at about two-month intervals between June 1970 and July 1971 should be rescheduled to allow for at least two HEXAGON missions, thus insuring a minimum overlap of CORONA and HEXAGON missions.

Naka’s report and increasing costs led to an EXCOM meeting. Somebody had reported the possibility of potentially massive cost growth in the program at a time when there were other cost pressures. In August 1969 the chairman of the EXCOM, David Packard, asked if they should cancel HEXAGON—only a few months after it had been revived from the dead.

One of the reasons that HEXAGON was costing so much money was that contractors had started going to double and triple shifts and overtime to get it done. This was not normal for defense procurements except when there was a critical shortage of a vital product, like ammunition. Perry noted that with other defense procurements it was possible to simply keep using the existing equipment (like airplanes) if the new ones were late. But satellites were different, because they were expended “regularly and inevitably.” Hence, the overtime, and the escalating costs.

But despite all these problems, the NRO’s leadership was running out of options. Lead time for building a CORONA was 18–24 months, meaning that a system ordered in December 1969 could not be delivered sooner than June 1971 (i.e. around the time HEXAGON was supposed to start flying). But by January 1970 the EXCOM decided to not order more CORONA vehicles.

In December 1969, Brigadier General Bill King, who ran the NRO’s west coast office with primary responsibility for launching the NRO’s satellites, among other things, convened a special meeting to assess the ability to meet the schedule. They all agreed that although it would be tight, they could probably remain on schedule if they pursued “vigorous action.”

Another major problem cropped up by summer 1970. J.R. Schlesinger, acting Deputy Director of the newly-created Office of Management and Budget, again proposed buying more CORONAs to fill the gap if this problem proved to be part of a major defect. But Naka explained that the last chance to order CORONAs had been in February, and if they continued CORONA launches at the planned rate, there would be a six-month gap before new CORONA systems could be delivered. That train had already left the station.

When the second KH-9 payload was delivered there were various camera and film-tracking problems that stalled the test program. But the details of these development problems, like so much else about the design and construction of the KH-9, remain classified for now.

By March 1970 problems were encountered in the acoustic and thermal tests of the first payload-vehicle assembly. According to Richelson, a change involving a resistor had been ordered but never implemented. During a thermal vacuum test this resulted in “catastrophic damage” to the film supply. Richelson stated that the second camera was then prepared to be the first to fly. However, another problem cropped up. Fortunately, once they diagnosed it, the fix turned out to be “almost simplistic,” according to NRO historian Robert Perry.

Throughout 1970 and into 1971 vehicle development progressed, and by summer of 1971, the massive new reconnaissance satellite made its way to Vandenberg Air Force Base and its first launch.

Next: The Big Bird Flies

Acknowledgements: The author wishes to thank Henry Check Jr., Nick Strauss, and Giuseppe de Chiara for his excellent artwork.

Dwayne Day can be reached at zirconic1@cox.net.

Home