Computing in the Soviet Space Program

Introduction

Bibliography

Computers

Discussion

Documents

Essays

Institutions

Interviews

Links

Sitemap

• Could you tell me about yourself: when did you join Pilyugin's firm, what did you work on?
• Did you attend any launches?
• When did the idea of using onboard computers in spacecraft control systems first arise?
• When digital computers were introduced, what was seen as their advantages in comparison to analog devices? And their deficiencies?
• When did the transition to integrated circuits occur? Why did it happen?
• Was there any study of the American experience?
• Have there been any attempts to compare American onboard computers with Soviet ones in terms of weight, speed and so on?
• How was the problem of division of functions between human and machine resolved?
• Who was doing programming for the onboard machine on the Buran?
• Could you tell more about the Bisser computers?
• Were there any computer malfunctions during the Buran launch?
• What was Sergei Korolev's opinion on the prospects of using onboard computers?

Gerovitch: Could you tell me about yourself: when did you join Pilyugin's firm, what did you work on?

Priss: I joined the firm in January 1948. The firm was organized by the well-known order of May 13, 1946, signed by Stalin. It was a general order to start the development of rocket weaponry. It contained specific assignments to various ministries. For example, it would read "on the basis of such-and-such factory, create an organization that will be doing this and that." A similar assignment was given to the Ministry of Electrical Industry, in which there was a factory that produced communications equipment: telephone sets, telephone exchange switchboards, and so forth; it was located on Aviamotornaya. On the basis of that factory, our organization was created. It had two assignments: the creation of autonomous control systems and the creation of radio control systems. Therefore, there were two chief designers in our organization: Nikolay Alekseevich Pilyugin and Mikhail Sergeevich Ryazansky. Pilyugin was the chief designer of autonomous systems, and Ryazansky worked on radio systems. All of them - Pilyugin, Korolev, [Valentin] Glushko - in 1945, after the Victory, were sent to Germany, where they studied captured rocket technology. There were doubts and debates in the Soviet Union: whom to assign the production of rocket technology? The Air Force refused; the Ministry of Ammunition also refused. In the end Nikolai Dmitrievich Ustinov understood the value of this technology - he was then the Minister of Defense Industry - and he took up this job. Then Stalin's order appeared, and all ministries received specific assignments.

The people who had worked at that factory now switched to rocket engineering. They were telephony specialists evacuated from Leningrad, from the Krasnaya Zarya factory. They were very competent in their work. They were inventors, who in the early days of telephone engineering were engaged in the construction of telephone systems - not only telephone sets, but also stations. The captured German rocket technology was based on relay circuits. Analog control devices used vacuum tubes. There were also some gyroscopic devices. The Germans had only five [control] devices in total.

Pilyugin, who had studied this technology in Germany, was appointed to supervise these relay specialists. He had graduated from the Moscow Advanced Engineering School (MVTU). In 1947, when we came, he had about 70 people. Back in 1946 he had just 30. In 1947 Pilyugin was a member of the graduation examining board at the Moscow Aviation Institute. He attended graduation project defenses and listened to the presentations. He used the following trick: he talked not to the students who came to defend their own projects but to those who came to listen to other students' presentations. They were about to defend their own projects, and they came just to listen. Based on their projects, Pilyugin selected the first group of six students and hired them.

I was in the second group. I met Pilyugin at the end of December 1947, when he attended project defenses of the first group. We still had about six months left to do our own projects. He told us: "Come to me, and do your projects with me." Because of secrecy restrictions, he did not tell us any details about his work, but just said: "Here are such and such research topics. If you want, I can hire you as technicians while you do your projects." We did not know any specifics. We thought, why not? What's the difference where to do our projects? So we agreed. We came to do our projects in January, defended in June-July 1948, and all stayed with him. There was also the third group, in 1950-1951. Stalin's order was all-inclusive: it had a provision for the organization of special student training and student job placement in the centers of rocket industry.

Pilyugin assigned topics by intuition, as I understand, and I got integration. There were six of us, and the majority began working on specific devices, while I was assigned to the integration laboratory. I defended my project, and since then I have been working on integration.

The Soviet rocket industry had a unique organizational structure. If you take aviation industry, you won't find there a "chief designer" of the control system for a particular aircraft. There is no such position, even though there are control systems, and they also have gyroscopic devices and computing devices, and their tasks are as complex as ours, and maybe even more. In aviation industry, the chief aircraft designers - Aleksandr Yakovlev or Andrei Tupolev - used the compilation method. They always had "equipment groups." Many of our college mates, who graduated the same year, before, or after, were placed in so-called "equipment brigades." They visited various aviation firms that produced autopilots, electric transformers, radio equipment, and so on, and searched for the equipment they needed. At best they could order equipment, but usually they had to take whatever they were given, and then they assembled these pieces and created a control system. By contrast, in the rocket industry the chief designer of [autonomous] control systems [Pilyugin] was appointed at the very beginning. Next to him was the chief designer of gyroscopic devices, Viktor Ivanovich Kuznetsov. The chief designer of radio control systems also was appointed. The Council of Chief Designers was created. It also included Sergei Pavlovich Korolev, the engine designer Valentin Petrovich Glushko, and the ground complex designer Vladimir Pavlovich Barmin. That's six people in total, the entire Council. They resolved all the technical questions.

What is integration? All these individual devices must be put together to make a unified cyclogram; there must be some equipment that would integrate everything. For example, we needed to integrate various parts of the power system: an accumulator battery and various converters into other kinds of energy (gyroscopes, for instance, require alternating current). We needed to prepare a cyclogram of the pre-launch sequence, to operate the engines and all onboard systems, and so on. It is the control system that turns on the engines at lift-off.

I graduated from the Equipment Faculty of the Moscow Aviation Institute. This Institute is distinguished by its broad education. They taught us how to design oxygen equipment, hydraulic equipment, radio equipment, and other kinds of devices. I was finishing college after the war, and at that time and during the war lots of new types of equipment appeared in Russia: American airplanes on lend-lease contracts, captured German equipment, and alike. We began studying all this. In the Faculty of Equipment and Instrument Building, the department of Viktor Naumovich Mil'shtein organized a group to study new equipment during an additional, eleventh semester. I got into this group. My diploma read "mechanical engineer with specialization in gyroscopic devices." For me, the idea of integration was familiar; I knew where every type of equipment belonged.

Since then, I have been doing integration. I was the lead integration engineer on the R-7 rocket, and before that on the R-5. I worked also on the very first rockets, the R-1 and the R-2, but as an ordinary engineer, not as the lead engineer. I was the lead developer on the N-1 and on Buran. Since 1956, I have been Pilyugin's deputy in charge of equipment and the test-and-launch complex. Pilyugin since died; his former post is no longer call the Chief Designer, but the General Designer, and I am still a deputy. Now I am the head of the Integration Department. We currently work on a joint project with Australia on the Aurora space-rocket system. We have the Integration Division, which includes several departments. Different departments work on the integration in different projects. When you finish the integration for one project, you switch to another. I also work on the problems of stabilization routes (the entire control system is divided into separate functional routes) and on communication with radio operators for other projects.

There was a period in the 1960s when I was the head of the entire Integration Division. In 1963 we moved from Aviamotornaya to the Southwest district, and after 1963 I was the head of this division and was working on the UR-500, on the Proton, and on the N-1. Almost 600 people worked in that division.

Gerovitch: Did you attend any launches?

Priss: Yes, of course. I attended all the launches of the N-1 and the first, unsuccessful attempt to launch Buran. Then I left for Moscow and did not return for the second attempt, but stayed at the Mission Control Center. I was curious to see the operations of the Center. I had already been to the launch site, and those who see the launch do not see the landing. Those who sit in the bunker during the launch cannot be at the landing site. And at the Center you can see everything and know everything.

Gerovitch: When did the idea of using onboard computers in spacecraft control systems first arise? Who suggested it first? What reaction did it provoke? Was it implemented right away?

Priss: By the time this idea emerged, the use of computers had already been widespread. There already existed ground computing systems; some information about the construction of specialized computers for guidance (not just spacecraft guidance) had appeared. Generally speaking, this idea was in the air. Computers had already spread quite widely. It was abundantly clear, especially for the main designer of control systems for spacecraft Nikolai Alekseevich Pilyugin, that computers could be used for solving problems of guidance and that they had many advantages over analog systems, which had been used before. It is hard to tell, to whom the idea of using computers for these problems came first. This idea was facilitated by the achievements of engineering at that time, including the achievements in computing. At the initial stage of the development of computers, it was simply impossible to use them on board because of the large dimensions and huge energy consumption, by the mid-1960s technology advanced so much - both in terms of basic elements and design techniques - that computers became quite acceptable by these external parameters.

The first onboard computer was installed on a space vehicle sent to Mars. It was part of the control system for a spacecraft intended for a flight toward Mars in the mid-1960s. Pilyugin decided to use a computer code-named Argon-11, which had been developed under the chief designer Krutovskikh. A Moscow organization [NIEM, later NICEVT] was designing computers, which by their external parameters were quite suitable for use as part of onboard equipment in spacecraft. And they modified their Argon computer to adapt it to the kind of problems in which Pilyugin was interested, and they code-named this new computer Argon-11. These problems were by and large problems of guidance. This was the first example.

After that, Pilyugin acted in his typical manner, in accordance with the engineering culture of his firm. He had been designing control systems since 1946, and he always adhered to idea of building an autonomous control system, one that would contain within itself all the subsystems necessary to solve all control problems. This was not an unfounded hope; this was a definite program. If the development of all subsystems of a control system is concentrated in the hands of one chief designer, then the control system can truly be optimized. This makes it possible to allocate functions among various parts of the system in the most expedient way in terms of weight, dimensions, energy consumption, and functional decision-making. Under this approach all main, as well as auxiliary, parts of the control system are being developed in one organization.

Even before Pilyugin applied this approach to onboard computers, about five years earlier, he had performed a similar trick with a major, core component of the control system - the gyroscopic unit. Until the end of the 1950s Pilyugin's firm had not developed inertial gyroscopic devices. There were specialized organizations that designed such devices. In particular, the Chief Designer Kuznetsov's firm developed various gyroscopic devices for the rocket industry. He also began working on gyroscopic platforms. In the latter half of the 1950s, Pilyugin incorporated a gyroscopic platform developed in Leningrad by the Chief Designer Arefyev into a missile control system. Pilyugin thus acquired expertise, obtained the results of on-site testing, and later started the development of gyroscopic platforms in his own firm. He used the same approach to facilitate the introduction of computers into the control system. The first experiment that I mentioned - the experiment with Argon - led to Pilyugin's decision to start the development and manufacturing of his own computer systems.

For the second phase of the development of the lunar complex N1-L3, Pilyugin decided to design his own onboard computer. In the first phase, the control system included some gyroscopic devices of the Chief Designer Kuznetsov and some analog systems. In the second phase, a totally new control system was created. This new design was implemented in the fourth (last) launch in 1972. The new control system included a gyroscopic platform and a computing system, both designed by Pilyugin. This computer, called the S-530, was based on the Tropa elements. It was intended for calculating all control tasks: guidance, control of the operating logic of the control system's own equipment, and control of the engines and of all other systems in the N1 rocket. TThe same computer was to be used in the control systems. Unfortunately, these plans did not materialize. The fourth launch did not result in the operation of the LOK or the LK, and therefore no experimental results were obtained about their control systems from that launch.

Afterwards all control systems developed by Pilyugin included computing systems designed at his own firm. After the C-530 we developed a computer series named Bisser: Bisser-2, Bisser-3, Bisser-4, and Bisser-6. The Bisser-6 computer represents today's technological level. Of course, these machines improved over time. In particular, the scale of circuit integration increased from medium-scale integration to large-scale integration to the current solid-state technology of very large scale integration. This provided an opportunity to increase the computing resources of the machine, to enhance its memory, and to raise its speed.

All these systems were based on the same design principle. High reliability was required of the control system in general, and of the computing system in particular. Therefore all these systems were designed with redundancy. They could withstand the failure of any single component. They had radial interfaces with various exchange devices, such as sensor data converters and operating component converters.

Along with the introduction of onboard computers, computer equipment was being installed at ground testing and launch facilities. Here we used the same approach. During the development of the first rockets, for example, the N1, the ground facilities used industrial production computers, such as SM-2 (developed at Severodonetsk). Later on they switched to special ground models of the existing onboard computers. Those ground models also had higher reliability. Using the same type of computers made it easier to establish computer-to-computer communication between the ground and the board, and this approach is still applied today. Since the structure of the control system for various space vehicles (with the changes in rocket stages and cargo weights) has evolved significantly during these forty years, a whole multi-computer system is now created, which includes a ground machine and several onboard computers.

Onboard computers could be designed for different purposes, and many specialized computers were developed. For example, some specialized computers handled the communications between the central computing system and various onboard systems, in particular, the radio systems and the information display and manual control system. Such specialized computers are of different types, and they may have different time cycles and interface types. Nevertheless, with the help of appropriate adapters and coordinating devices, it is possible to create a unified system. This principle was implemented not only in Pilyugin’s projects. The Mir space station also included a multi-computer control complex developed by specialized organizations, in particular, by the Zelenograd Science Center. It is the same today with the International Space Station: different types of computers are included in a unified computing system.

Gerovitch: Researchers from NICEVT, who developed the Argon-11, say that this computer was installed on board the Zond spacecraft that circled the Moon.

Priss: Zond is a name for the L-1, which is also our project, but this happened after the flight toward Mars. Argon-11 was used in other projects too. Speaking of priority, the very first use of this computer was on the spacecraft sent to Mars. The L-1 was launched with the Proton rocket; this project preceded an expedition to the Moon. There was one successful launch of the L-1, which returned to the Earth and landed with escape velocity. If this project were to continue, this computer would have been replaced with the S-530 machine too, but this did not take place.

Gerovitch: In the initial period, when digital computers were introduced, what was seen as their advantages in comparison to analog devices? What were their deficiencies? Was it obvious that a transition to this new technology was needed?

Priss: At that stage this was not obvious. First, the basic elements at that time were not sufficiently reliable. There were numerous failures, and therefore a decision to launch could be made only when redundant devices were installed. There was hope that back-up systems would ensure reliable operation in flight. During preliminary tests all back-up systems, including the computing system, were checked, and, as a result, a very large number of elements were tested, and failures occurred more often. For this reason, preparations for the fourth N1 launch (when an onboard computer was used) proceeded with great difficulty. Various computer elements failed all the time, repairs were repeatedly made, and there was always a hanging question whether we would be able to move on to the next stage in testing.

Second, digital computers at that time did not have any advantageous external operational characteristics. That is, they were not much lighter or more energy-efficient than analog systems, and in this sense there were no advantages. One had to learn how to program them, how to test algorithms, and how to test software. All this involved great difficulties, and we placed our hopes on future improvements. At that time, it was already clear that this technology had good prospects, since the development of new basic elements was under way. It was clear that computers would be able to solve many more problems than they did at the initial stage. Today computers handle control logistics and systems control, not to mention new, advanced guidance systems, which would have been very difficult to build with analog devices. Innumerable changes in the launch inclination, in the direction of flight, in the trajectory, transitions from one trajectory to another - all this could in principle be solved with analog methods, but it would have been much more difficult, not to mention the speed of calculations, which is very important for logical tasks. It is less important for guidance tasks; in the latter case, all processes run more slowly, than, say, the processes of starting-up and regulating the engines. Initially all these problems were solved with analog methods. As computers improved, these tasks were shifted on to them, and thus the functions of the control system were substantially expanded.

TThe use of computers made it possible to eliminate entire classes of equipment and whole subsystems. For example, today a computer-based gyroscopic system, incorporated into the control system, fulfills the task of the initial azimuth aiming of a rocket. All the first rockets of the 1940s, 50s, and early 60s were aimed with the help of an external geodetic system, which somehow had to agree with onboard gyroscopic devices. One had to perform a geodetic locality orientation, one had to install special geodetic devices on the ground, and then, with the help of optical communication devices, one could turn and aim the rocket. The launch installation had to provide means for turning the entire rocket. Later on the turning operation no longer had to be done on the rocket itself but was applied just to its gyroscopic system. And when a computer complex was added to the gyroscopic system, the entire system was greatly simplified. Onboard pulse generators, which had earlier been used for timing coordination, were eliminated; electrical current distributors, which put a cyclogram into effect, were also eliminated - all this is now done by a computer complex. As a matter of fact, the computer complex and the gyroscopic system are the only essential components of the control system; everything else are just various converters and capacity amplifiers, which control operating components, rocket automatics, engine automatics, and power source. That’s it. Actual control problems, which earlier had been solved by complex devices, are now solved by a computer. Today ground computers prepare the flight task and recalculate it on the fly if necessary. All serious functions of the control system are now performed by computers.

Gerovitch: You mentioned that the first computer made at Pilyugin's firm was built on the Tropa elements, which were not very reliable. When did the transition to integrated circuits occur? Why did it happen?

Priss: The S-530 computer was developed in the late 1960s, and it was used on a spacecraft in 1972. Subsequent machines in the 1970s were based on integrated circuits of various types; the first had medium-scale integration. Here again Pilyugin chose an original approach. Different options were available at the time. In our branch of industry, a new, so-called “open-frame” technology was available: a designer could build the required circuits with open-frame elements – tiny diodes and triodes – and thus could in practice, not just on paper, implement something similar to medium-scale integration elements. This technology was being used, but not at Pilyugin’s firm. He chose an alternative method: he let contracts to various specialized organizations of the Ministry of Electronics Industry. According to his specifications, large-scale integration elements were developed. Pilyugin’s firm received finished, well-tested elements, and then assembled the required units out of those elements. Such was our approach in the 1970s, and it was used not only at Pilyugin’s firm. In the late 70s the same approach was used by another firm engaged in the construction of control systems, the Kharkov Institute [Khartron]. All other design organizations that worked on various elements of a control system also introduced one technology [of integrated circuits] or another. The “scattering” technology was widely used with good results at the Research and Production Association of Measuring Technology, which built telemetry systems, sensor equipment, transformer equipment, and so on. I believe they still use this technology. Our Bisser-6 computer is built on large-scale integration circuits developed at Minsk or Zelenograd firms.

Gerovitch: The Americans chose integrated circuits as the basic elements for the Apollo Guidance Computer in the early 1960s. Was there any study of the American experience? Was it taken into account? Did it have any impact on technical decision-making?

Priss: I am not aware of that. In 1960s NICEVT was engaged in this more deeply than we were, and they possibly knew something. In the past several decades, this information has widely spread. Studies [of foreign-origin hardware] and comparisons with basic elements and capacities of our electronic technology are conducted all the time. For a long time the military system of quality testing blocked applications of these [foreign] types of systems or elements. Until the last decade, everything [in the space program] had to pass quality checks by military organizations, and one of the requirements for control systems was the use of domestically manufactured basic elements. In recent years this rule has not been enforced. If such [foreign] technology is used, this requires detailed knowledge and understanding of what is being done abroad, and this also requires the creation of domestic analogs to foreign models. To create an analog takes a lot of time. To repeat always means to lag behind.

Gerovitch: Was there any discussion of the American design of the control system for Apollo?

Priss: I do not recall seeing such materials or publications; I do not think that we had detailed knowledge and understanding of the hardware implementation of the Apollo control system. There was some information available about general design solutions and trajectory calculations for Apollo, but these issues do not belong to the control system. This information could be of interest only to the lead organization, which was selecting trajectories and strategies for space flight. Despite the substantial openness of the Americans and the availability of endless material on the Shuttle (there was less available on Apollo), there were almost no publications that would specify hardware design. One could find some general characteristics of the equipment, for example, the gyroscopic platform or the computer (speed, memory, and so on), but more substantive information was absent. Perhaps, this information was available elsewhere, but it did not reach us, the developers.

Gerovitch: Have there been any attempts to compare American onboard computers with Soviet ones in terms of weight, speed and so on?

Priss: I know that in the early 1980s there were attempts to compare the parameters of systems designed for Buran and for the Shuttle. And I must say that, based on external parameters, we did not lag behind. Our machine on Buran was Bisser-4. In terms of computer performance per se, the parameters of our machine were close to theirs; perhaps, we even had a lead. As for the functional operation, here you have to take into account that the overall structure of our system was different. Speaking of computers, the overall picture was as follows: there were four machines on the Shuttle; they worked in the regular mode during the ascent, and they were supplemented with the fifth machine. This fifth machine was of the same type as the other four, but it had different software, different "mathematics." It was used as back-up mathematics in an emergency. According to the American plan, this back-up mathematics was to be developed by another team, not by those who did mathematics for first four regular machines. The idea was obvious. In the American case, a switch to the backup system was to be carried out by an astronaut. If the astronaut realized that the first system was not working properly, he could switch manually to the second system. We did not adopt a similar approach for two reasons. There was no other organization that could develop the backup mathematics. Even if such an organization were found, it was still necessary to test and debug their mathematics. Our testing program included an extensive series of ground tests and later in-flight tests. The second mathematics would have required additional spending and cause time delays, and we could not afford that.

All these issues were widely debated by technical specialists engaged in the development of Buran. There were many organizations involved. The lead aviation firm that built the Buran glider and was responsible for landing strongly insisted on considering this option. But eventually we managed to prove the reliability of our regular system. Together with the Rocket-and-Space Corporation Energia, we managed to demonstrate that we foresaw a large number of potential emergencies, and for all these potential emergencies we developed automatic modes of coping with them and rescuing the crew.

For example, we introduced the innovative mode of "restoration of control" on Buran. Let’s say, for some unclear reason the control system breaks down in orbit, all four machines break down, the orientation is lost, the spacecraft begins to make strange and unpredictable maneuvers. For example, this can happen because of an interruption in power supply. If power is not restored, even the Americans with their fifth machine could do nothing. If power is restored, however, then a special mode would be switched on, which would restore the orientation, tell you where you are, restore all the processes, and make it possible to rescue the crew and to return them to Earth.

Nevertheless, the idea of an additional, backup system was discussed all the time, and as a result, it was implemented - not in hardware, but mechanically (a hand controller and so forth) on the model 002 intended for horizontal flight tests. This model of Buran was tested in the Moscow region with numerous lift-offs and landings. On this model, a manual landing control mechanism was implemented, but it was never used, since the automatics worked without failures.

Comparisons were made all the time. Not only the computers were compared, but also the gyroscopic systems. We had the same amount of information about their gyroscopic devices. Only external parameters of the gyroscopic platform were known, not the internal ones. Here we did not fall behind either. If we take the entire systems - the control system on Buran vs. the avionics system on the Shuttle - it is impossible to make an informed, accurate comparison, because there was no information anywhere on the specific tasks carried out by the Shuttle avionics system. Based on a rough idea of its composition and on information from some marginal sources, one could conclude that their avionics system did not carry out all the tasks assigned to our control system. Certainly, the avionics carried out guidance and controlled both automatic and manual modes, but that was all. What else could it do, what equipment was included in the avionics system to carry out other tasks - all this was not at all clear.

Except for the gyroscopic platform and the computer, whose functions were completely clear, all the other equipment [in the Buran control system] was developed as a complex, that is, it not only controlled operating components, but also carried out other tasks. A computer route - stretching from the cabin, where the computing system was set up, to the exchange devices, which were installed in the tail compartments of the Buran - transmitted not only information necessary for the control of aerodynamic surfaces on landing, but also data required for the control of engines, for the control of small-thrust nozzles, and so on. Therefore, it would be incorrect to compare our route with the route in the Shuttle if it did not serve the same function.

Gerovitch: How was the problem of the division of functions between the human and the machine resolved at different periods? How did the functions of onboard computers evolve?

Priss: If we talk about the first machines, in the case of the Mars probe, it was a fully automatic spacecraft. The N1 carrier was also an automatic system; the orbital ships were supposed to have a pilot and manual controls, but unfortunately, as I said, they did not get a chance to be tested. The only case about which I can say anything is Buran. All the systems designed by our firm before or after Buran are automatic systems. In some projects carried out by the Rocket-and-Space Corporation Energia, on orbital stations, some functions were assigned to the crew. I do not know the details. I am not aware whether any such functions related to guidance.

As for the Buran and the lunar ship, they had the so-called "manual control loop." The implementation of such loops in the presence of an onboard computer was limited to the following. On Buran, the manual control system, which included a steering wheel, hand controllers, and pedals (like on the Shuttle), had two modes of operation: the landing mode (with tasks similar to aircraft landing) and the orbital control mode for docking and undocking. There were different types of controls for these tasks: the aviation controls were the steering wheel and the pedals, while the orbit controls were the hand controllers in three dimensions. Inputs from these manual controls through converters were entered in the computer. There were no special operating components for manual control; they were the same components that were used in the automatic mode. It would have been ridiculous to put additional ailerons or vertical or horizontal rudders [just for manual control].

The main question is this: what algorithms are used to process the information received from manual controls? There are two possibilities here. The first is this: no function is assigned to the computer, except for simple transmission, that is, whatever the pilot has done is transmitted to operating components one-for-one. The second variant is to limit his possibilities. Say, not to let him give a silly instruction or an untimely instruction. The latter variant is in fact chosen. There are some restrictions on the pilot's actions.

Another issue is presenting information to the pilot. During docking, he must have a visual image of the situation to be able to aim correctly. Here, too, there are several possibilities. First, he can make observations through separate optical devices (for example, on Buran, there was a pilot's sight and some other optics through which he could observe independently from the computer). Second, he should receive information about the status of onboard equipment, during the landing you must show him the landing field, and he must have arrow indicators, as in an automobile or on an aircraft. The latter task [of information display] was entrusted to the computer. This information came from inertial system sensors, from the air-velocity measuring system, and from telemetry systems, and it fed into the machine. The machine needed this information for automatic control. But besides this use of information in the machine, it was also displayed for the pilot.

The Americans assigned a greater role to the pilot than we did, and the pilot had more opportunities to observe the situation. In the Shuttle, for example, he even had some indicator panels above his head. In our case, the setup was a bit different. It was decided from the very beginning that we would design a shuttle capable of working both in automatic and manual modes. Therefore, everything had to be automated for the automatic mode. And so it was done. For example, an automatic landing system was implemented. Then, when this implementation was combined with manual control, naturally, an idea arose: since everything is already in place, the pilot’s role may be limited. Indeed, the machine thinks faster, it executes commands and evaluates the situation faster. The Americans faced a different situation, and for this reason they assigned wider functions to the pilot.

On a lunar complex we had manual control, because landing on the Moon was to be performed manually. There were special controls made, information display units, and so on. All this was done through the machine, in the same way.

Gerovitch: Was the control system for the lunar ship completed?

Priss: It was completed; it passed almost the entire cycle of testing on the ground. It was ready for operation.

Gerovitch: Let's come back to Buran. If the computer determined that the pilot was making a potentially dangerous move, what happened in this case? Would the computer cancel this action?

Priss: No, the computer simply limited the action.

Gerovitch: Say, the pilot could not make a turn by more than a certain number of degrees?

Priss: He could not make it greater than the number that the machine accepted.

Gerovitch: That is, the computer physically limited possible changes?

Priss: Yes, it did.

Gerovitch: Was there anything the pilot could do to override it?

Priss: He could not pass the preset limits. He could operate only within the limits set by the machine, or, more precisely, by the developer of the algorithm. The pilot could do something else. He could manually switch to a different control mode. All the modes were automated and had a specific algorithm (that is, a program). If he had serious reasons, he could, for example, cancel docking. Say, he is on a shuttle which is docking with a station in the active mode, and he receives information - either from the station, or from the ground - that for some reason the docking cannot be performed, while the docking has already began. He can manually cancel this mode and switch to another. He can choose the control mode. This switch could be performed either automatically according to machine criteria, or by a command from the ground, or by the cosmonaut.

Gerovitch: That is, the cosmonaut could only switch between the modes, but could not turn off the computer altogether?

Priss: In principle he could turn it off, but this did not make any sense.

Gerovitch: Who was doing programming for the onboard machine on Buran?

Priss: This is a very broad issue. If we talk about small projects, such as the Mars or the N-1 (in terms of computing and programming tasks, those automatic projects from today's perspective were not so big), then the development of mathematics was done as follows. There are three circles of programs for the machine.

The first circle is a set of tools for the machine itself. The machine must have an operating system, which directs the computing process: how to address various types of memory and how to receive and send out external data; there are also channel programs, input-output programs, and so on. All this mathematics does not depend on the task carried out by the machine. It must control the execution of any task assigned to the machine. A special group of programmers is assigned to create this set of tools. They are “pure” programmers who must know the computer, its system of instructions, and the interfaces very well. One must add to this set also the algorithms of flight mission input, flight mission control, and exchange with other machines (if this is multi-machine system, then some computers must be active, and some passive, and so on). This set of [general] programs is required for every type of computer. If I shift this machine from one set of [specific] tasks to another, this set of [general] programs will remain the same.

The second circle comprises guidance tasks. This includes the processing of information from various sources (inertial systems, optical systems, and telemetry systems). Then the computer solves the guidance problem, taking into account various dynamic characteristics of the object of control: inertial and weight characteristics of the spacecraft, the characteristics of various operating components, dynamics requirements, and limitations imposed by the rocket design (for example, there are limitations on rotation angles and rotation speeds). The programs that we include under the guidance rubric, as a rule, are developed by programmers from the main contractor organization that designs the control system. If we design the control system, then these are our programmers.

Gerovitch: Are these programmers trained as engineers? Do they understand the logic of the control system well?

Priss: Certainly. They design it! They design guidance algorithms. For example, one can direct a flight along a fixed trajectory, in which case any deviation from this trajectory is detected by sensors, and then the spacecraft must return to the straight path that you have planned beforehand. This is yesterday's technology; nobody does this today. Today science has developed the technique of terminal control: it does not matter how I fly; what is important is that I arrive in the final destination with the given speed, angular parameters, and so on. This means that I don't have to force the product, to twist it this way and that; I save fuel and power; I don't have to make an extra durable spacecraft that would resist the wind. If the wind carries you away, you don't have to fight it; you can correct for it later. This is the meaning of terminal control.

Programming is done by people who know, understand, and can do all of this. Besides, they know the computer and its operating system; they know the capacities of the machine. They try to solve problems without spending much memory or processor time, that is, they try to make an optimal use of its resources. As initial data, they use specific trajectory requirements set by the developers.

There are some exceptions to this approach. For example, V.A. Trapeznikov's institute, that is, the Institute of Automation and Remote Control [currently the Institute of Control Problems], develops mathematical models of complex dynamic processes for large rocket complexes. In particular, they develop algorithms for regulating the propellant component ratio for rocket engines. This line of research was started there by the academician B.N. Petrov. They do the initial development of such algorithms and then transfer further development to our specialists, who link these algorithms with their own algorithms that relate to motion. For example, if an engine is regulated, then velocities also change, and these algorithms must be coordinated.

The third group of programmers deals with the issues of control of all other flight-related processes, with the control of other onboard systems. Such issues also have two parts. Say, a developer of the control system formulates an algorithm of engine control and requirements for the separation [of stages]. For example, there are certain strength requirements for the separation. Here the third group of programmers must come into contact with the second group. For example, the task is to turn off the engine when a certain velocity is reached. Velocity is measured by dynamics specialists, but they don't know how to turn off the engine. This task is not a simple instruction but a whole cyclogram. The dynamics specialists only issue a command to turn off the engine, while the entire cyclogram - all the subsequent operations with the engine, with separation, and so on - is done by integration programmers, who know little about motion and dynamics but understand the logic of operation of the system as a whole.

It was precisely at this stage that the developers of Buran faced tremendous problems caused by the very scale of the task. There were 52 onboard systems that had to be controlled by the computer complex. Some of them were part of the control system, and some were not. Take, for example, the telemetry system. It has its own chief designer who creates the hardware and knows how it should function. Telemetry must be turned on and off, it may work in different regimes, and so on. Only its chief designer knows the algorithm of working with this system, but it is the computer that will actually be operating it. This means that there must be a separate program for the telemetry system. To find a place for this program and to agree upon its interaction with other programs is a task of the integration programmers working for us. Then a question arises: who will write this specific program? On simpler spacecraft, where we faced relatively few such problems, we used the following method. The developer of a system draws a conventional cyclogram, without translating it into a software program, and then our integration programmers translate it into a program. This method always resulted in discrepancies, misunderstanding, and errors, and several iterations were needed to correct and debug it.

It was not feasible to use the same method on Buran: in this case, the design stage would have taken 20 years. For this reason, high-level programming languages were created that enabled the developer of a system to write up an algorithm without knowing any characteristics of the computer, its system of instructions, or any other internal requirements of the machine. He would write an algorithm in a simple, clear language, and then a programmer on the basis of this initial material would write a program for the machine. During the design of the Buran, the translation of this algorithm into a program was done mainly by hand with the help of some semi-automatic tools. Today new systems exist that can process this information and produce a complete program as a result. After that, this program, of course, has to be tested: first on specialized simulators, and then on integrated simulators. In my recent article, I have included a list of organizations involved in this project as part of the Buran program [G.M. Priss, "Nekotorye aspekty razrabotki sistemy upravleniia 'Burana'," Aviakosmicheskaia tekhnika i tekhnologiia, no. 3 (1999): 35-42.].

Gerovitch: Did the Institute of Applied Mathematics take part in this project?

Priss: Yes, they participated most actively. The academician A.N. Tikhonov supervised this project; M.R. Shura-Bura was among his top aides. They worked together with us, since our specialists knew the specific characteristics of the computers. They created the PROL language for programming this type of tasks - not guidance problems, but logical problems.

Gerovitch: Could you tell more about the Bisser computers?

Priss: Bisser-1 did not develop beyond the design phase. Bisser-2 was designed in the late 1970s-early 80s, and until recently it was installed on Zenit-2 rockets. This machine was used quite widely and was installed on many different models. The development of Bisser-3 went in parallel with the work on Bisser-4 for Buran. These two machines were principally different. Bisser-3 was also widely used. Today it is installed on Zenit-3, on the boost stage of the Sea Launch. It also flew on the Fregat boost stage; there have been four joint launches with the French.

Bisser-4 stood apart from these models; it was an original machine. It was used only on Buran. Its main feature was that each channel was structurally autonomous, that is, each channel was a separate machine. It was possible to combine as many of them as necessary. There were four machines in one complex and four machines in the other complex, and a special external logical switchboard implemented the majority vote output.

Gerovitch: Would it have been possible to install five machines?

Priss: Yes, it would have. The only thing that would have change is the logical switchboard so that it could process information from five sources. It would have also been possible to install only three machines. Our choice of this logic was influenced by the American structure. They have a fifth machine, and it works according to its own program. But since it uses a different program, it is impossible to apply the majority vote rule. This principle requires a comparison of similar information from all sources and suggests choosing the outputs that are identical. For example, out of three machines two results must be identical; if one output is different, then I simply stop paying any attention to it and take information from the other two. Out of four machines, three results must be identical. It is possible to build more complex logic: to compare pair by pair and so forth. But if the fifth machine has its own program, it cannot be part of this process. Before producing the final output, all the computers in the Bisser-4 complex exchange information among themselves. This way the reliability of this system sharply increases.

In all current models Bisser-3 is used. The next step is Bisser-6 machine on integrated circuits of higher integration scale, newly developed by organizations in Minsk and Zelenograd. This computer has improved communication with external users, a higher speed, and a different type of memory. So far this machine has been used in ground testing complexes. It is hardware- and software-compatible with Bisser-3 (with small adjustments). When the testing is completed, it can be installed on all models where currently Bisser-3 is used. Bisser-6 is cheaper, its elements are smaller, and it is lighter and more power-efficient. It is our next step.

Gerovitch: Does the Bisser-6 complex operate on the same principle as Bisser-3, that is, its computers are integrated, rather than autonomous?

Priss: Yes, that's right.

Gerovitch: Let's go back to Bisser-4. Say, one of four machines produces an output that differs from the others. Would this machine be turned off completely, or only its output during this one cycle would be discarded?

Priss: Only the output during this cycle would be discarded. This could be a sporadic malfunction, and the computer may operate normally in the future.

Gerovitch: Would this machine be included in further operation on a par with the others?

Priss: Yes, of course.

Gerovitch: Were there any computer malfunctions during the Buran flight?

Priss: On Buran there was not a single malfunction - not only in the computer, but in the entire control system. The entire prelaunch testing cycle was repeated twice. The first launch of the Energiya booster with Buran was scheduled for November 30, 1988. But they called it off right on the pad. The system did not take off because the mechanism for retracting the targeting optical devices on the booster malfunctioned; I already talked about it earlier.

Today optical systems are not used during launches. Our gyroscopic platforms have a gyro-compass mode, that is, the platform itself can determine the cardinal points; there is no need for geodesic or optical devices. Wherever the platform is, it will determine its location and direction of movement. The Americans did not have that, and they had to do aiming. The matter is that the Americans do not have a booster rocket separate from the Shuttle. They have solid-fuel boost engines and a tank that works with the Shuttle engine. Therefore, a control system - the “brain” - is installed only on the Shuttle. They had three gyroscopic platforms, and it was necessary to aim them. For this purpose they used a ground optical system. But they were able to put their platform on the left side of the cabin and somehow managed to establish an optical link with the ground to aim correctly.

We have not been able to place our gyroscopic devices on the Buran in a similar way. This happened not because our platform is bigger; in fact, it is rather smaller than the American one. The matter is that the requirements for automatic operations - including such operations on the orbit as docking, determining the location, and so on - resulted in expanding the structure of the control system with a whole complex of automatic optical devices: stellar-solar gauges, navigation antennas, and so forth. It was necessary to be able to adjust their position with relation to the platform. All our gyroscopic devices were installed not in the front, but behind the cabin, in the cargo compartment. Three platforms (a triple-redundancy system) and all optical devices were put on a very powerful adjusting plate. But the cargo compartment is closed! It opens only on the orbit. And they were not able to make a slit to reach these devices. At that time, we did not have platforms with the gyro-compass mode. We needed an optical orientation to set the initial azimuth, but this could not be done. Therefore, we had to do optical orientation for the platform that stood on the rocket.

Our Energiya rocket had its own control system, separate from the shuttle (which the Americans did not have), and therefore Energiya could be launched without Buran. Its gyroscopic system had to be aligned. A ground optical station aligned it with the help of special optical devices. We had to aim two beams at each platform, and there were three platforms in total. On the N1 we solved the same problem. An optical alignment device was created, which was needed only on the ground. It was not needed in flight, because the rocket had already been aligned and took off. On the N1 we simply dropped this device; it fell on the ground and shattered into pieces. It could be used only once. This was a waste, and we decided to preserve such devices on Energiya. On the N1 this would have been impossible: the devices were placed too high, the service trusses moved away, and there was nothing that could pick up those devices. With Energiya, an opportunity presented itself. Energia was much shorter, smaller than the N1. A special truss construction was designed that supported those three devices. It was fastened to the rocket, but at launch it would automatically separate by way of pyrotechnical locks and move away from the rocket.

During the first attempted launch on October 29, 1988, this system did not work. More precisely, the disconnection occurred, but there was a tilt, and a confirmation that it worked did not come through. The launch was aborted. But up to that moment all preparations on board passed without a flaw. Then the fuel tanks were emptied, and another attempt was scheduled in two weeks, for November 15. We repeated the entire routine. This time this gizmo worked and moved away, we reached the orbit, flew two orbits, and then began slowing down and descending. At this stage, there were other difficulties. We landed in the automatic mode. There was a side wind of 17 meters per second, but we landed within three meters of the mid-line of the landing strip. Not a single flaw! There was none - either at the take-off, or during the ascent, or during the flight.

Gerovitch: What was Sergei Korolev's opinion on the prospects of using onboard computers? He evidently had contacts with the design bureau of Philip Staros and Joseph Berg in Leningrad and asked them to develop an onboard computer in the mid-1960s. Some people contend, however, that he was skeptical about onboard computers. Do you know what his opinion was? To what extent could his opinion have influenced Pilyugin's decisions?

Priss: Pilyugin always was on very terms with [Sergei Pavlovich] Korolev. They were friends, and therefore Pilyugin always listened to Sergei Pavlovich's opinion. They never disagreed on principal issues. They had some disagreements on specific questions. They often had quite harsh and vigorous disputes. But Korolev died in 1966. It was just the beginning. I was pretty close to the these circles, and I never heard of Sergei Pavlovich's negative attitude toward computing.

I am not aware of the contacts that you mentioned. I can only make the following assumption. At that time Sergey Pavlovich's organization had already formed a large division under the direction of Boris Yevseyevich Chertok. They too were designing control systems. This happened for the following reason. When Korolev started the development of manned spacecraft, he offered Nikolay Alekseyevich [Pilyugin] to build control systems for them. From the point of view of control, satellites are very different from manned spacecraft; satellites are rather passive objects. Before that, nobody else designed control systems. In the late 1950s, however, a firm in Kharkov was created to build control systems for the Chief Designer Mikhail Yangel. In the same period, a special design bureau under the direction of Nikolay Semikhatov was set up in Sverdlovsk; they began doing the same for the Navy. And the directors of both firms were former employees of our organization.

The period when Sergey Pavlovich offered Nikolay Alekseyevich to participate in this project was very difficult for us in terms of work load and also personal relations. Gagarin flew in 1961. His spacecraft was a primitive device. He landed on a parachute; the spacecraft had almost nothing to do. Korolev envisioned a whole series of advanced piloted spacecraft, and later this was indeed implemented. At that very moment [the Chief Designer] Vladimir Chelomey began working on spacecraft, and Pilyugin was assigned to do some projects for him. Thus we did work for Korolev, Yangel, and now Chelomey. Pilyugin realized that he could not carry out all these projects at the same time, and he declined Korolev's offer.

Sergey Pavlovich began working on this problem in his own organization. Boris Chertok, Boris Raushenbakh, and others were involved. Apparently, they coped well, and this division still exists and solves these problems with good results, which are in no way inferior to what we could do. There is, however, a difference in approach. We work by the method I described earlier: we gather everything under one roof. What they do is, in effect, a compilation. They buy a machine; they buy gyroscopic devices; they buy optical equipment, and so on. We try to build everything ourselves, except for the optical devices. Apparently they realized, just as we did, that it was necessity to use computer technology. I never heard anyone, including Sergey Pavlovich, to speak against it. He was not in this sense conservative, and I can hardly believe that he could have taken a negative stance. Both he and Chertok understood all these problems perfectly well. It is hard to imagine now that someone could have objected [to the use of computer technology].

Gerovitch: By the mid-1960s, analog technology had achieved a certain level of reliability, while the new digital technology might have raised doubts. It is often said that Korolev insisted that it was necessary to rely on well-tested technologies, to choose simple solutions without unnecessary complications.

Priss: This principle has always held true. Everyone supports it. In this sense the West is more conservative than Russian engineers. We now often face this problem. Today negotiations are going on with Boeing, which has created the Sea Launch company, about further steps, in particular, to work not just on a sea launch, but also on a ground launch. We put forward various proposals, for example, switching to Bisser-6. They replied, "No need for that! The old computer has already flown, and the new one has not." We said, "It will be well tested by the deadline!" They reply: "No; there is no need for that." I believe they have the rule that all new models must have no less than 70% of old equipment. But even in the remaining 30% one might include a computer.

Gerovitch: Thank you very much for the interview.