Translation of a seminar on the development of the JA 37 Viggen's avionics

[renhanxue]

I've posted this on SomethingAwful before, but here's a repost for those of you who haven't given Lowtax tenbux. Warning: it's very long.

The Swedish original is called JA 37: pilot och system and is a transcript of a witness seminar held at the Stockholm Museum of Technology in 2007, with a lot of old geezers who had worked on the development of both the Viggen and other Saab aircraft.

A very brief glossary:

- FMV is the Defense Materiel Administration, which was (and is) the government authority responsible for buying and developing all kinds of equipment for the Swedish armed forces. At the time when the Viggen was being developed, FMV still had a lot of engineers in house, which was considered important since it was thought at that time that in order to buy or develop a good system you had to know how it worked.

- SRA is Svenska Radioaktiebolaget (Swedish Radio Company). Acquired by Ericsson in 1983.

- LM Ericsson is just Ericsson. The name comes from the full formal name of the company, "Telefonaktiebolaget L M Ericsson", which in turn is named after its founder Lars Magnus Ericsson.

- JA 37 is the fighter version of the Viggen.

- AJ 37 is the strike version, which predates the fighter version by about ten years (entered service around 1970; the JA 37 around 1980).

- 35, or aircraft 35 is the Draken.

- 37, or aircraft 37 is the Viggen airframe in general.

- 39, or aircraft 39, or JAS 39 is the Gripen.

My comments are in parentheses.

Please ask away if there's anything that seems unclear.

 

Gunnar Lindqvist (technical lead at FMV for the JA 37 project): In the early 1960's, work on aircraft 37 was initiated, and it was going to be a "main battle aircraft". That is, it was going to be a multirole, just like the Gripen is today and like aircraft 35 also kind of was, although with different versions of a common airframe. For various reasons, maybe because the Draken F existed and was equipped with decent missiles, the strike role dominated early on. That was sort of unfortunate, because it's the fighter role that puts the greatest demands on a combat aircraft. That meant a great deal of modifications had to be done to turn the strike version, AJ 37, into the fighter version, JA 37. A completely different targeting system using a pulse-Doppler radar was required, the aircraft's agility had to be improved and new weapons had to be introduced, some of which - like the radar-guided AA missile - were also based on Doppler radar. And in addition to that, further modifications or modernizations of other avionic systems. That added complexity, while at the same the aircraft's reliability had to be improved.

Many of these changes meant the utilization of digital technology, especially in the targeting system. But other systems were also digitalized, down to the engine control system. This was possible in part due to foresighted research in key areas in electronics, and also in part - of course - due to the inherent know-how in our aircraft industry. Maybe also in part due to a good working cooperation between the customer - FMV - and foreign government authorities and subcontractors.

[...]

Bengt Sjöberg (then at Saab): It's hard to entirely ignore the AJ 37 heritage. Because really, it was in the early 1960's that we pioneered the use of computer technology in an aircraft. And that was a time of "sturm und drang", I'll have to say. The ice was really thick in some places, especially among the aircraft engineers down at Saab. They were thoroughly suspicious towards electronics. I think they might have had some bad experiences a while back with aircraft crashes due to unreliable electronics. And a computer, that was something extraordinary, that seemed very unpleasant and dangerous.

But we did have some enthusiasts that understood the emerging technology. There was some experience with a transistor-based machine, SANK ("Saab's automatic navigation calculator", completed in 1960 and intended for missile guidance) that people had prepared themselves with and that we encountered in the beginning, when I came to Saab in 1960. It could fit on a tabletop at least, unlike BESK that took up a whole room. And we had some preliminary ideas back then, but we still couldn't fit it into an aircraft. But then came the development of integrated circuits. Viggo Wentzel and Bengt Jiewertz at Saab were all over those things and traveled around the US and picked up components straight from the labs. We could see the light of dawn in regards to being able to pack a decently powerful "generic calculator", as we called computers back then, into an aircraft.

And we tried to help in various places. I was thinking about how you could use a doohickey like that in an useful way and wrote up something about it. I talked a lot with our operations and tactics people about mission planning and what you need on board a combat aircraft. In general we believed early on that it was mainly about navigation and maybe some aiming aids, because we had done some work on gun sights and the like. But we came up with a long list of functions that we thought seemed important and essential, especially in a single-seater where we were going to get rid of the systems operator/navigator. And I wrote some documents about this.

There were a lot of different opinions about how much of this would be possible in practice. Our prototype was called NSK for a while, and a lot of people thought that stood for "numeric sighting calculator". But I thought it was better to call it "numeric combat calculator". Later it would eventually become CK 37 ("central calculator 37"). And we were fully occupied with getting even a single generic computer into the system. Some, especially at the avionics bureau at FMV, were of the opinion that we should have two computers - one that they would use for navigation and one that the weapons bureau could use for targeting and weapons - but I opposed that as much as I could.

Because the technology at that time had one great weakness, and that were the input and output systems. Analog-digital conversion, that was breaking new ground and we had to build from scratch. We thought that once we had gotten the signals into the computer there it was very easily possible for different functions to communicate with each other. But getting in and out of the computer was a big problem. And I remember when one of the computer engineers managed to run an analog simulator linked to SANK, that was a big deal. We didn't know if anyone in the world had managed to do that kind of link in real time, but at least in Europe I don't think there was anyone. But we succeed and started to believe in it, just to give you an idea just how tentative and fumbling everything was. Everything was new.

Later, we built what would become CK 37 with integrated circuits. Saab was a huge customer at the US manufacturers during the years when CK 37 was in development. The fact that it was one central computer wasn't just a device issue, it was also a philosophical issue. Because we gathered so much information in a central device that could be managed with software, we got a firm grip of all the important functions in the entire avionics system. We organically grew a personnel organization where, in a single corridor with everyone just a few steps away, we had full control over the development of the entire system. And that turned out to be very important for the future. We sort of inherited this central computer from AJ 37 to JA 37.

Those of you are interested can take a look in these pictures handed out. My memory isn't what it was and I have to look a bit at it before continuing.

(I guess I should translate this at some point)

Well, it's a small diagram showing the avionics system when we got started with the JA 37 project. In the center there's a pretty big central computer. It's not big because it's physically big but because it keeps then entire avionics system together. And these central computers, both in the AJ and the JA, they came to be filled with a lot of what we could call "pilot support functionality". The cooperation with the pilot went from input data via the computer and to the presentation equipment, and from the pilot's control equipment to the various parts of the system. The data traffic was coordinated. One thing we learned from the AJ 37 was what a surprising number of logical operations that grew up during development. We re-did the instruction list in the CK 37 a number of times during the 60's. We had four separate prototypes before reaching CK 37.

All of this grew organically during violent but friendly altercations between us and the computer engineers. We discovered the importance and the computational load of various logical operations and mode selectors. We built those into the computer, which became very good at managing logical operations with a small memory footprint. And this central computer was the pilot's partner, you could imagine the pilot's brain working in parallel with the central computer.

If you look at the diagram, you'll see some boxes with doubled border, those are the functions that became computerized in the JA 37. Not just the central computer, which we kept in principle, but also for example the autopilot, which became the first operational digital autopilot in a combat aircraft. It was the result between a partnership between Saab and Honeywell. The basic system was a mechanical-hydraulic one, inherited from the AJ 37, and on top of that an autopilot that adds or modifies input signals in various ways. Then we also see that the atmospheric measuring system was also computerized, which was a big improvement because especially when you run back and forth through the sound barrier you get a lot of issues. We were sick of old mechanical cam discs and other miserable things in the old measuring devices, we could do better functionality in software. Then we also have some other sensors and some old purely mechanical instruments in the cockpit.

Additionally, a computerized or digital inertial navigation system was introduced in the JA 37, a good deed that was largely dependent on the people at the avionics bureau at FMV being proactive. Down at Saab the thinking went along the lines of let's save money, we did fine in the 35's with the old platforms. But this was a great improvement in quality. We had managed to create a "pseudo-inertial" system in the AJ 37 purely in software, via filters, complementary filters and such things. And that was very well suited to porting over to the better quality when we got a real platform for the inertial navigation. It turned out to be very important, both for the position reporting during fighter missions and for the quality of detail, the dynamic quality of all values shown to the pilot. The HUD in particular was especially sensitive since the symbols - like the artificial horizon, for example - had to be stable in relation to the outer world.

Then we have the electronic presentation, it was called EP 12 if memory serves. It had some digital solutions in a presentation generator and a "Digital Interface Unit" that talked to the radar (the DIU was made by Hughes in the US according to a Swedish specification and named such because of export restrictions; it was also called a "scan converter" and transformed the polar coordinates from the radar data into rectangular ones that were used by the cockpit radar display). The radar also introduced some computers, which you might hear more about soon. And all of this meant a lot, both signals processing and general control of various functionality via software. The HUD we inherited from AJ 37 and improved. There were some issues with the field of view, but some of the engineers mucked around with it and got double reflex glass which improved the FoV and made the thing more manageable. Then, of course, we have the radar display. The guys in Mölndal and Kista (SRA and Ericsson, respectively) hacked up some new technology with some kind of digital processing of the radar data, which made it very simple and nice to overlay symbols on the radar display in the same way that you displayed radar returns and tracked targets.

The radar display is in the center of the cockpit front, behind the stick. If you look to its right and slightly upwards, there's a new display there that wasn't there in the AJ 37. That's the "tactical indicator" (TI) or tactical display. There was a lot of gargling about that. The avionics bureau at FMV had agreed with SRA that it was entirely possible to add another display with the same kind of info as the target display ("målindikator", MI; the head-down radar display). But they didn't really know what to use it for. They started by drawing something like a course instrument for showing the bearing to the target. The air force test center's test pilots were really skeptical and wondered what the hell they were gonna use that for, and went "give us an ordinary mechanical target bearing instrument"!

But we had a guy at Saab that had previously been an air force air combat controller. And he was very excited when we started talking about drawing maps on the tactical display and was like "it should be possible to use this for mission parameters in a better way and show some targets and stuff". He was all over that idea and wanted to combine the target display with some kind of simple map. Since he was an air combat controller he was able to convince the controller people as well. We started working on some suggestions. SRA started thinking about electronic map generation and came up with what would later become an ELKA (an abbreviation for "electronic map"). However, when we got to the time to make the decisions, we got a document from the test pilots that said that they by Jove didn't want no tactical display, it should be thrown out of the system. They wanted a bigger mechanical target bearing instrument. But down at Saab we all agreed that the tactical display would be a big step forward and FMV agreed with us.

In the end it was decided that the tactical display was in, and that turned out to be one of the big improvements in how data was presented to the pilot. Among other things, this was because until then, we had used what was called "direct combat control" on the 35's, that is to say all commands and courses and altitudes were decided and calculated on the ground and transferred to the aircraft via data link. But this new technology opened up a world of possibilities to do what we called "indirect combat control", in part due to a good inertial navigation system and in part due to this tactical display. But the air force chief of staff, I think it was the choleric man Gösta Tullson, he forbade us to do anything related to indirect combat control at all. We ignored him, though.

Together with the air combat control people we programmed the computer with all of the different data link messages and did all of the ground work for this, so we could transfer everything the ground installation knew of the target's position to the aircraft. And up there we had the complete freedom to calculate intercept vectors and target bearings and such things ourselves. This eventually turned out to be very important functionality in the JA 37 and all of the hostilities were forgotten. But to throw the test pilots a bone we had to extend the panel above the tactical display, as you can see in the picture of the cockpit. We had to make a bump on the right and take some space from what had previously been holy, namely the field of view over the nose during landings and low altitude flying, in order to put in a mechanical target bearing instrument that resembled the one in aircraft 35. Because the 35 was holy to our fighter pilots back them. But that bearing instrument really wasn't necessary. The tactical display had the same information plus a considerably better overview of the situation, a map, targets, friendly aircraft and so on. I've met fighter pilots who wondered about this mechanical instrument, because they had hardly looked at it. For symmetrical reasons we had to raise the panel on the left as well. Suddenly the field of view over the nose wasn't so important anymore!

Well, that was a long parenthesis about the tactical display. There's one more thing to say about this, though. The communication in and out of the central computer, between all of the computerized subsystems, it was done point-to-point using serial binary links. And that was a huge leap forward, because they were programmable in both ends. We could grow flexibly as our knowledge grew, we didn't need to change the hardware. What remained of analog things usually entered the computer world via a conversion unit that turned the analog data digital and then let it into the central computer. The communication with the combat controllers on the ground was of course digital. Digital tape recorders were also an important thing that SRA developed while working on various presentation screens, and they came to be used a lot for training. You could record everything that was displayed to the pilot and show it to them on the ground after the mission, and they learned a lot from that. This was used more and more as the JA 37 was developed. Other tape recorders were used for maintenance and testing.

[...]

Moderator: Bengt, I have a question. Early on you hinted that it was hard to get traction for this idea with a central computer. Was it some big isolated thing that led to you guys getting approval for it, or was it more like a gradual acceptance of the idea? And who were the key people involved in this?

Bengt Sjöberg: Well, we had some enthusiasts down at Saab. Then we had a lot of support from FMV, the avionics bureau and also the weapons bureau. Despite those two being kind of at odds with each other, or maybe more like competing with each other in some respects. There were some people up there that tried to be at the cutting edge of technology and cut in at certain critical points. The big skeptics were down at Saab. The closer to the aircraft building and sheet metal bending you got, the more contrary they were. They saw all avionics as a complication. They were used to Saab just making aircraft and that FMV ordered empty spaces that they put their secret black boxes in. And who was responsible for the entire system, that's beyond my ability to judge.

Ulf Frieberg (test pilot): Well, there's been some criticism flying through the air here. If we consider what kind of people could help with with development, I saw with more and more clarity over the years that there were serious problems with the test pilot training, with the different specializations available. There were those who knew the aircraft and all sorts of ergonomic things very well. And there were those who were starting to tentatively look at avionics systems. But not even at Saab they weren't listened to very well. And that's something I'm definitely remembering for future generations. But then I had the great privilege to join those who had been involved since the start of the AJ project. And when we got started on JA people were suddenly all ears from the beginning. One big obstacle was actually the Draken, which was a very good aircraft for its time. But in terms of development it was a surprisingly bad aircraft. If the pilot got into a combat situation where decisions made in the matter of a second decided if he was going to live or not, he was presented with choices that he would need time to consider. Some of our people saw that, some didn't. And that was the big dilemma in JA 37. Those of us who had been involved in the AJ 37 development were considerably better prepared for discussing these things.

But when we saw Bengt Sjöberg's vision about systems and improving pilot efficiency that was met with a great deal of enthusiasm from both sides. We saw the possibility of improving pilot effectiveness a great deal, you could not just do things but do them during very high levels of stress. And, like I said, there were all to few that had the possibility of developing themselves through the insight of a superior, instead there were some individuals that educated themselves, and of course this resulted in conflicts. But these things were irrelevant considering what actually happened. Bengt came along with his enthusiasm and I needed to support him. Unfortunately, at the air force test center there was a faction of very talented guys who knew the J 35 very well and really wanted to use it at its full potential, which was far above the level most regular pilots could get to. That meant it was one hell of a job to get everyone on board. I guess used some tricks, but that's probably forgiven today.

As an example we can talk about this little mechanical target bearing instrument above the tactical display, which Bengt mentioned. It was absolutely nothing until we suddenly realized that SRA had improved that map above a certain level. And suddenly everyone agreed that the tactical display should be in. We had an experimental model installed in the simulator at an early stage and the entire mission overview was ready in a second. I mean everything was clearly laid out immediately! But if you wanted to get the same overview through conventional instruments you had to look in a lot of different places and gather the information up yourself. That was a big step in convincing the test center and this faction. Without this development step you'd have to be continually keeping up with new directives and commands from the ground. But with this tactical display, that wasn't necessary. The pilot was far, far ahead of the data link directives and he was aeons ahead of the air combat controller on the ground and could see the entire situation continually and correctly. I've never heard any pilot saying he actually used this relatively expensive mechanical bearing instrument.

Kim Bengtsson (computer engineer at FMV at the time): Well, I can relate a ridiculous anecdote. Bengt was talking about integrating the inertial navigation unit. Originally it was completely analog, that unit, and had the software running in the central computer. It was FMV that bought the inertial navigation unit and paid for software development for it in the US. It worked very well. Then Kearfott came up with an improved unit that had a built-in computer, a pretty powerful machine, so we could move all the inertial navigation calculations to that computer. In theory, that is. But then we ran into the problem of how to replace the unit in the test aircraft, which still had the inertial navigation program running in the central computer. In order to make that transition smoother we let the inertial navigation unit's computer simulate the old analog inertial navigation unit, so we kept the same interface but without actually making meaningful use of the computer in the new digital unit. As far as I know it still worked that way when we scrapped the JA 37's. We sometimes talked about how possible future researches would wonder what the hell we were doing.

I also have something to add regarding what Bengt said about direct vs indirect combat control. There were a lot of big discussions about that. From the start we were more or less forced to make JA 37 simulate and act like a J 35. And that was really stupid, a lot of people were pissed off about it. But when thinking about it in retrospect maybe it wasn't so stupid after all, because we had one leg firmly on the ground in the connection to the ground systems, and we adapted the aircraft to that. And after that we could make the transition to a more complete combat control system in the aircraft. At that point we had gotten to know each other, understood each others problems and so on. So maybe it was a bad thing that turned out to be good in the end.

Gunnar Lindqvist: The Draken had a complicated targeting system. It wasn't just a radar, but also an infrared sensor that unfortunately wasn't too reliable. And this made it pretty hard to switch between different working modes. You took information from one sensor and combined it with information from another and there were a lot of different cases that made it pretty hard on the pilot. But under not too stressful circumstances I maintain that it wasn't such a bad system. In particular it wasn't bad compared to what everyone else had at the time. You shouldn't compare the 35, which dates back to the late 50's, to aircraft introduced a decade later.

When it comes to the central computer, its origins are further back. There were a lot of discussions regarding if aircraft should have one or two engines and if they should have one or two crew. And most of our strike aircraft have been two-seaters, but some that have been modified from fighters have been single seaters. That was a big issue during the development of the AJ 37, if it was going to be a single pilot in it, or a pilot and a systems operator. But finally the faction that wanted a central computer instead of the operator won. That was a risk, but it turned out to work well. But I want to remind you that we have aircraft called JAS 39 B and D, and those are two-seaters. Sometimes you need two crew.

[...]

Leif Åström (strike aircraft squadron commander at the time): I'm going to back up a bit and talk about what Gunnar mentioned, taking the step from flying two-crew aircraft to single seaters and the worries we had about that. At that time I was in a strike squadron and flying the Lansen and experienced this change personally. In the Lansen we were dependent on having a navigator/systems operator in the rear seat to be complete as a leading aircraft. It wasn't certain that all aircraft had one, only some did. We were also dependent on having a radar, and not all aircraft had one of those either, so only some aircraft could lead. If you were flying with the squadron commander he was usually in charge of one group while his second-in-command led the other. Really, those two were the only ones who were fully equipped. If you lost one of them, or even worse both, we were very ill prepared to continue the mission. In practice the unit lost its decision making powers. Happened a lot. It was the operational consequence.

From a pilot perspective the obvious thing was that we had a system that was based on top-down control. The commander led, the rest of us hanged on to his wingtip and followed. And that was an organization that motivated the flying personnel about as well as you'd think, of course. When we switched to aircraft 37, a single seater, where every aircraft was fully equipped with radar, navigation systems and so on, we suddenly had the capability that anyone in the unit could take over command at any point if necessary. It could be because we lost someone or maybe the guy taking over had the best view of the situation, or whatever reason. And that, I'd say, was a very important turning point for us. We now had an equipment system that let us switch from a top-down organization to a management-by-objectives one. We could use the squadron's intellectual capacity in a completely different way. So this was a very important step in many ways, not just important technically but also important operationally.

If we then look at the JA 37 and the tactical display it was basically the same thing. Before, it had been the air combat controller on the ground that made the decisions and controlled his fighter aircraft with data link directives. With the introduction of the tactical display, the air combat controller's behavior changed. He became the one that served the pilot with information so that the pilot had a better view of the situation, a better basis on which to make decisions. And then air combat controller wasn't the bottleneck he had once been. Not to say anything negative about the air combat controllers, but if you have a demanding decision making role, you can only manage a certain number of fighters at a time and the air combat control was often a limiting factor. When we started serving up information on the tactical display that limitation mostly disappeared, and at the same time we started getting pilots who were used to making their own decisions. The pilot's task was to make decisions based on what he saw on the tactical display, and we created more decisive pilots, pilots that were better prepared to handle the unexpected and better at reacting to quick changes. So I'd say both of those changes were fortunate.

Lennart Alfredsson (then at SRA): You're talking so much about that tactical display, but I guarantee that back then when you were going on about it, there wasn't a chance in hell to actually make it work. We had no working solution. Map instruments in aircraft did exist, optical ones. Ferranti had implemented one, it was based on a 35mm film running around like crazy, rotating and whatnot. Then there were those who built small extra ports in the back of cathode ray tubes and projected the map on the back of the phosphorus layer, it was white so you could see it, but it had to be pretty thin to be visible. Since we didn't have an aluminum backing we'd lose half the luminance if we did it that way. So we didn't really have any good way to do it. On top of that our project lead didn't want the thing to be too expensive.

Well, we decided we were going to try with a digital tape deck, and the ones we had could hold about 14 megabits unformatted. But it took five or six minutes to run the tape from one end to the other and that made things difficult. Sweden's a very oblong country so we had to split the tape in segments. Possibly this "Raymond Tape Unit" we had could work, it was in use in aircraft but not for this purpose. So we really didn't have any solution that would let us draw a decent electronic map. We worked for over a year with trying to find new solutions that we eventually discarded, further new solutions that we also discarded and so on. For about a year. I had a colleague called Stellan Nennerfelt, many of you knew him. One Friday we went home and suddenly we both had an idea. And we both rushed back into the office and said "I have the answer!". We had the same solution, the two of us. Exactly the same.

Simple, but not something you come up with immediately. The time for a presentation on the tactical display is about 20 milliseconds, and it refreshes all the time, so you have to do everything in 20 milliseconds. We callously stole two milliseconds and said that two milliseconds, that's what we got for the map. But no longer. We wanted a lot of other things on the display as well, after all. And then we made one long chain of segments, starting in Sundsvall. Ordinary seven-segments, sometimes you could jump a bit further and sometimes a bit shorter. Sometimes it was silent and nothing was drawn. We started in Sundsvall and just followed the chain as fast as humanly possible, along the coast, up rivers and down rivers and up to Umeå. Along the entire Finnish and Norwegian border and the entire way down the western side of the country, some islands in and out and up and down. And when we got to the place where the aircraft was, we stopped and went segment by segment and illuminated the entire map within the area the display showed. And when we got to the far end we sped up again.

This actually worked, this was something you could do in a small country. Hughes Aircraft Company, our partner, they couldn't do it in the US, it was too big. And when we tried with Australia when we were going to travel down there, that didn't work either. Sweden was just small enough. Those of you who were involved back then will remember that there was always a demo map over Scania (southernmost part of Sweden). That's the perfect place in Sweden to draw an electronic map of. Try it up north and see how that works out for you! There's one single building! And that building was put there on the map just to make it possible to figure out where the hell you were.

But the electronic map turned out to be cheap, actually. The entire system was 4000 four-bit words, 16000 bits total, and that covered the entire country. You can imagine how much that is. 16000 bits is about two kilobytes. Saab's travel expenses form is 65 kilobytes today, and that's before you fill it in. It wasn't easy to get this electronic map in, really, it took us over a year to find a workable solution.

Ulf Frieberg: But there's no doubt that once that step was made, all resistance was gone.

What the tactical display actually looked like. The aircraft is the small triangle in lower center; it's heading south by southwest over the island of Öland. In the lower right corner the aircraft weight (VIKT) and max allowed angle of attack (ALFA) are shown.

Lennart Alfredsson: We had simulated the electronic map earlier in a simpler way in aircraft 35, or maybe it was in 32 Gamma (32 Gamma was a modified Lansen that was used as a flying testbed for the JA 37 radar and some other avionics). We had the resources, we could actually fly these experiments. In combat aircraft, too, unlike the yankees who got to do it in transport aircraft early on.

Hasse Olofsson (at the avionics bureau at FMV): I worked on the presentation system and related developments directly. What I think was a major factor in all of these parts was actually our experimental and testing activities in parallel with the regular development. Without those we would never have dared to push as hard as we did. You mentioned 32 Gamma and that's where most of the work was done. It was the same with the HUD in JAS 39 which was the first of its kind in the world. We've had an incredible experimental and testing environment.

Bengt Sjöberg: I thought since we were discussing the usefulness of the tactical display I should jump back a bit and point out a bit of icing on the cake, if you will. More specifically the fighter-to-fighter datalink that more or less came skidding in on a banana peel at some point. They were working on a project called A 20 (proposed new low-cost strike Viggen version with the capability of carrying more and heavier weapons) and couldn't afford to put a radar in every aircraft, some development work was done more or less by feeling and then the entire thing was cancelled. But one thing they did come up with, namely that if only some aircraft had a radar then you could make those transfer their radar data over a radio link to their squadron mates. They squeezed a solution out of the radio system that made it possible to send that data during idle periods. The aircraft with a targeting solution could transfer both the radar image and the targeting solution to others. From this the real fighter-to-fighter data link grew and was later integrated on the JA 37. We ended up with outstanding cooperation capabilities within a fourship, through this datalink and the tactical display. And that functionality JAS 39 ended up inheriting as well.

Leif Åström: I have a few things to add to that, it might be worth commenting on from the user's perspective. The first step, the introduction of the pulse-Doppler radar, that marked the end of the days when the men of honor and silence rushed along between the treetops of pine and fir and nobody could see us with radar. We could hide pretty successfully from the 35's back then, but with the Doppler radar that came to an end. They dug us up and saw us down there too. On top of that, when the fighter datalink came, when I got a radar warning from one direction, suddenly another one appeared from a different direction and had me boxed in without me having discovered him at all. That was a very unpleasant development for a strike pilot! It was a very big step forward.

Initially, this was something we were pretty careful with. There was a lot of secrecy around the fighter datalink, it was nothing you could talk about carelessly, it was some years before we showed it to anyone. And what I was actually going to talk about was the first time we showed it to an international audience. It was at the 6th air wing and we had a visiting squadron of Jaguars from the UK. It was kind of a special visit, we met up on a Monday and gathered up as usual in the briefing room and presented ourselves to each other. We talked about what we were doing, about our aircraft, such things. Their squadron commander did the same thing for their side. They had their wartime assignments in Norway, as it turned out. That was interesting. But then he said something that was pretty surprising and very un-British. He said "well, I really gotta ask you one thing I've been wondering about. Why on earth are you, in this small country with a population about London's size, developing your own combat aircraft for silly amounts of money when you could just buy off the shelf, from us for example?" And that was pretty hard to answer then and there.

But then they got to follow us for a week. They got to ride in the 37, see its performance for themselves, performance that wasn't just marketing but actually true in real life. And they got to see some other things. At that time, we showed - for the first time, as far as I know - the fighter-to-fighter datalink. They got to see it at the 13th air wing in Norrköping through the UTB (the UTB was a recording and presentation system mainly used for training; it recorded almost everything that the aircraft showed to the pilot, as well as pilot input and some tactical data, and the entire mission could then could be presented - including pausing and rewinding - and analyzed on the ground in what basically amounted to a simulated cockpit) and we noticed how they grew silent. They started whispering among each other, talking and pointing, and it was obvious that this was something that had impressed them. And the week went on and the time came for them to go home. And the same squadron commander stood up again, and he was a bit embarrassed, because remembered what he had said that previous Monday. He hemmed and hawed for a little bit but then he finally got to the point and said "I remember what I said last Monday and well, now we've seen some things here".

Then he talked about the experiences I mentioned and he said "You should know that right now we're discussing if we can afford a system like this in our next aircraft, the Eurofighter, and in that case how many of the aircraft we could afford to put it in. And then we get here and see that you already have this system operational, and at a fraction of the cost we're going to spend on it. Keep building aircraft, it seems like a good idea!" He actually said it, just like that. I think that might be some kind of indication of how valuable that system was, because sometimes it's hard to see internally how far you've gotten. This was in 1985, I think, and only today are other systems like this starting to become operational. Sometimes we've been far ahead without knowing it.

Kim Bengtsson: Yeah, I just want to relate another anecdote of how the datalink was perceived. I remember the first time the test center pilots were up and flying with it and when they landed they were like, "This is goddamned awesome, now we can turn those guys down in the cave off".

Moderator: Who were the guys down in the cave?

Kim Bengtsson: The air combat controllers. But maybe I shouldn't call them that here.

Bengt Sjöberg: Well, "turning the guys down in the cave off" might be a bit of an exaggeration.

Kim Bengtsson: Well, yes, slightly, but that was the general feeling you got.

Bengt Sjöberg: These days though you want encryption and ECCM on everything, so it's not so easy to detect and identify what's flying around up there. You can out yourself by sending out a bunch of signals.

[...]

Moderator: Well, we've talked about the system, but I was thinking we should focus a bit. We've already talked a bit about user interfaces, but maybe there are more things about the machine end of things that we'll want to cover?

The target display in the JA 37, in its original incarnation.

Lennart Alfredsson: Yes, well, you can't avoid talking about user interfaces when you've been at SRA working with displays for a significant part of your life. In the very first solutions for the JA we used parts from the AJ, of course. In the AJ you could turn the reflex glass for the HUD up and down, but Ulf Frieberg didn't like that, he wanted a different solution. And we did end up with another solution in the end.

In the AJ the central display in the cockpit was a round thing with an outer diameter of 110 mm (4.3 in) and on that an ordinary PPI was shown. The useable area was about 100 mm in diameter. We couldn't go back to the system we'd used on the 35, because there the radar returns were "stored" with simple electroluminescence that showed a "memory" of the pulses, and the visibility wasn't great. We wanted something with much better visibility than that, and so we started off with the AJ's radar display.

The first thing we tried was drawing a regular B-scope. An analog B-scope has pretty big distortion errors at the bottom of the display, things float out towards the edges no matter what you do. We implemented some B-scopes and showed them on the display we had then; as Hasse Olofsson mentioned we had a pretty good amount of money to use for various related development. But the result wasn't very good, nobody at all thought it was a sound idea. Instead we returned to the solution they had in the US, which was based on a "scan converter", which meant you took the radar input signal, put it through a special cathode ray tube and then you could read a signal suited for presentation in the other end and show that on a display. You needed an extra glass tube, about 40 centimeters (15.75 in) long with a weird double anode in the center and two electron guns, and this was supposed to fly.

I know the yankees did that. We had steady contacts with Hughes Aircraft Company. Their solution was originally part of a system called CORD (Coherent On Reception Detection, a precursor to fully coherent systems), which some of you might remember. And that was no good. It was nothing. Nobody at FMV, especially not at the avionics bureau, found that particularly attractive. But then, in the morning of May 8, 1968, my colleague Acke Axelsson called me and said "I'm back from my visit to Hughes and I have something to tell you". And he came to me after lunch. He told me there was a new component, a linear shift register, which was a thousand bits long.

A single component that could store a thousand bits, that was something that hadn't existed before. We got started, Acke and I, and we kept on until nine PM and we drew and drew and drew. We built chains of these shift registers. We needed about a hundred thousand bits (~12.2 kB, about two thirds the size of the text of this post) to make a B-scope. And we really wanted the possibility of storing and "aging" the radar returns. On top of that we had the problem that while the radar had a refresh rate of about 1 Hz, the presentation system ran at a refresh rate of about 50 Hz. Thus we had to be able to present the information continuously while reading in new radar data and removing old information from the system. It actually took us nine hours until we were convinced that it would work. From that day, when Acke and I went home at about 9 PM, the old solution for the central display and everything behind it was completely dead. There wasn't a chance we'd even get the idea of getting that back.

We had solved all of the problems. In the central display on the AJ there's radar integration pulse by pulse, there's brightness control, there's memory, there's tons of things. Everything was a huge compromise, an extremely fine balance of voltage potentials that controlled various grids and cathodes in the system. It was all gone, we could divide and conquer. The radar did its own thing, it delivered a neatly packaged video signal. We could draw symbols separately, we could solve the storage of radar pulses separately, we could solve the brightness control separately. Everything came together in a single evening. And there was no talk of going back to any other solutions.

But it was hellishly expensive. When we got started the military variant of these shift registers were eleven bucks per bit. Eleven bucks per bit (in today's dollars; I adjusted the original Swedish numbers for inflation and converted to USD). And we needed quite a number of them. We were talking millions, and there was no way we could spend that on trials, so when we built an experimental system we used the civilian ones and they were about $2 per bit. In the end when we had to quote the cost for the system the military variant had gone down to $2.2 per bit as well, so that was 220,000 dollars for that one component. Everything, the entire system, five boxes of electronics, was $287,000 in all. It wasn't easy to convince the management that we should blow $220,000 on a single component.

Anyway, we built these things. From that point analog wasn't on the table, everything was purely digital. This turned things on their head, we could process targeting data, we could do all sorts of things. And then Hughes had another flash of genius. They suggested we should switch to drawing the targets character by character as ASCII. What we did was we chose only the targets and marked them, which meant the memory footprint was greatly reduced. In the system we had the capability of showing 128 targets. That was all. We implemented "aging" so we could see the target tracks, we could save two, three, four, five or six, as many sweeps after each other as we wanted. So we could see where the targets were going. We could draw their course vectors directly on the display. But that information might not be all that easy to interpret correctly when you're going really fast yourself and on top of that you have to take the B-scope distortion into account. So then we implemented compensation for the aircraft's own movements and for the B-scope distortion. More or less, the tracks you saw on the display were a pretty good representation of how the target was actually moving in relation to the aircraft's "track-up" in the system.

May 8th, 1968, that was the big turning point for us at SRA. The symbols weren't a huge problems, we had those in the computers, we did a lot of experiments with those, we tested all kinds of things. It was obvious that computers and digital was the solution. But for the radar it was on that day that we received the information that solved everything. Sometimes you're in luck.

[renhanxue]

Continued:

Göran Tode (visitor at the seminar, worked with strike aircraft ops at the air force HQ at some point): I have a question, let's see if I can formulate it decently. I'm rather impressed when I hear you talk about these things, how you took advantage of new ideas. I wonder if there's something special with corporate culture when it comes to aircraft in particular. In that you take advantage of all opportunities, and that it's so dynamic, it changes all the time. Many years ago I got take a ride with the navy while they were hunting subs, and I took a look at their command and control system and I understood immediately why they never caught anything. It had brass tubes and whatnot. No, I mean, through the aircraft, through high-tech and the danger and everything, you immediately get to know if things are good or bad in a different way than you do at sea. And if we look at administrative control systems that's even worse, then you never see anything until it's too late. Does anyone want to comment on this?

Lennart Andersson: Well, it's a good thing. In my world there's always been positive and negative things. If we at SRA had some good ideas that we believed in, that we though we could do - and it was certainly the same at Ericsson in Mölndal - FMV was always up for it. Sometimes you had to advance some money but it always came back in the end. We had lots of opportunities, we had it better than the researchers at any university. However, in my opinion, what was kinda wonky was the corporate executives, you couldn't trust those. At all. When we got started with digital technology at SRA, which was about 1964 to an order of magnitude, the circuit boards were huge clunky things. And then a directive from the execs came down, saying that digital circuits were for the development division only, and those guys were on the sixth floor somewhere far away. The rest of us were not to work with integrated circuits. They forbade us to work with these things, but we didn't give a fuck, we didn't take that seriously.

The next big thing was when Saab got started with computers, or calculators as they were called back then. There was a directive from Marcus Wallenberg himself to all of Ericsson and SRA (Wallenberg was one of the big owners of both Ericsson, Saab and SRA; he was the chairman of the board at Saab at the time) that we could not do anything with computers. Period. And of course we didn't, we called them program units, and in Mölndal they called them control units. But they were surprisingly computer-like, I'd say! Anyway, those were the orders, we were not to work with calculators. When new things show up the execs have to let them out all over the business and let everyone use them, no restrictions, check them out and poke at them. There's always someone who has an idea somewhere.

Moderator: Can I add something to that? You'd think that the universities would be on the cutting edge, and I became a professor in 1965. If we were to purchase a computer at the university, it had to be approved by the computer committee, the computer central. And they were kinda like corporate execs, they said we couldn't get a computer of our own. But then we called it a process control unit instead. And then it was fine to buy one. So it's not just the corporations that has management like that, the kind of progressive places called universities suffer from that too. Sorry, I shouldn't talk so much myself. Gunnar.

Gunnar Lindqvist: Well, regarding the skepticism against electronics. I suppose I should point out the background information that in the case of aircraft 32 and 35, the electronic systems were heavily delayed and couldn't keep up with the aircraft deliveries. That's why the aircraft engineers a bit doubtful about newfangled things. At Saab they had a new systems division and the guys who "banged at the sheet metal", as someone irreverently put it, wondered what these systems people were doing. When it came to delivery deadlines the avionics industry wasn't as mature as the sheet metal benders. When we were getting started with aircraft 37 we got a directive from the commander of the air force, that we couldn't put a single new device in the 37, it was all to be stuff from the 35. This was changed later but the IFF unit remained. One of my coworkers said it was only interesting to museums and souvenir hunters. That was some background regarding the skepticism against electronics. As always, it was about risk management.

I almost forgot the most important thing I was going to say, the issue of the customer's knowledge compared to that of the provider. We at FMV had the mindset that we had to have very good engineers with a vision of the future, who could communicate with the people in the industry, our suppliers. The Royal Academy of Engineering Sciences conducted an investigation in the 70's where they found that one of the reasons that Swedish industry had been so successful was the relationship between customer and supplier. And that wasn't just between FMV and Saab and Ericsson, but it was at least as important between, for example, Vattenfall and ASEA Atom (nationalised power company and nuclear power generation respectively), and between SJ and ASEA Traction (national railways and train builder respectively). And certainly not the least, Televerket (the national telephone administration). The national telephone administrations in Norway, Sweden and Finland came up with a standard for cell phones. That meant that the Nordic countries were more than a decade ahead of other countries in that area. But in today's investigations made in the armed forces they find that the engineering people at FMV are an unnecessary double competence that just bumps up the requirements and that those engineers shouldn't be there.

Bengt Sjöberg: When it comes to working methods maybe I should direct some attention to the working groups that more or less grew up during the time of the AJ 37. For example, when it came to to adapting the JA 37 to the needs of the pilots, we made sure that was a more formal working group. It was called the PM group and it consisted of representatives from Saab, FMV, the air force test pilots and Saab's test pilots, as well as some people from Mölndal - the radar was important - and some presentation guys from SRA, usually one or two from each. And we had some pretty intense meetings. First we had some smaller groups meeting up to discuss every single system function, and then we had more formal meetings at regular intervals. Other than the PM group there was also, for example, a radar integration group, called Målinmätningsgrupp 70, MIG 70.

Lennart Alfredsson: We weren't allowed to call ourselves 37. MIG 37.

Bengt Sjöberg: No, and this was in the early 70's. I think we met every two weeks for several years. We alternated the place where we met, and here we benefited from Sweden's limited size, we could always call for a meeting just a day or two in advance if we had to. We could gather the knowledge that existed within our borders. Then we had a communications and combat control group where the corresponding people at FMV were members. The pilots had their say, we always had excellent cooperation with the test pilots. Some of them were very good a understanding us engineers and could describe details they observed during the flight, as they were observing it. In a single-seater aircraft it's not easy to hitch a ride and look over the pilot's shoulder. Then we got good simulators and there we could have intense discussions.

When this PM group came with a recommendation that was sometimes a pain for the project management, because who the hell was responsible for what in this group? Who was in charge of the economic issues, and so on? But we always replied that we were just issuing recommendations. Then it was up to every project manager to count his own purse and for the project management in concert to decide. And it was very hard for the project management to go against these recommendations, because there was a lot of power behind them if you look at who was represented in these groups. The way the contracts were written for the 37 was also pretty good for cooperation in an area of technology where the understanding of things was constantly growing. Because the man from FMV could go home and talk to Gunnar here, and come back with a small bag of money, most knots were untangled. FMV had some resources in reserve.

Since we could talk freely between the industries it worked very well. If we didn't like something you could always call up Acke Axelsson or some representative at FMV and say "We're having some trouble with this detail, can't you go talk to the guys in Mölndal about it?". And after a while the problem was usually solved. Usually the FMV man in question had a small bag of money with him because he saw the long-term economy in getting a better system, while the industry was working with a more limited perspective. But, well, you didn't win any popularity contests with your bosses at Saab because of all the suggestions we came up with. Because it cost money. But there was a solution for that in the 37 era.

Those solutions weren't there when we were developing the JAS 39, there we couldn't criticize each other in front of FMV. I couldn't say that now things are fucked at Ericsson in Kista, they're completely stuck. We shut up about that. We didn't tell our friends at the avionics bureau. I really do believe though that it was basically a good way to write a contract for the 37, and these working groups worked very well. Many solutions and ideas grew from there. The PM group, for example, just decided many hundreds of details that we flat out sat down and signed off. No project manager could even be bothered to get up to speed and understand the details, so it became as we had decided if the costs didn't skyrocket too much.

Lennart Alfredsson: There was one downside to these contracts, and that was that it was very easy to change, add or remove things during development. And nobody thought about the upcoming series production. We added a small casing here, we laid a cable there and there was some of this and some of that. The electronic map and the other things weren't free. Things got added here and there. And then when we had to come up with a quote for the series production it turned out to be pretty expensive compared to what we had said originally, we had gotten a gradual increase of costs in this system, at least at SRA. They wanted to get rid of that in the coming discussions about JAS, and so they did and unfortunately they did it way too well. There they connected the development to a limited series production run and there were no changes allowed for something like seven or eight years. That was sort of unfortunate. But there are downsides to any contract.

Moderator: I'd like to return to the discussion regarding the user interface system. Are there some parts we've forgotten or should discuss further? The systems are interconnected. I'll start with Leif this time.

Leif Åström: Okay. We've been talking a lot about the background, how the different devices worked together. And the sum of all this for the pilot, that could be that you saw a target on your tactical display, you caught it with your own radar. With a few simple touches, a few button presses, you could choose the desired target, choose your weapon, a radar or IR guided missile. Unlock the seeker, fire your weapon. It was breathtakingly easy to work the system. In a lot of other (foreign) systems that I've had the opportunity to compare with, you have to do a lot of configuration, you have to set the appropriate radar mode and weapon mode and so on. You suddenly realize that we had gotten pretty far with this, even on this level.

Then, if we start looking at the short-range modes, where you have shortcut functions immediately accessible on the throttle, you don't even have to let go of the stick or throttle, you can just keep flying the aircraft and set both radar and weapon modes right there. That's the explanation, I'd say, as to why the JA 37 is so competitive. The user-friendliness. It's not just that you have very good technology, it's that you make it possible for a human to use that technology in a simple way. And that it all works in extremely stressful situations such close-range air-to-air combat. The fact that JA 37 was relevant all the way from the 80's until it was retired in the 2000's, that's largely thanks to its user-friendliness and the good balance between different systems.

Oh, and I'd like to return to something we talked about earlier. You asked earlier if I could give an example of foreign reactions the aircraft, and I recalled one. We had an American general visiting, I think he was the equivalent of our Air Force chief of staff, or something on that level. And he said to someone next to him, I think addressed to his aide: "Take a good look - that's the best fighter in Europe you see there". He actually used just those words, and that was an American general who said something like that. Of course, I don't know if they're always up to speed on everything, but that was the kind of respect we had on the other side of the Atlantic.

Moderator: You mentioned that it was very easy to work the cockpit systems. Was there no other foreign aircraft that was as easy to fly?

Ulf Frieberg: I'll field this one, since I was there from very early in the user interface development. My view is that most others were far, far behind us. They hadn't really understood that it was the pilot who should decide what kind of workload was manageable, and then make sure that the test pilots were successful in getting that message across to the system engineers. That was the big thing, and the test pilots were not trusted with that kind of responsibility for many years, elsewhere in the world. But that has changed and today they are.

Moderator: So if I put this in very oversimplified terms, you could say that the JA 37 was the aircraft equivalent to an iPod, if you're looking at the user interface?

Leif Åström: Yes, I think you could say that. And I, who got in a few years later and have seen the end result and been able to compare it to how it's done in the MiG-29, in the F-16, in the F/A-18 and others... Without mentioning any specific aircraft, a frequently reoccurring thought has been "how do you manage to make something so simple this freaking complicated"? That way, I've come to the conclusion that we were further ahead than I could ever have imagined back then.

Moderator: Bengt's waving over there.

Bengt Sjöberg: I have a few short words to add regarding one more diagram. I don't need to talk a lot about it, it's just that the pilot was in the center of it all, as the headline says. You see the boxes with the different system components, where every component has an integrated presentation interface that connects to the pilot, who needs to work with all of this. You can also note the communications link to the command and control on the ground. You had to cooperate with the air combat controller during fighter missions, and we worked a lot on that cooperation. When we got started on the JA 37 project, we got a bunch of long lists from the old dogs who had flown the 35 all the way up to 35F, and the they said "Here's a long list of problems. Never, ever repeat these mistakes". We worked through these lists systematically and checked every item. In the 35, the pilot workload had been way too high during missions. I remember one of our test pilots who said "Guys, keep in mind that once I put the flight helmet on my IQ goes down by 50%." With that in mind, we were constantly trying to get rid of every detail that could be a distraction. The central programmable logic was very important, since we could divide the system into straight, simple function chains (or at least we hoped they were straight and simple).

We had various simulator stations, both for testing and for development. We had a pure PM (Presentation and Maneuvering) simulator where one of our best known biotechnologists worked a lot on the buttons and levers. The radar control panel and the radar maneuvering in general were his babies for a while and we had to wrangle it a lot in the MIG 70 group. Doors were slammed, and there was a lot of arguing. But in the end a controller grew out of it, you can see it in the pictures of the cockpit. If you look down at the left side panel you can see the throttle lever there. There's also a horizontal hand controller with various buttons and knobs, and underneath that there's a radar panel with different buttons and knobs. And we tried to simplify all of this as much as possible. Then, as Leif Åström mentioned, there was a lot of buttons added to the stick, which we tried to squeeze in while wrangling with the "base aircraft" people who were used to having a lot of other functions there, radio controls and trim controls and whatnot.

We also had several other simulators, other than the PM simulator that is. The main development simulator, which was developed near the end of the AJ period and then became a given during the JA 37 era, was what we called SYSIM, the SYstem SIMulator. The fact that it was necessary we learned hard way from the day when we, in the alpha rig for AJ 37, installed an ordinary chair and some oscilloscopes that represented the instrumentation. And then we put Ulf Frieberg with a control stick on the chair. And we discovered that we had to disconnect all the hardware devices, accelerometers, air data sensors, angle of attack sensor. We couldn't use those, we had to simplify the simulation.

Ulf rejected the entire arrangement as unflyable. And we realized that he was right, it wasn't workable. We started thinking about it and came to the conclusion that we had invest a lot more effort in simulation, we had to work with a lot of different simulation models. The aircraft and all of its dynamics had to work in the simulation, and we started with building a real cockpit. Eventually SRA got involved and provided various "life-like" presentation devices that we could add. And of course we had to emulate a lot of actual aircraft devices, especially the computerized ones.

For example, there was a lot of work done on trying to generate sane input signals for the radar stuff, in order to make the simulated radar work like it did in on a real aircraft. The presentation devices eventually became the same as in the real aircraft, of course. The central computer too, and so on. Then we started to try to run all of this together in SYSIM with all the software and everything, until it worked. And then we involved the pilots. Every button, every knob, every single instrument was subjected to trials in the simulator and in a cockpit mockup. Long trials which were documented in writing and then conclusions and decisions in the meetings with the PM group.

In the end, you could sit in SYSIM and fly complete fighter missions. Over the years the simulators became better and better at mirroring the aircraft's characteristics in an accurate way. The flight controls were pretty good, and they were increasingly perfected into the JAS 39 era. SYSIM was our most important development platform. Eventually the flight trials became more of a verification, a confirmation that SYSIM was sane. Sometimes defects in the simulation were found and then we had to fix the modelling. But as I said, it was one of our most important development stations, and we had a lot of discussions there, since you could stand behind the pilot and literally look over his shoulder and see what he was pointing at.

-- coffee break --

Moderator: And we're back. I figured I'd let a member of the audience ask a question. Bertil Wennerholm.

Bertil Wennerholm: I was at FMV for many years, I was head of the avionics bureau for a few years in the mid-90's which most likely didn't have much of an impact on the JA 37 design. It was more the work of my friends and coworkers that did it. After retiring, though, I've become a historian and that makes you pretty mean. After listening to all this talk about how good the JA 37 was, I'd like to ask if it was good enough? The reason for my question is, of course, that we fielded a lot more aircraft of earlier generations. In the 50's we had a fighter force of about 600 aircraft, peaking in 1957 with 760 fighters. In 1990 when most of our fighters were JA 37's, we were down to 200 fighters. At the same time, during those intervening years the threat level rose significantly. In the 50's and 60's Soviet fighters hardly had the range to reach large parts of the country. But in the 80's that had changed significantly with the MiG-23's and MiG-27's, and thus I'm asking what the strength ratios were like against this fighter threat?

Moderator: Who dares answer that?

Gunnar Lindqvist: It's a difficult question you're asking. Once we had 55 squadrons, and then later as you said 200 aircraft. You were talking about the JA 37, right?

Bertil Wennerholm: Yes, and the J 35F.

Gunnar Lindqvist: Man has a hard time learning from experience. You can't learn from the experiences of others, and a lot of the time not even from your own experiences. And what happened in 1939, that's just fairytales for the people who make the decisions today. Simply put, it's taking a risk. They figure it's not necessary to invest in a good military defense. Now we're supposed to have four squadrons of fighters, that'll be enough. Enough for what, asks the man with some historical background. But that's a greater political risk, they have different priorities.

Göran Tode: This thing with fighters and the JA 37 in general, it sounds great. But no combat aircraft is good if it doesn't have armament. Haven't you been itching to work on the air-to-air munitions? Picking from the shelf is one thing, but did you ever get to influence radar and IR missile development so you could fit them into the system? Or did you just get to choose from what was available on the market?

Gunnar Linqvist: Well, really, you should ask yourself that, as a staff officer. But I think that there might be certain things that maybe could've been done differently when it cames to weapons. We had the best fighter gun in the world on the aircraft, in fact significantly better than the one that's on the JAS 39. The gun had a kind of renaissance, as I understand it. And we had the best radar missile that could be found in the 1970's (the Skyflash), and later we added an even better radar missile (he's talking about the addition of the capability to carry AMRAAM that the JA 37D got in the late 90's), so that was good too. But when it came to heat seekers, maybe we should've chosen another option. But that's where the economy comes in, it's a dilemma. Should you buy something decent off the shelf for a reasonable price, or start your own development?

We had the choice between the Sidewinder 9L from the US or developing a Swedish missile with the development designation Robot 72-12 (in military Swedish, a missile is for historical reasons called a "robot" - which in ordinary Swedish means the same thing as it does in English). The difference was that the latter would have become a good missile if we had spent a ton of money on it. But we thought the development would take too long, so we bet on the Sidewinder instead. The thing was that we had to wrangle with the Americans for nine years before we were finally allowed to buy it, and in the meantime it became obsolete. That's a representative example of a bad decision, I admit that. But other than that, I don't think you can complain about the armament of the JA 37.

Moderator: All right. We had a few things left to talk about regarding man/machine interfaces. Lennart, if you would?

Lennart Alfredsson: We had another bit of functionality in the 37 that was unique in the world for many, many years. And that was the data recording we did in the aircraft. We could record everything, every single variable and every system that was in the aircraft, from the start. And then you could play this back at the base and you could go through the mission and see exactly where you had flown, exactly what the pilot had done, which buttons had been pressed and everything that went on in the aircraft. This was only possible because we had chosen to use a single central computer. SRA tried to sell this idea on export, but we couldn't find a single aircraft that we could even fit the equipment into. Because they all had one weapons management system, one navigation system, one aiming system and the engine of course did its own thing. And all of them, every single one was different. You couldn't fit them all together at a single point without remaking the entire aircraft from scratch.

But we did this, and we had it from the start. The recorder was probably the only device at the active service units that wasn't painted gray, I seem to recall it being blue or possibly orange. Those of you who flew with it, I hope it was useful to you out in the field because I never got any feedback. But it was pretty fun to design it. The reason it was coupled to the user interface equipment was that of the roughly 900 signals that we recorded, everything the pilot did and all variables that changed, around 700 were used in the electronic displays. The majority of everything could be shown to the pilot in some form.

I did a small survey of how many such signals we've used in different aircraft. In the 35F there were seven symbols that could move or change when you were looking at the air-to-air radar display. It's possible that others might have a different opinion, but that's my count, I got seven. In the AJ 37, there were exactly 23 symbols that could change. And in the JA 37 there were 700! That was a huge expansion in pilot supporting functionality, he could do a lot more than in earlier aircraft. And that was all because the system became digital, it was easy to do it because you had access to everything in all the systems. And it continued in the same vein. When we got to the JAS we thought that it might be a good idea to go big from the start - every one of the four displays got a thousand variables. But it wasn't long before Saab came complaining that that wasn't enough for the tactical display and we had to increase it to 2000. The only thing that's still at a thousand is the HUD, and that's just because it's hard to display too much on it, it's the high brightness that sets the limits. Roughly speaking you can only draw and see about 10 diameters of lines (unclear what this means even in Swedish).

Göran Tode: There was a saying in the air force, that before the UTB, before the UTA (UTB and UTA were the recording systems for the Viggen and for the Gripen respectively), more air combat exercises had been won in the break room than anywhere else. Whoever talked his performance up the best won.

Moderator: I want to ask you, Lennart. Wasn't this a result of having this central computer with all of this information readily available? Wasn't that an imporant prerequisite?

Lennart Alfredsson: Yes. We talked to Englishmen, Germans and Americans about this, and none of them could fit this functionality into their aircraft.

Moderator: Gösta Elg, you wanted to get a word in?

Gösta Elg: I was going to comment on what Lennart talked about. Because this is a matter of having an integrated system. There are two other important examples of this, and the first one is data recording in general, not just flight data for training purposes but also recording test results and functionality monitoring, which were both available in central units and helped a lot with troubleshooting. I think that was a big strength. The yankees I talked to at various times had never heard of anything like that either. And they simply couldn't do it, as Lennart points out. So we were ahead there.

The other strength we had was that we all knew each other. I graduated from the Royal Institute of Technology in Stockholm in 1961. Lennart was my classmate. I had to find some nice place to work where I could do both computer technology and control engineering at the same time, because I thought that was fun. And then Lennart said "I think I know a place, we could at least go take a look, right?". And that was the radar bureau at what was then the Air Force Administration. They were very grateful that anyone at all was interested in coming to visit. You walked up to Westergård's office and talked to him for a bit and then you were hired at the Air Force Administration. Just like that. You get to talk to people in an entirely different way when you work in integrated systems and I think that's a practical experience that we should pass on to engineers and tech people. That's it. Thank you.

Moderator: Kim, you wanted to comment?

Kim Bengtsson: Just a few words on the UTB/UTA. It was a very powerful tool when you started flight testing new software editions to certify them and get them out into service. SYSIM (the system simulator) was for program testing and verification, for approval, it was the absolutely necessary device or piece of equipment, which was complemented well by this data recording equipment.

Moderator: OK, who else? Leif?

Leif Åström: Yeah, I can only agree. I'd also like to get back to where Göran was just now - like he said, before the UTB and the UTA, many air duels were won in the break room. The guy who argued the best, talked the loudest and showed the best with his hands what he had done, he won the air combat exercise that had just been fought in reality. But with these recording systems, you could see exactly how it had gone down. And that led to a really commendable improvement in discipline, I'd say. On top of that it was huge for our tactics development. We could actually see exactly what worked and what didn't, and we could reward good behavior and get rid of bad behavior. It should not be underestimated. At the same time it's a bit surprising that we were alone in doing it. Among the pilots, there was a lot of interest for it. After each flight, you went to the UTB as a matter of course, to check out how you had done. It was very exciting to see. But we noticed that many other air forces, they were a bit hesitant in the face of this. They seemed to look at it as the scary big brother watching you. So maybe it's a different attitude here that caused it. But that doesn't diminish its importance, it was immensely valuable.

Lennart Alfredsson: Just want to clarify a bit here. What you do is that you record the nav system's data in the UTB/UTA for each aircraft, so you now how each aircraft has moved in all three dimensions. Then you can present this information for several aircraft at the same time, so you can show their tracks in relation to each other. The alternative approach is to have a lot of big honking radar stations on the ground that track the aircraft positions from there and then add all of that data together. If you're coming from that background you can see that what we did wasn't so bad at all. It was a lot cheaper.

(...)

(Translation note: the seminar starts to get increasingly technical at this point. If you have trouble following, I recommend brushing up on your radar understanding. The Wikipedia page on pulse repetition frequency might be of particular interest, and radartutorial.eu might also be quite helpful.)

Moderator: I would like to segue over to a few other areas. We still have the radar to talk about, and development methods and software. I think I'd like to move on to the radar, if there isn't anyone who has anything important to say about these things we discussed earlier? In that case I'd like to pass the word to Ingvar for a short introduction. The radar was very special because it incorporated a computer, which was very important if I understand it correctly?

Ingvar Sundström: (b. 1937, electronic engineer, responsible for developing the digital signal processing in the PS-46 radar for the JA 37, later head of the development team for the PS-05/A radar for the Gripen) I was going to talk for a little bit about how things were around 1968 and a few years into the 1970's. As I experienced it. I worked at an engineering division at LM Ericsson in Mölndal and I was responsible for servo technology and some signal processing. In production at the time was the last bit of deliveries for the Draken series production. There was both the IRST, which we built on license in Mölndal, and the radar, which we had developed ourselves. And we had just started delivery of the series production of the AJ 37's radar in the early 70's. Both the Draken and the AJ 37 had entirely analog radar systems, pulse radar systems, you only used the information contained in the amplitude of the pulse. And the communication between the radar and the CK 37 (the AJ 37's central computer) was entirely analog too.

There was computer activity close to me, we were working on a computer for the tele-test truck which was used for the AJ 37 (the tele-test truck was a truck containing automated testing equipment for the AJ 37; you hooked it up to the aircraft and it would test all the onboard systems; there's a long article in Swedish about it). LM Ericsson prototyped it and delivered five sets of equipment to FMV for evaluation, and I think that was the first computer stuff we did in Mölndal.

Tele-test truck on the left.

Then it continued with process control, close to us was an in-house-designed computer called UAC 1601 which was used for that, and it later became UAC 1610. They tried to get into a lot of fields - train signal control, hospitals, banks, things like that. And when it came to the radar in the JA 37 there were intensive studies and experiments during most of the 1960's. I think they had gotten started for real around 1967. Is that right, Jörgen? Jörgen Nilsson here in the audience was head of the systems division back then. He was very open to new ideas that came up and the requirements for the JA 37's radar were set very high. I think Gunnar said at some point that if we can't make the detection range requirement we're not gonna have a project. Was that right?

Gunnar Lindqvist: Yes, the radar was a key piece of equipment in the aircraft, which made it the best in Europe for a while.

Ingvar Sundström: And it was obvious that we were going to use pulse-doppler technology in order to be able to fight at all altitudes. When you're looking down towards the ground with a radar you get a ton of ground returns with all kinds of amplitudes and frequencies. That makes it very hard to see flying targets. But if you use doppler, you can separate flying objects from the ground. You talk about something called waveform, how you modulate your outgoing microwave signal in order to get desirable properties. And there were studies conducted over several years on what this waveform should look like. After a lot of work and talking to American experts we figured out a solution that should be possible, the details of which I don't have time to go into now.

The only possible way to do signals processing in a radar like this was to do it digitally. You handle the complex numbers which you get when you start using both the phase and the amplitude in the microwave signal. To begin with you need an extremely "pure" microwave signal for it to be possible at all, and when receiving the signals you must use both the real and the imaginary part - or phase and amplitude, if you want - when processing the signals. And you need to get this into the digital system and then you need an A/D converter and those things were not readily available back then. You either had to build them yourself or pay for someone to develop them for you. It was a pretty big thing. And this complex signals processing, I think we who were working on radar were among the first to do anywhere. These days it's completely ordinary technology that's everywhere in communications equipment: digital television, satellite phones, cell phones and so on. Everything relies on using complex signals processing, but back then it was completely new.

Digital signals processing became possible because of the development of "medium scale integration", like the shift registers Lennart was talking about. They became useful for DSP as well. We didn't need quite as many bits in the radar, but those shift registers were among the first components that became interesting to us. And of course the radar was a tracking one, where you had to filter out good target data. Earlier, this had been accomplished by hooking up gyros to the antenna. There were three such gyros in the Draken system and they built servo systems that kept the antenna pointing in the target direction regardless of how you flew and which way the aircraft was pointing. These were pretty "jumpy" systems and it wasn't all that pleasant to mount sensitive gyros to a moving antenna that banged into things pretty hard sometimes. But you wanted to go as fast as possible.

In the later half of the 1960's it suddenly became really hard to read magazines on automatic control engineering, you started seeing a lot of weird matrixes. In March of 1969 Karl Johan Åström (b. 1937, became a professor of automatic control engineering in 1965 and is the moderator of this seminar) held a course in modern control engineering that really opened our eyes in Mölndal. Suddenly you got a completely different understanding of the control theory when you could start using these state variables. For the first time, you heard the words "Kalman filtering". It was a stable theoretical model for how to build filters, it was as if it was made for the explicit purpose of filtering radar signals.

Suddenly you could use the uncertainity measurements that are a part of Kalman filtering and suddenly you understood what you were really doing. Earlier, we had worked a lot on our intuitition and by practical trials. But if you want a Kalman filter for a radar, you need a computer. You can't do it without one, and that's where the idea of putting a computer in the radar came from. Bengt Sjöberg was a bit hesitant at first. He did some simulations, got a bug in the program and the results were no good, but then he got on board with the idea too. I suggested to Jörgen that we should have a computer and he too agreed. When you started looking at what the radar really needed, the computer became obvious, really. In order to manage the signals processing you have to constantly keep track of how you're flying and you kinda have to calculate what the ground clutter will be like and compensate for that and so on and so on.

Another thing that was very nice was that we could use data from the inertial navigation system in the radar directly. We subscribed to updates on roll, pitch and course data and could calculate ourselves, inside the radar, where the antenna should be pointing. And make a Kalman filter. The coordinate transforms that were needed from radar data in polar coordnates were to a ground-fixed coordinate system where the target aircraft resided. We could do that ourselves and didn't need any gyros. We worked with very high data rates, the shortest cycles in the computer were just under three milliseconds long. The control of the DSP was slightly slower.

Of course, FMV knew the technical risks of a complicated radar and wanted an American company that could verify the system and cooperate with us on the development. That's how Hughes Aircraft Company entered the picture. In 1972, both parties had proposals ready for the radar architecture. It turned out when we visited them in the US in 1972 that we were both thinking the same things about computers and DSP, and it was pretty easy to cooperate and make a basic proposal for what this radar was going to look like together. HAC also worked with us for a few prototype versions for the first prototype generation of the DSP software. We built a computer with a core memory of eight kilo-words - 16-bit words - (16 kilobytes, in modern parlance) for the first prototype generation. It was an assembly machine, pretty similar to the minicomputers that existed at the time. Then Ingemar Carlsson came with an expert from the Chalmers Institute of Technology, Gunnar Carlstedt was his name, who took a look at our solution. He said it was way too clumsy, what we had done. Too much hardware, he said.

Microprogramming (referring to programming a CISC processor) of computers had become pretty hot at the time. And that was what he suggested, that we should use microprogramming and do less in hardware. And that's actually what ended up happening. When we entered series production there was FAMOS memory on the market. That is EPROMs where you could write a program electrically and then erase it with ultraviolet light. The first units were shipped with those and it lasted throughout the entire 1980's. The radar was delivered with a 32 kilo-word program memory and a data memory of 2 kilo-words. Then there were program cycles with lengths from 3 milliseconds and up to 72 milliseconds. It was a bit different depending on a number of factors. Internationally they called this a "software controlled radar", which was used in the marketing.

As a host computer, we bought a Honeywell H316 mini-computer early on. We used it for most of the 1970's as a host computer for the software development, until we got our VAX system in the late 1970's. One important part of the computer was that we had trigonometry lookup tables which meant we could do sine and cosine operations and such things extremely quickly. A lookup only took a few microseconds and I don't think I've seen any computer after that with such fast trigonometry. Later on the computer was replaced by a Texas Instruments signals processor, TMS320, with performance in a completely different league. That came in the 1990's. Suddenly radar functionality became easy, suddenly you had plenty of memory and processing capacity. Maybe there's questions?

(...)

Moderator: I'll pass the word on to Jörgen, who as we've heard was a benign boss.

Jörgen Nilsson: (b. 1926, electronic engineer, head of the PS-46/A development project at Ericsson) It was very nice to hear what Ingvar had to say. If you want to criticize or show you a different aspect of it. It wasn't all that easy as a project manager, to keep track of what everyone was doing. Sometimes they just came to you and told you what they had done. Sometimes it could be very surprising. Acke Axelsson was involved in air-cooling the transmitter, which meant finding an entirely new transmitter tube. I was completely against that. But our guy who worked on it, he simply said, well I talked to Acke, and it's gonna have to be air-cooled. Then we had several years of issues with trying to find tubes that could handle air cooling. Sometimes it was like that, you just had to roll with it. But it was very nice of Ingvar to call me a benign boss - it was a very interesting job for me.

To us down in Mölndal, it was a very big step, in a lot of ways. In the 1950's and 1960's we were very dependent, we more or less did what FMV told us to do. But then we came to stand on our own legs and I think that had a lot to do with the JA 37 project. We discovered that we were actually pretty good at this radar business and we started getting more customers, we could export and so on. And we could manage ourselves, we came up with new things to do on our own.

Ulf Frieberg: To me, getting to the MPD thing (MPD is short for "medium pulse doppler") was more or less a detective thriller. Because without MPD we would've had nothing. Can you tell that story?

Moderator: Well, Jörgen, can you?

Jörgen Nilsson: Yes. I was working with radar development during the 1960's. Pulse radar, pretty simple things you could say, looking back. The Air Force Administration contracted studies of pulse-doppler radar to a whole bunch of companies, to us, to SRA, Saab, AGA and Teleutredningar AB. We got our share of the money. The contract was split into parts, we got to do pulse-doppler and some others should work on continuous wave (CW) or intermittent CW, and yet others on other things. We got pulse-doppler and many regarded that as being pretty hopeless with a lot of stability issues. We didn't care all that much about those naysayers and built a prototype for our own money that we tested on the ground around 1965. And it turned out it actually did work pretty well.

That gave us the courage to work on aircraft-mounted radar. The big problem there was that the radar was mounted in an aircraft that was rushing along above the ground returns at a certain speed. There were vibrations and a horrible noise up there, we were in the very tip of the nose. And that affected the radar, the stability and everything and we were very worried about that. I asked one of my engineers to make a study of the movements between the radome and the antenna, if that was going to affect anything. He concluded that it didn't look good at all, that it probably wasn't going to work. But I realized that it had to work, so I took his report and put it at the bottom of my desk drawer. Then I don't know what happened to it, I think it laid around there for many years.

The big issue was the ground returns, you had to track the speed of them. The thing we heard was that we had to approach a resolution of 50 Hz with a 9 GHz carrier, which seemed pretty hopeless. But we tested it in a flying prototype and to our surprise it actually worked. We got a 1-sigma of about 50 Hz. And then we weren't so afraid anymore. Then the work started for real and it was very good that we got to meet with the Americans and figure out that they knew about the same things as we did. They had tested a bit more than we had though, and that was good. I don't know if I should go on?

Ingvar Sundström: Then there were a lot of discussions about whether we should use low or high PRF?

Jörgen Nilsson: Oh yeah, that was what we were going to talk about. Using low PRF was a very attractive option, since you could just take the radar we had had in the AJ 37 and with some small changes, adding some equipment, make it into a pulse-doppler radar. But then we figured out that that wouldn't work very well and that we should switch to what was a called a coherent transmitter system. But many argued in favor of keeping the low PRF idea like an ordinary pulse radar, but with a coherent transmitter that was nice and pure. But it turned out that such a radar would get us in trouble with moving targets on the ground, such as cars, trains and other vehicles. We did some studies of it, I think we even test flew something. Volkswagen Beetles resulted in big targets and we weren't interested in that.

Eventually it became apparent that in order to be able to use a low PRF on doppler radar you had to approach the target in a very specific way, because you could get into "blind spots" for the radar. And there was quite a bit of discussion in these project groups if you could mandate certain attack paths and so on. But when looking back on that today I'll have to say that was very unrealistic with regards to the pilot. The pilot wanted as little trouble as possible and a great deal of tactical freedom, of course. This led us onward to what is called MPD, which did give you that tactical freedom. You can approach the target from the front, from the rear and so on. There was only one downside, that the range was reduced somewhat, but that was acceptable. Then the range has increased a lot during the work on the radar, so it was a pretty fortunate choice. And the pilot did get his tactical freedom of choice, didn't he?

Leif Åström: Yes, definitely.

Moderator: Gunnar wanted something.

Gunnar Lindqvist: Well, Jörgen kinda talked about this. But there were high PRF radars back then, in the F-4 Phantom, I think it was a Westinghouse radar in there. And we chose, for the tactical freedom among other things, a MPD. But it might be a good idea to clarify two things further. When we went on to the JAS 39 we got a multi-mode radar, where you can mix both high and medium PRF. So the first question is, why couldn't we do that on the JA 37? And secondly I think you should talk a bit about the importance of the antenna studies you've done. Building a low-sidelobe antenna and a transmitter with a spectrally pure signal are the two fundamentals for a doppler radar.

Ingvar Sundström: When it comes to multi-mode radar, if you have a high PRF you need a shorter pulse length. There are various types of traveling wave tubes you can use. They're usually designed for a certain work factor, that is to say the relation between the pulse length and the pulse repetition frequency. In the JAS 39 they solved that with pulse compression in the transmitter and the receiver, pulse expansion when you transmit and pulse compression when you receive. And that kind of technology wasn't very well developed in the early 1970's, I think it would've been very hard to do that. Later transmitter tubes with two different work factors, "dual mode tubes", and that would also have solved the problem. But those didn't exist in the 1970's either. It was hard, really, you had to chose one or the other.

When it comes to the antenna, the side lobes are looking straight down into the ground in level flight. They're usually very powerful and you can't filter them away like you do with the main lobe clutter. Instead, you simply have to design an antenna that minimizes the side lobes, and there was a fantastic amount of work done on designing a Cassegrain antenna with small side lobes. Olle Dahlsjö was responsible for managing that. Additionally they did some things to the radome, they called it the "necktie". It was a grid of radar-absorbing stuff in the lower part of the radome that helped minimize the side lobes. In the JAS 39 they have a different type of antenna, a phased array antenna with a fixed grid of about a thousand radiating elements, which also gives very small side lobes.

Acke Axelsson: (b. 1922, so 85 years old at the time of the seminar, radar engineer at FMV) You've probably already mentioned what I was going to say, I'm a bit hard of hearing and can't really follow the discussion. But I wanted to say something about the choice of medium PRF. We chose to use a magnetron for the trials. You take the signal that you just transmitted, you take the phase of that and make a reference signal, and you use that in the receiver. Then you get all the echoes from the target within an unambiguous distance so you can give them the doppler treatment. But when it comes to the second, third, fourth time around, ambiguous echoes, you can't handle those. So we just skipped those immediately. Other than that, Hughes Aircraft did signals processing just like we did at the air force test center.

Then there was the question of medium or high PRF. The tactical methods of the time were either the gun, or that the air combat controllers guided you to a distance of about 10 or 15 kilometers from the target, and then you approached from there to attack with radar or IR missiles. High PRF wasn't suited to that, so medium PRF which could be used as low PRF if necessary was chosen. And that's the background for that choice. I don't know if it's been a downside really, I think it's worked out pretty well.

Then there was another factor, and that was the result of the Air Defense Investigation of 1967, LFU 67, which among other things studied ground-based missiles. In a search radar for a system like that all the clutter, all the side lobe clutter, it has a frequency of zero, so it was easy to get rid of it. On the other hand, if you mounted the radar in an aircraft, the clutter spread out over basically the entire doppler spectrum. Since it's impossible to filter out all the clutter you have to try to choose an antenna that prioritizes the direction the target is in. And it's really tricky to both get a good antenna and to do the signal processing right.

When we visited Hughes Aircraft in 1972 Ericsson presented their radar and Hughes presented their proposal for signals processing. They were trying to develop a PRF pattern, the standard was to choose three main PRF's, each of which had two partial PRF's, in order to sort through all the ground clutter. And then you ran that pattern in a few variations. But eventually it was concluded that it's really only one PRF that discovers the target, or maybe two at the most. Then you just waste a bunch of time transmitting the other PRF's! In the last stage of development for the JA 37 around the year 2000 they figured out that you ran with one PRF until you found the target and then added a sub-PRF to determine the range. That was mostly based on the fact that the target fluctuated a lot in amplitude, depending on its path through the air flow and whatnot. The theory was that if the target happened to be in a position which gave a very strong echo, you kept it for a few milliseconds, or maybe even about a dozen milliseconds. If you had one PRF that discovered the target you should quickly take the opportunity to choose a sub-PRF to determine the range. And that was one of the last software modifications that was done at Ericsson for system 37.

Ingvar Sundström: Yes. It wasn't long ago at all.

Acke Axelsson: No. And that, together with some other things, resulted in roughly doubling the detection range. Doubling the detection range, that's a lot, that's 12 decibels. But there were some other things involved too, a microwave amplifier and so on. Well, that's what I had to say.

Moderator: Thank you. Kim?

Kim Bengtsson: I was thinking about the radar and how it hooked up to SYSIM. In the beginning, we only had a model that provided radar output. But with the development of computers, we could later build a radar simulator, RADSIM, which generated simulated inputs to the radar's computer and gave us the possibility to run the real radar software in a realistic environment, with the actual flight-verified programs running against each other, which was pretty valuable. And the engineers could sit there and discuss what was going on with the pilot.

Ingvar Sundström: Yes, when they came out with TTL-series multiplicators it became possible to make a signal generator that could create the signal vectors we needed inside the radar. And then you could put the real devices, the real flying ones in the SYSIM, and test the real flying software together with the entire system.

Kim Bengtsson: That was in the mid-70's?

Ingvar Sundström: Yeah, mid or late 70's I think. And that was a big step forward.

[Belesarius]

TL:DR:RN

 

(Too long, Didn't Read, Right Now)

 

Will be coming back to this later when I can give it appropriate attention.  Thanks for posting this Ren.

[Collimatrix]

Some takeaway:

 

 

 

The Draken had a complicated targeting system. It wasn't just a radar, but also an infrared sensor that unfortunately wasn't too reliable.

 

This is part of why I have become skeptical of the purported stealth-busting ability of IRST.  The unit on the J-35 didn't work well, and test pilots who have flown in East German MiG-29s have been unimpressed with the Russian models as well.  

 

 

 

You're talking so much about that tactical display, but I guarantee that back then when you were going on about it, there wasn't a chance in hell to actually make it work. We had no working solution. Map instruments in aircraft did exist, optical ones. Ferranti had implemented one, it was based on a 35mm film running around like crazy, rotating and whatnot.

 

And now we have all-digital systems in smart phones that do this easily.

 

 

 

The first step, the introduction of the pulse-Doppler radar, that marked the end of the days when the men of honor and silence rushed along between the treetops of pine and fir and nobody could see us with radar.

 

 

If you'd asked me before I read this what the single most important innovation in the JA-37's avionics was, this is what I would have said.  The Doppler filtering was a big deal and not many other aircraft at the time had it.

 

 

 

We had the best fighter gun in the world on the aircraft, in fact significantly better than the one that's on the JAS 39.

 

The GIAT 30M791 and GSH-30-1 beg to differ, but the KCA was a very good gun.  It's a pity it wasn't more widely used.

 

 

 

 

We had the choice between the Sidewinder 9L from the US or developing a Swedish missile with the development designation Robot 72-12 (in military Swedish, a missile is for historical reasons called a "robot" - which in ordinary Swedish means the same thing as it does in English). The difference was that the latter would have become a good missile if we had spent a ton of money on it. But we thought the development would take too long, so we bet on the Sidewinder instead. The thing was that we had to wrangle with the Americans for nine years before we were finally allowed to buy it, and in the meantime it became obsolete. 

 

I wonder what their beef with the AIM-9L is.  In 1982 the Brits had an 80-something percent hit rate with the 9L against the Argentines.

I mean, sure, the R-73 runs rings around the AIM-9L, but there were no weapons outside the Communist bloc as good as the R-73.

 

 

 

When we went on to the JAS 39 we got a multi-mode radar, where you can mix both high and medium PRF. So the first question is, why couldn't we do that on the JA 37? 

 

 

Interesting question.  If DCS World is to be believed, the N001 on the SU-27 had variable PRF.

[renhanxue]

This is part of why I have become skeptical of the purported stealth-busting ability of IRST. The unit on the J-35 didn't work well, and test pilots who have flown in East German MiG-29s have been unimpressed with the Russian models as well.

Really? The IRST on the Draken was early 1960's analog technology. Making arguments about today's IRST based on that is, uh, a bit of a stretch.

I wonder what their beef with the AIM-9L is. In 1982 the Brits had an 80-something percent hit rate with the 9L against the Argentines.

I mean, sure, the R-73 runs rings around the AIM-9L, but there were no weapons outside the Communist bloc as good as the R-73.

I think the actual problem there was that while we were waiting for the 9L to materialize we had to make do with the old 9P, which was hardly impressive in 1980. We didn't get the 9L until the late 80's and by then the 9M was everywhere and I think people were jealous.

[LostCosmonaut]

I'm not sure you could even generalize the issues with MiG-29 IRST in the 80s to now (for instance, look at the gap between the -9M and -9X). Unfortunately, any info on modern IRST systems is probably classified. But I doubt the Russians would invest that much in them if they were completely useless.

[Collimatrix]

AIUI Russian IRST is still a non-imaging system, even on the PAK-FA.  It's just a very sensitive IR detector that looks for hot blobs and gives a bearing.  It can't identify targets and it can't give bearing or range data (unless they're close enough to tag with the built-in laser RF or if the radar is slaved to the IRST).  Compared to radar that's really bad.

 

I suppose that's a useful auxiliary if your plane's radar is horrible (see: MiG-29), or if you don't have AESA and you're afraid that going radar hot will suck up AMRAAMs (see: MiG-29), or if you're somehow stuck fighting people in JSFs and your radar is literally useless.

 

F-35's EOTS is an imaging system; essentially a FLIR that can do IRST things too.

[renhanxue]

And so are all the Eurocanard ones. If you want to say that the current Russian ones are worse, sure, I'll buy that, but it doesn't really have anything to with the state of the Draken's IRST in the 60's or the MiG-29's in the 90's.

Speaking of the men of honor and silence who rushed along between the treetops of pine and fir, though, I recently saw a former Lansen and AJ 37 pilot relate this anecdote and I figured I might as well stick it here:

 

On the strike side of things, we did radio silent exercises pretty often. As in, everything was pre-planned in the briefing room. Exact times for engine power up, taxiing out, take-off, strike time, as well as the type of attack, what formations to use during the mission etc. The air traffic controller was either there when the orders were given or was briefed separately so he could keep the runway and the airspace free. Then we flew the entire mission without speaking a single word on the radio until after the strike was delivered. After the strike, the flight lead called for everyone to report in. Sometimes, it could actually turn out that someone was missing, usually because they had missed the take-off or encountered some technical problem. In an emergency, we would of course break our silence.

Usually though, you'd know if someone was missing. We had a system where the flight lead was parked the furthest away from the runway, number 2 next to him and so on - on the wartime bases, that is. So when I as a flight lead started taxiing, I went past everyone else and looked into their cockpits - they'd do a thumbs up or thumbs down, and I'd know. Of course, pilots would sometimes have to abort at a later stage for whatever reason but that was usually not cause enough to break radio silence. The problem was if I myself as a flight lead got in trouble. Then I had to be really damn quick about informing someone, usually number three in the flight, who got to take over the lead. And I had to communicate this with maneuvering and sign language. What to do in a situation like that was naturally pre-planned too.

Things got even more interesting when we did exercises with other squadrons that took off from other bases and had to gather the flights in the air under radio silence. Each flight got a waypoint and a time when they were supposed to meet the other flights to gather up into a bigger strike package. You really had to be careful when calculating flight times and that you had your clocks set right.

It actually worked surprisingly well. The coolest experiences you got on some big joint exercises when we got to do all-E1 strikes with live weapons. First you gathered the entire strike package up, 12 or 14 flights (of four aircraft each) that all took off from separate bases, and then attacking with a 30 second separation between the strikes. If you arrived near the end, it was usually really hard to see the targets because of all the smoke.

The Swedish cold war era air force did not believe in loitering.

[Belesarius]

 

The Swedish cold war era air force did not believe in loitering.

48 Aircraft doing a preplanned form up and strike, radio silent?  Jesus, that's gonna take some training to get right.

[Collimatrix]

Really? The IRST on the Draken was early 1960's analog technology. Making arguments about today's IRST based on that is, uh, a bit of a stretch.

 

 

I took "not very reliable" to mean "not an adequate means of detecting aircraft because of problems inherent to the concept of IRST."  Did they actually mean "it was made of vacuum tubes and animal sinews and other pre-transistor technologies and thus caught itself on fire all the time" kind of unreliability?

 

 

48 Aircraft doing a preplanned form up and strike, radio silent?  Jesus, that's gonna take some training to get right.

 

Yeah, that's nuts.

[renhanxue]

I took "not very reliable" to mean "not an adequate means of detecting aircraft because of problems inherent to the concept of IRST."  Did they actually mean "it was made of vacuum tubes and animal sinews and other pre-transistor technologies and thus caught itself on fire all the time" kind of unreliability?

To be honest your guess is as good as mine, the "not too reliable" part is a pretty literal translation of the original Swedish. Personally I'm inclined to think you're right though, it simply wasn't all that good at detecting aircraft. But it could also be a little from column A, a little from column B.

The JA 37 didn't get a similar unit, after all.