Computing in the Soviet Space Program

Introduction

Bibliography

Computers

Discussion

Documents

Essays

Institutions

Interviews

Links

Sitemap

• What were the basic problems of creating a computer capable of working in space?
• Were any special elements developed for onboard computers?
• Were there any disputes over the weight and size specifications for the Argon-11C?
• With which spacecraft design organizations did you work?
• Was the Argon-12S modified in comparison with previous models?
• Were the designers of Salyut onboard computers your competitors?
• Did you have any interaction with the organization in Zelenograd that developed the Salyut computers?
• Besides designing hardware, did NICEVT participate in the development of software for onboard machines?
• By what deadline did you have to deliver a machine?
• Did any software errors appear in flight?
• Was there any attempt to study American experience on a regular basis?
• On what technological basis were onboard machines created?
• How was high reliability of the Argon-16 achieved, given such unreliable elements?
• Did you develop any information display units?
• Is there any way a cosmonaut can enter information into the machine?
• Did you consult with cosmonauts how to present information to them in the most convenient way?
• Have the technologies developed for onboard computers been used in other areas?
• Did secrecy seriously limit the spread of this technology?
• Did you develop any mobile computers for the civilian sector?

Gerovitch: Let's start from the very beginning of the development of onboard computers for the Soviet space program. What were the basic problems of creating a computer capable of working in space? Was it clear that such digital machines could indeed be built? Were there any doubts whether this could be done?

Przhiyalkovsky: Certainly, there were some doubts. At first the space program did not require onboard digital technology. Analog systems were quite sufficient. However, as the space program advanced, this need did arise. The main problem was technical: how to create a digital device that could carry out the required program within the prescribed weight and size limits.

Gerovitch: Were any special technologies developed for onboard computers?

Przhiyalkovsky: No, not at the beginning. I can tell about the Argon-11S computer, which was designed at NIEM/NICEVT (the Scientific Research Institute of Electronic Machinery/the Scientific Research Center for Electronic Computer Technology) in the latter half of the 1960s. The Argon-11S was based on general-purpose hybrid integrated circuits of the Tropa-1 series. Naturally, equipment quality control in the space program was carried out by the military, and quality standards for integrated circuits were higher than usual. As a matter of fact, the functions of the first machine were not very broad. It was not a universal but a specialized computer for solving only one guidance problem.

Gerovitch: Were there any disputes over the weight and size specifications for the Argon-11S?

Przhiyalkovsky: I can't really tell, since I did not work at NICEVT at the time. I am sure those were typical disputes between a system integrator and an equipment manufacturer. The system integrator needs smaller size, lighter weight, and higher reliability. The manufacturer, on the other hand, is trying to bargain, while taking into account what can actually be delivered. Anyway, it all ended well. There were several launches in the Zond program, and everything worked. Once on a test launch the parachute system did not work well, and the entire landing module together with the onboard machine hit the ground with excessive speed, and the computer was dented a bit. It is now on display in this dented condition at our Scientific Research Institute Argon. They cut some pieces off and took printed-circuit boards for souvenirs.

Gerovitch: When you came to NICEVT, did you work directly with subsequent models of the Argon?

Przhiyalkovsky: Yes, certainly. One of the areas assigned to me was the responsibility for onboard machines. I was responsible not only for the Unified Series, but also for onboard computers.

Gerovitch: With which spacecraft design organizations did you cooperate?

Przhiyalkovsky: We worked both with the Research-and-Production Association Energia and with the Research-and-Production Association Mashinostroenie [Machine-Building] (led by Vladimir Chelomey). For Chelomey, we were designing the Argon-12S. A whole series of computers was being planned for the Salyut space station, both for the station itself and for transport ships. Chelomey's program, however, was not completed. They had a real competition with Energia. Chelomey designed an Orbital Piloted Station (OPS), while Energia was building its own orbital station. At some point, the preference was given to the Energia program.

Gerovitch: Was the Argon-12S modified in comparison to the previous models?

Przhiyalkovsky: Yes, this was a completely separate project, even though it was being carried out at the same time. Of course, we also used some standard technologies and parts, for example, the Tropa-1 integrated circuits. In this case, a special technology of printed circuit mounting was used. While in the Argon-11S double-side mounting was used, in the Argon-12S we used multilayer mounting by the method of pair pressing. In later onboard models, a new method of "rod growing" was used. The method is as follows: they take a printed-circuit board and another board on top of it with openings in specific places, and then they "grow" metallic rods through those openings. The contact between one board and the other is through this rod. This is a very labor-intensive technology, and it takes a long time, but as our experience has shown, this technology is one of the most reliable.

Gerovitch: Was this new technology used only in onboard computers or in other machines as well?

Przhiyalkovsky: Only in onboard computers. Higher quality standards applied to them. This was a Ministry of Defense contract, and they had special quality control units and special quality standards. This technology was used in the Argon-16, and look at the results: the Mir space station, on which it was installed, orbited the Earth for 14 years, and there was not a single breakdown of the Argon-16! There were some computer breakdowns on Mir and some repairs, but those were not ours.

The Argon-16 is not intended for repair; it is built for repair-free operation. Replacement of large modules is possible, but this did not happen. On one of Chelomey's Almaz space stations there was indeed one instance when our computer was repaired. The Argon-12A two-channel onboard computer system broke down. And it so happened that in one channel one module failed, and in the other channel another module failed. It was not clear why this happened. There was a hypothesis that a memory module broke down because of a solar flare. But it was not proven, since this was a one-time event. In the end we tinkered a bit and combined those channels into one. We quickly made a new connector to link working modules and shipped it to the station. The connector replacement was a simple operation, and the crew performed it successfully. Chelomey ingeniously presented the whole affair as "the first computer repair in space." Like any other "first," this was met with understanding [at the top]. This was serious business, and there could have been negative consequences for us.

Gerovitch: When was it?

Przhiyalkovsky: It was on one of the Almaz stations, Salyut 5, I think. I was already in NICEVT, and together with our experts, I searched for a solution how to restore the operation of this system.

Gerovitch: On some Salyut stations there were onboard computers also called Salyut.

Przhiyalkovsky: Those were not ours. They were made at the Ministry of Electronic Industry.

Gerovitch: Were they your competitors?

Przhiyalkovsky: No, you cannot call them competitors. Under favorable conditions, we could have handled all the computing equipment for spacecraft ourselves. But at that time there was no market economy, and we gained very little from participation in the space program, except for prestige. The first flights - Gagarin's flight and a few after that - brought the participating firms great prestige, but as the space program developed, it all came down to rush job, because when the launch date is fixed everything must be finished on time. Under favorable conditions, we could have done it all. At some point there was a decision by the Military-Industrial Commission of the USSR Council of Ministers (which let us contracts), according to which NIEM was to become the lead organization for the design of onboard computers in this country. And NIEM carried out these functions, that is, it provided standardization guidelines in order not to allow unjustified proliferation of incompatible computer types. It acted like a curator, a monitor of activity of various onboard computer developers. But later NIEM merged with NICEVT, whose primary function was the design of the Unified Series of computers for all socialist countries. NICEVT was then relieved of its duties as the lead organization for the design of onboard computers.

Gerovitch: Did this function pass to another organization?

Przhiyalkovsky: No, this function did not pass to another organization. Those are usual bureaucratic maneuvers. A top official does not really understand what is going on far below him. He would think: "The lead organization - what a big deal? Let's assign this role to somebody else!" it turned out that this was not so simple. Competent experts are needed, experience is needed, and so on. It is not so simple to pass the lead role to somebody else. Eventually everything returned to the way it had been before. The lead role was returned to NICEVT, which formulated a program for the development of onboard computers for the entire Soviet Union. There were many participants, but NICEVT was responsible for the formation of the overall program. The most important part was the standardization of computers and parts. When people learned how to make onboard computers, many started designing their own machines suited for their specific goal. As a result, an incredible number of different computer types proliferated. This caused great problems with servicing and so on.

Gerovitch: When did the lead role return to NICEVT?

Przhiyalkovsky: In about three years, perhaps by 1973. An authoritarian decision of a top official had not worked out, and it was reversed.

Gerovitch: Did you have any interaction or information exchange with the organization in Zelenograd that developed the Salyut computers?

Przhiyalkovsky: Yes, we had. We had good relations, for we did not lay claims to what they were doing. The situation was quite complicated. In the late 1970s our Minister [of Radio Industry] P.S. Pleshakov faced a dilemma: either to expand drastically the range of different types of onboard (and not only onboard) computers and to increase the production volume accordingly, or to start narrowing down the area of their application. At the beginning, in the 1960s, the Ministry of Radio Industry required very few specialized computers for its own needs. Therefore, NICEVT could do some work for the space program as well, and it all went very well. In the 1970s, however, a huge demand for computers arose among the organizations that belonged to our Ministry. In particular, we designed the Beta-2, the Beta-3M, and the MSM computers for the Ministry. Those were not onboard computers, but mobile, transportable machines, which could be transported even on caterpillar vehicles. Those were even more complex than computers for spacecraft. In space, there are no such vibrations and temperature fluctuations as in an armored troop-carrier. For various systems developed at the Ministry, we designed the Argon-10, the Argon-15, the Argon-30, the Argon-40, and the Argon-50 computers. As a result, NICEVT became overloaded with mobile computer designs, and the Minister gradually began to drop contracts with other Ministries. The Research- and- Production Association Energia belonged to another Ministry, and they were told that our Ministry would no longer make computers for them. This caused great discontent among the Energia leadership, Valentin Glushko and Boris Chertok. They could not understand our Minister's position. At that time, it became necessary to modernize the Argon-16 and to develop computers for other purposes on board long-term orbital stations. The termination of this contract also upset our engineers, the developers who had worked with Energia for a long time and wanted to continue; they had worked together and understood each other very well. Nevertheless, we carried out this order and stopped all work for the space program. And then Energia began talking to other computer designers, including the ELAS Institute [the Scientific Research Institute of Micro-Instruments (NII MP)] in Zelenograd. They developed the Salyut computers, including those for the Mir space station. Our Argon-16 worked on Mir without failures. You can see it on TV: when there is a docking of any spacecraft - a cargo ship or a manned vehicle - to the station, they show their approach on a computer display. This display is part of the Argon-16, which takes part in the control of the docking operation.

Gerovitch: Is the same display installed on board?

Przhiyalkovsky: Yes, it is part of the computer.

Gerovitch: Is the onboard Argon connected to any computer on the ground?

Przhiyalkovsky: The Argon-16 is not connected to anything on the ground; it is completely independent. It controls station orientation, docking, and separation maneuvers. On the Soyuz-T spacecraft, the Argon is located in the instrument compartment, and it burns down during the descent; it is not included in the landing module.

Gerovitch: Besides designing hardware, did NICEVT participate in the development of software for onboard machines?

Przhiyalkovsky: Yes, we developed systems software and various test programs. Specialists from Energia and Mashinostroenie wrote specific applications and debugged them on their simulators.

Gerovitch: Who wrote the operating system?

Przhiyalkovsky: On onboard control computers there was no operating system. Besides the central processing unit, the Argon-16 (as well as the Argon-11S and the Argon-12S) contains a interface module for communication with various sensors. Designing this module is a serious task. All sensors are sending data at the same time; the data is read and processed, and then a control instruction is issued. Everything works in real time without any operating system.

Gerovitch: That is, the same program is running all the time?

Przhiyalkovsky: Yes, various parts of the same program.

Gerovitch: By what deadline did you have to deliver a machine? How long before the launch was it necessary to hand a finished computer over to the guidance system developers so that they could write their specific program? Was it possible to change anything after the submission?

Przhiyalkovsky: The first machines had to be finished long in advance. After that, we could change only small details by permission in order not to mess up the program. Space engineers needed a year or two for software debugging on a simulator before the launch. During that time onboard machines for other space ships were being built; we launched a whole set of computers into production. Later on, in order to shorten the application development span, they used simulation on universal computers. Since 1973, our onboard computers began using the Unified Series architecture, and application debugging became much simpler.

Gerovitch: Were any software errors revealed in flight?

Przhiyalkovsky: There were no big errors in the software for onboard computers for spacecraft. There was only one case in which, I suspect, there was a software error. If it were a hardware defect, we would have had a lot of trouble. A special committee would have been appointed, an investigation would have started. It's serious business. There was nothing in our line. From this I can conclude that it was the fault of mathematics [i.e., software algorithms]. Here is what happened. A spaceship approached a station. They came close, and the docking system's radar (made by others) showed "capture," that is, it showed that the ship has found the station's docking port. The Argon-16, which turned on orientation and other systems, received this signal. I was at that moment at the Mission Control Center. On the screen the space ship was very quickly approaching the station. The flight director on the ground asked, "Isn't your speed too high?" The cosmonaut replied, "Yes, it is. Perhaps, I'll slow down." Possibly, he lowered the speed more than was necessary. But the work of automatic systems had already begun! As soon as he did it, the Argon display suddenly began to flash: "Alarm! Alarm! Alarm!" There was no time to investigate, and they gave the order: "Manual docking!" And he docked the ship to the station manually. Then they began investigating what the problem was. In the end, there was no breakdown: the machine was fine and everything worked. For myself, I figured that when he lowered the speed he possibly left the capture zone, the capture was lost, and this was interpreted by the machine as a malfunction. This was a false alarm. This happened on one of the first launches of Soyuz-T. Was this the true reason or not, I did not find out. In such serious business people are very cautious. Specialists from Energia then took appropriate measures. There were no other incidents.

Gerovitch: Is the Argon-16 still used on spacecraft?

Przhiyalkovsky: Yes, even now the Argon-16 flies. It really boggles my mind how they do it. There is a great shortage of necessary parts. Those Argons are made of gold! They are individually made. I've been told that the Argons have been approved for installation on cargo ships for the International Space Station, which is flying now. This is a joint program, and Energia works together with American companies. The Americans, I was told, looked at it and said that it was surely bulky but extremely well-tested, so it was worth keeping.

Gerovitch: When the Argons were being designed, was there a requirement that they meet or surpass the level of comparable foreign models?

Przhiyalkovsky: There were general requirements of this kind, but we did not really know any specific details about foreign onboard computers, and we proceeded on our own.

Gerovitch: Was there any attempt to study American experience on a regular basis? The work on the Apollo Guidance Computer was not classified, and information about it was published. Was there any discussion of how to use this information?

Przhiyalkovsky: And how could it be used? In practical terms, it is impossible to copy a computer. Let me cite another example. At some point we adopted the IBM-360 architecture (and later IBM-370) as the basis for the Unified System of computers for socialist countries. This architecture was adopted because it could unify everyone: we did not have to argue which architecture is better, ours or yours, Bulgarian or Hungarian. Everyone agreed. Many specialists in this country were very unhappy; they believed that we needed to develop our own architecture. To this day we hear accusations that we have allegedly copied the American series. It cannot be copied! In order really to copy a machine, it is necessary to copy the entire industry required for the development and production of this machine. It is necessary to have precisely the same integrated circuits, precisely the same sockets, materials, technologies, and so on. It is necessary to copy the IBM company with its entire manufacturing facilities in order to make the same machine. It is only possible to borrow the architecture, that is, the logical structure. By the way, it is open, it is not patented.

In the same way, it is impossible to copy the Apollo Guidance Computer. It would have required copying too much: all the equipment, all the technology. And this does not make any sense. We must design computers with our own parts and our own industry in mind. We did not take from the Apollo even a general idea, nothing. There was no need: we knew how build small control devices. The main problem is to make a machine that would fit its specific place on board within the required weight limits and the required conditions (temperature variation, vibration, and so on).

Gerovitch: What was the technical prototype for an onboard computer? Did mobile computers for aircraft serve as a prototype? Where did this ideology come from?

Przhiyalkovsky: No, space onboard machines were completely independent. The work on computers for aircraft went in parallel. In the case of aircraft, the task is even more complex: there are greater vibrations and, most importantly, huge temperature variations. In onboard machines the greatest effort is spent not on computing but on how to make the computer withstand the climate, vibration, acceleration, and so on. Appropriate tests must be carried out, and they take a long time. Take the Argon-16. When it was already flying on spacecraft, for technical reasons we needed to replace the production method of "rod growing" with a more advanced technology of metallization of openings. In the latter, the boards are connected not by a continuous rod but through the metallization of the walls of the opening. The opening remains open, and it is possible to insert a leg of an electronic component and to solder it in. This technology is more advanced; it is accepted all over the world, but it is somewhat less reliable, since these walls may sometimes break apart. The question whether to apply opening metallization in the Argon-16 was thoroughly discussed. The Argon-16 is manufactured in huge series, hundreds of copies. Eventually we came to the conclusion that this was impossible for one simple reason: in order to use the new technology, we have to repeat the entire cycle of tests already done on the Argon-16. This is simply unfeasible, for it would hold us back for a long while. Very serious, long tests are required.

Military technology in general is very conservative. It is extremely difficult to change anything because of very laborious testing. Such tests usually require the participation of many other organizations. It takes a huge amount of work to get a machine admitted to flight tests. And then the flight tests come - first unmanned, then manned flights.

Gerovitch: I have been told that the Tropa elements, on which the Argon-11S was based, were very unreliable.

Przhiyalkovsky: Yes, that's true. The Tropa and the Posol, which followed it, were hybrid elements, like those installed on the first series of IBM-360 machines. In this case, micro-transistors were soldered to the chip, that is, the chip was not monolithic, but it consisted of separate, independent micro-transistors. Since soldering is not the most reliable technology for electric contact, the reliability of these elements was insufficient. For this reason, they were quickly replaced by solid-state integrated circuits TTL and ECL, which allowed for a fast increase in the scale of integration on a microchip. Small-scale and medium-scale integrated circuits of the TTL type remain now only in the Argon-16, since this computer has not been suitably modernized. Today the weight and size of the Argon-16 could have been reduced by an order of magnitude, but its modernization was not carried out in time. The greater the scale of integration, the more reliable are integrated circuits.

Gerovitch: How was the high reliability of the Argon-16 achieved, given such unreliable elements?

Przhiyalkovsky: In this case, reliability is ensured by structural methods. The Argon - 16 is a triple-redundancy system. Information passes simultaneously through three channels, and it is processed step by step. On the first level, the information is processed in three channels, and the results are passed on to a comparison unit. If all three results are identical, then in all three channels the information is passed to the next level. If one result is different, it is discarded, and the other two are passed to the next level in all three channels. Therefore, even if breakdowns or malfunctions occur in completely different places, the end result turns out to be correct.

Gerovitch: Are all three channels included in a single computer?

Przhiyalkovsky: Yes, there is only one computer on board: a triple-redundancy machine with majorization. On each level, there is triple calculation. If a malfunction occurs in one channel, the machine corrects it, the information flows around it, and triple calculation is resumed. During a long term of operation, this system accumulated up to 50 malfunctions in various places, but it still worked. This system proved extremely reliable. But it required a large size and weight. These days, perhaps, there is no need for it, but back then, with the Tropa elements and the small-scale and medium-scale TTL circuits, there was no other way to ensure the required degree of reliability.

Gerovitch: Do onboard computers today still use the Tropa?

Przhiyalkovsky: No, only the Argon-11S and the Argon-12S were based on the Tropa. The Argon-16 uses small-scale and medium-scale TTL circuits. I cannot even imagine where they obtain those circuits now. The Ministry of Electronic Industry has terminated its production, and this technology is long forgotten. It's the same with vacuum tubes - they are now worth their weight in gold!

Gerovitch: How was the problem of reliability solved on the Argon-11S?

Przhiyalkovsky: The Argon-11S was not intended for long-term operation. Circling the Moon took only a few days. One processor was enough to ensure its reliability. Chelomey's Almaz was intended for a much longer term, and the Argon-12A, built for the Almaz station, had two processors.

Gerovitch: In case of a single backup, how can you decide which channel malfunctioned, if two channels produce different results?

Przhiyalkovsky: The Argon-12A worked with a different guidance system. If there is a breakdown, then the testing system turns off the broken channel, but if ithere is just an individual malfunction, then the result is simply discarded, and the curve is smoothed out. Most often only simple malfunctions take place, just fleeting incorrect results. Guidance always follows certain regularity, taking into account, say, the laws of guidance, and if a single point suddenly jumped aside from the curve, then it is discarded by logical reasoning. The program already knows within what approximate range the result should fall.

Gerovitch: From the outside, the Argon looks simply like a box with sockets for connections with sensors. Did you develop any information display units?

Przhiyalkovsky: For the Argon-11S this was not necessary, since those were unmanned flights. In the Argon-16 there is such a unit. It is a small, light display, which shows only necessary information. During docking, the data from its display is transmitted to the ground, and it shows speed, angles, distances, and so on.

Gerovitch: Does this machine have a keyboard?

Przhiyalkovsky: No. One can't intervene in its operation.

Gerovitch: Is there any way a cosmonaut can enter information into the machine?

Przhiyalkovsky: This is not recommended. There is simply no need for him to do it. The Salyut computers have both a keyboard and a display. The Salyuts are similar to personal computers in terms of functioning; a cosmonaut can work with them directly. Initially the Salyuts were not needed; the Argon quite sufficed.

Gerovitch: During the development of the Argon-16, did you consult with cosmonauts how to present information to them in the most convenient way?

Przhiyalkovsky: No. We had sufficient contact with specialists from the Energia Association, and dealing with the cosmonauts was their problem.

Gerovitch: Were all such issues resolved inside NICEVT?

Przhiyalkovsky: Inside NICEVT together with Energia. The Argon-16 is included in the control loop for the orientation of a station or a ship. There is nothing for a human to do in there. The cosmonaut has means for guiding a space vehicle, but he does not operate through a computer. The computer only monitors parameters of the station and controls its orientation without any coordination with the crew on board; the computer only shows these parameters on its display.

Gerovitch: Is any information transmitted from the computer to the ground?

Przhiyalkovsky: In the Mission Control Center, there is a huge screen which shows the same picture as seen by the pilot.

Gerovitch: Can Ground Control send any instructions to the Argon? Can they turn it off, for example?

Przhiyalkovsky: Yes, they can.

Gerovitch: That is, the only thing that the pilot can do with this computer is to turn it off?

Przhiyalkovsky: Yes. Only why would he do it? With other computers it happens sometimes that the program is changed. And here the program cannot be reloaded. Here it is well-tested and hardwired. Nothing can be done with this program; it will be carried out blindly.

Gerovitch: Have the technologies developed for onboard computers been used in other areas?

Przhiyalkovsky: No, they are narrowly specialized. Perhaps, this is wrong; perhaps, they could have been used. If it were today, this may have happened differently, but back then we did not need this. First, secrecy was hanging over us. Furthermore, how could they be used? Those machines were too specialized. What would you operate with it - a tractor?

Gerovitch: Did secrecy seriously limit the spread of this technology?

Przhiyalkovsky: No, technology itself was not classified. The problems that this computer was solving were classified.

Gerovitch: As you said, the algorithms of problem-solving had been hardwired. Was the computer as a whole classified?

Przhiyalkovsky: This computer could not be sold to just anybody. There was also another aspect. The Ministry of Defense acquired a taste for computers and demanded that they be installed on a whole bunch of [weapons] systems. To fulfill such intense demand, unification of tasks was needed. For example, the Argon-15 was intended for two types of systems: it can be used both as a mobile computer, and as an airborne computer. As a rule, mobile, airborne, and marine systems are completely separate. They have different requirements - mechanical, climatic, and so on. The Argon-15 worked in more than 50 systems of different kinds, including those on wheels, on caterpillar tracks, in air defense systems, on board an aircraft, and so on. It really sold well and spread widely, but only for defense systems. It did not go into the civilian sector, because it is too expensive.

Gerovitch: Did you develop any mobile computers for the civilian sector?

Przhiyalkovsky: No, not the mobile ones. For the civilian sector, there was the Unified Series (ES) of stationary machines and the Small Machines (SM) series of the Ministry of Instrument Building (Minpribor). When the SM series was being planned for all socialist countries, we held endless talks and in the end decided to take the DEC architecture as the base. We designed special models for factory shops, for harsh environments. If a mobile computer were needed (for civilian use), they would have just taken the military model, the Argon-15. The problem was that it was too expensive.

Gerovitch: Was it because of the high reliability standards set by the military?

Przhiyalkovsky: Quite right.

Gerovitch: Thank you very much for the interview.

See also Viktor Przhiyalkovsky's essay Scientific Research Center for Electronic Computer Technology: The Early Years, 1968-1975