This page has quite a number of pictures, all of which link to larger versions of the same image. All the larger images are from 40 to 80 KB in size, so don't click on them unless you want to wait for the download.
Around late 1994, I approached a company I was dealing with for other reasons with a design for a simple, cheap full motion interactive simulator system, designed to be networked together in clusters of up to 16 units. To cut a long story short, they agreed to buy the parts and provide the space for me to build a proof-of-concept prototype, with a view to making, installing, and operating a network of at least 8 units. The design goal was a complete simulator with the following features:
The last two items were the really important ones. The low height clearance was because the building it was intended to use the things in had low ceilings and it wasn't economically viable to raise them, and the low cost was to allow a sufficiently large number of units to be installed to make the exercise worthwhile, while still meeting a quite strict budget. At the time, the nearest commercially available system cost around 72000UKP, which meant that, in theory, we could install almost five UMVS systems for one of the competitions.
A few months of experimentation on simple control and motion systems ensued, while I worked out what was possible for the price allowed. I discussed the matter of hydraulic versus electric actuators with a number of suppliers and colleagues, and eventually came down on the side of hydraulics, mainly for reasons of cost, availability, and power density. Electric systems were available, but cost considerably more, were much bulkier, and didn't move as fast. A couple of primitive prototypes were built of proposed hydraulic systems, but unfortunately no pictures survive of them.
By early 1995, I had built a working prototype of a simple two DOF motion base, and a prototype single seat pod. I and a co-worker, Rob Voisey, designed the shape of the pod together, and then I built it over one weekend. The motion base was fabricated from heavy-duty 1"x1" and 2"x1" steel box section tube, and the pod was made using much lighter grade 1"x1" box tube. Some pictures of this version are shown below. The pod was temporarily covered with 6mm ply, to light-proof it during tests with the video system. The box at the front housed a large monitor, which was visible from the seat via an arrangement of mirrors. This was used because no projector was available at the time. The seat was from a Namco Ridge Racer game, and the system was controlled via the PC joystick visible in a couple of the pictures.
The system used two single-acting rams for each axis, in a somewhat unusual push-pull method of control. It had a couple of disadvantages, mostly in mechanical complexity, but these were outweighed in my opinion by the advantages, namely very fast and precise operation, instant stopping capability, and identical speeds in both directions. Normally, of course, using a single, double-acting ram, the speeds of movement are different in either direction, assuming pressure, flow rate, and load are constant. This is because the area of the two sides of the piston differs, due to the pushrod attached to one side. Compensating for this is quite straightforward in software, but increases the complexity of it considerably, since the difference varies with any of the three factors listed above, which change in operation. As a result, you have to monitor all of them and compensate accordingly.
This first test system proved most of the individual components worked very well, giving a range of speeds from so slow you couldn't see it moving, to +30 deg to -30 deg and back in less than half a second. This last speed was far too fast for safe operation, and at one point managed to catapult me several meters across the workshop! However, it also showed that the motion base, while amply strong, was much too flexible, and would have to be redesigned using significantly heavier materials.
By late 1995 a number of further tests had resulted in the next prototype. Progress was considerably slower than expected due mostly to funding problems, but eventually I built a new motion base, using six inch section, 1/4" thick. I wanted to use six inch box tubing, but didn't have the facilities to cut it in-house. As a result I had to weld up my own box tubing from 6"x1/4" plate, which took quite a lot of work. The end result was functional, but somewhat more flexible than it should have been. Even so, it solved most of the problems and the new base worked very well. The following are some pictures from this period.
As you can see, the pod had also been modified. It now had a pneumatically operated door, and the framework for the joysticks, seat, and screen had been fitted. The joysticks settled on were designed for arcade use, since they were incredibly heavy duty, and easily serviceable. There were two sticks, one for each hand. The seat mountings were designed to take a standard after-market car seat on slider rails, complete with a five point harness. The hydraulic power pack I built ran from a single-phase 13A 240V power outlet, and produced a pressure of 180 bar at a flow rate of around 9.5L/minute. The system could be run at a considerably lower pressure, but was usually set high enough to give a fast response. It had a tank capacity of about 25L, which was not strictly speaking enough, mostly because of cooling constraints, but worked fairly well as a testbed system.
The motion base was now modular, so it could be disassembled for transport. All the parts would fit through a standard doorway, and could be moved (just!) by one person, weighing a total of about 300kg. In fact, on one occasion, I disassembled, moved, and reassembled the entire system single-handed, although I wouldn't like to repeat the process. At this point the base was purely two DOF, although a third axis (yaw) was in design. The hydraulics were very odd in design, using off the shelf parts in ways the makers probably hadn't intended, but it worked. The pod was the main problem, as it was a single unit, that couldn't be removed from the room. The door design was not ideal, either.
Over the next several months, until early 1996, a lot of work went into refining the motion base mechanics and hydraulics, and working on the drive electronics. A 3 CRT Zenith projector was purchased and suitably modified. I had to rebuild it to have the CRTs and the high-voltage circuitry in a custom-made framework in the front of the pod, and the rest of the electronics in a rack-mounted case external to the pod. This was because the drive electronics were quite sensitive to vibration. Anyone who has ever had to set up such a projector knows how long it takes, and I didn't want to have to do this every couple of days. The results of this system were quite good, but still weren't really what I was aiming for.
Unfortunately, at this point, progress came to a screeching halt. I had put in some 5500 hours over a period of about 18 months, and the system was showing a lot of promise. Most of the drive electronics had been designed and wire-wrapped, the mechanical parts worked pretty well, and some software had been written to control everything. However, continuing funding problems and personnel issues within the company paying for the parts and providing the workshop space caused the project to be virtually abandoned. Eventually, however, after a hiatus of several months, work resumed on the nearly completed prototype. The motion base prototype was working satisfactorily, and my attention turned to the pod.
By mid 1996 I had redesigned the pod and built another one, and also been employed full-time in June of that year by the company involved. I had previously been working on my own time, being self employed without a lot of other work. The new pod was also modular, once more designed to disassemble into parts that could be fitted through a standard door. The pod was based around a substantial chassis, which everything else, such at the sides, seat, projector, and so on, bolted to. This was to allow different versions to be made easily, and so the thing could be field-upgraded. No pictures from that period seem to exist, so the following are from several months later, in early 1997.
As the above images show, by now the new pod had its inner and outer skins fitted. The outside was made of polyester powder-coated aluminum, and the inside was 6mm fireproofed plywood with flame-retardant carpet glued to it. The carpeting was partially for effect, but mostly for soundproofing and cushioning. The door had been revamped, now opening in two parts, but was still operated pneumatically. The intention was to hide the door mechanism inside the front of the pod in the final version, mainly for safety reasons, but for this version ease of access was more important. The joystick mounts were covered with the ply at the time of these photos, but not yet carpeted, the same with the footwell and floor. The floor was made from 1.6mm mild steel sheet screwed to the bottom chassis.
A throttle control had been added to the console on the left side of the seat, and the videophone system was in place in the control console, at the front left of the cabin. This used a small CCD camera pointing at the occupants head, together with a 5" colour composite video monitor. It was intended to allow combatants to talk to each other during the game, either when they pressed the appropriate switch for the pod they wanted to contact, or if they had auto-talk turned on, to any other pod they targeted with the crosshairs.
The screen was made from 10mm Lexan with a translucent projection material bonded to one side. It was bolted into a solid framework fastened securely to the pod structure, to prevent vibrations distorting it during operation. The thing was so solid that you could punch it as hard as you wanted, and you'd just hurt yourself. It had to be very robust to withstand the abuse not only of normal use, but of users who didn't like losing.
There were six speakers fitted for the sound system, one on either side of the screen, one on either side of the seat at the back, one in the middle of the ceiling, and one in the floor under the seat, along with the bass bin. This was a 45L ported reflex unit made from 25mm MDF, and had a 450W Kenwood 12" driver fitted. It gave excellent performance, driven from a 150W MOSFET amplifier with a 200Hz low-pass filter on the input. The speaker was directly behind the seat, about six inches away, and you really felt explosions. It was an odd shape, basically a wedge, to fit the inside contours, and occupied the entire bottom rear half of the pod.
The sound system was driven via an 8x8 analogue fader board I designed, based on an idea of another co-worker, Dan Wilson. You can see him "enjoying" a test ride in the first picture from late 1995. He suggested the theory behind the system, which was given the name 'SoundSphere', and designed the multi-channel amplifier board for it. The system would basically allow the steering of any of the eight input sources to any or all of the six speakers forming a face-cubic array within the pod, with any desired fading level. It produced the most amazingly accurate 3D sound I've yet come across, and could be used in conjunction with any other sound processing system, such as SurroundSound, etc. The fader board also controlled the bass bin and the videophone audio system.
During this period of time, I was also writing driver libraries and test software for the sound system, the motion control, and the other peripheral input and output devices. I produced a number of PCBs for the SoundSphere controller, the hydraulic valve driver, the amplifiers, joystick and position sensors, and so on, which took some time away from the mechanical design work. One of the PCBs is shown below:
This is the new SoundSphere controller, which also had a 24 bit digital IO port (used for various control functions, such as communicating with the high-speed serial ADCs used for the joysticks and position sensors), and a bipolar 8 channel DAC which drove the valve power control board. This updated controller increased the original 8x8 matrix to a 12x10 one, with the eventual aim of driving an eight-speaker corner-cubic array to increase the 3D effect of the sound.
I also rebuilt the power pack, uprating the motor to 5.5kW, which necessitated switching to 3-phase power. The new unit could supply 24L/minute at 180bar, which was sufficient for two complete simulators. It was intended by this time to make another unit as soon as possible, to allow us to test the networking and communication software between multiple units. We were also going to hire someone to produce the final engineering drawings for the system, and have it 'productionised', ie. have the design checked over, and if necessary modified, by an engineer specialising in making it easier to manufacture in a factory environment. Once I had built the new power pack, I also took the opportunity to move it to the next floor up, which was unused. This was to reduce the noise of the simulator test area while in operation, and also to improve the cooling. We had a hell of a job drilling two 45mm holes through the floor, which was 10 inch thick reinforced hardened concrete. It took several hours with an enormous drill, and made a real mess.
By late 1997, the prototype was mostly finished. All the major bugs had been worked out, and a new motion control library I wrote had improved the smoothness and accuracy of the movement a lot. We had switched to a new projector in mid 1997, using a Proxima DLP unit that was smaller, lighter, and much higher-resolution. This required a certain reworking of the inside front of the pod. We had also switched to using different hydraulic valves, which simplified the hydraulic system quite a lot. In addition, the position feedback sensors on the rams had been changed to use a contactless system, which improved positional accuracy.
I had also added the bright yellow panic switch you can see in the above image, and several other safety features. Everyone involved in the project at any time had made suggestions for safety-related subsystems, and almost all of the suggestions were used. The power behind even a small hydraulic system is unbelievable, if you've never seen it before, and it is very important to have methods of shutting the whole thing down and dumping any stored pressure safely and quickly in the event of any fault. The hazard is made even worse because of the eerie lack of noise. If you can't hear the pump, the rams just make a quiet swooshing noise, even as they throw a half a ton of simulator and motion base around at really silly speeds. People don't expect that anything so quiet can be so dangerous, until they get in the way!
The final system required the harness to be on tightly, both joystick triggers to be pulled simultaneously, and both feet to be on the floor before the door would shut, and it wouldn't start up until the doors were fully closed and locked. If the harness was removed during operation, or the panic button was pressed, the system would shut down completely and open the doors, and it couldn't be restarted until the fault was cleared and an external reset switch was triggered.
A full-time programmer had been hired earlier during the year, and the first task he tackled was rewriting my sound control library using a much improved algorithm. This took some time, but the eventual results were fantastic. It was possible to fly the demo program around and follow the target plane just by listening to the noise of its engine, even with your eyes shut. The realism was incredible. He then switched to writing the actual game code, which was to use 3DFX voodoo cards for the graphics output. My other library routines for controlling the rest of the system were kept intact.
After interviewing a few potential engineering companies, we finally ended up hiring a friend of mine who used to work for Bendix as a draughtsman and engineer, in precisely the capacity we needed. His job had been to take prototypes that worked and convert them into products that could actually be made at a reasonable cost, and to produce the engineering drawings. He took a lot of pictures of the motion base, measured it, and went "Hmmmm" a lot. After talking it over with me, he went away and produced some drawings. He and I looked them over, argued a bit, and he went away again, coming back with some more drawings. After this had happened several times, we agreed on the final design, which was functionally identical to my original system but much easier to make in a production environment. It used 3", 4", and 6" x 1/4" box tubing throughout, which was cut for the initial unit by the company that supplied it to us.
The hydraulic rams were upgraded considerably, as well, to overcome some potential failings of the original, smaller ones. We had a new motion base welded together by an outside company, and installed it next to the first one. The drive software had to be modified slightly to take into account the somewhat different characteristics of the new system, but everything went quite smoothly.
We went through the same exercise with the pod, lengthening and widening it slightly in the process to make more room for the new projector and to allow two seats to be more easily fitted. By mid 1998, the final production prototype was nearly finished. The door now opened by means of an electric winch and cable, which enabled all the potential danger spots to be removed, so people wouldn't get fingers removed if they ignored the safety warnings. The new pod had an even larger screen, and the new motion base moved slightly further than the original had done, but overall it was almost exactly the same.
At this point, rather unexpectedly, disaster struck, at least from my point of view. The owner of the company wandered in one day and said words to the effect of "We're dumping this system and buying in a much more expensive third-party electric motion base and pod. As a result we don't need you any more, so you're redundant as of now." It turned out that they were having a custom made electric motion base manufactured, and a new pod, and were only keeping the currently unfinished game.
Well. Four years, a lot of money, and a hell of a lot of work down the toilet. I had designed and built mechanical, hydraulic, and electronic systems, laid out PCBs, written software, you name it. I'd nearly finished the thing, virtually single-handedly for most of the design process, despite many delays and problems outside my control, and it was considerably cheaper than the original specification. We could have made the motion bases, even with the third axis that no-one but me ever thought was needed (funny, really, since one of the aspects of the new electric thing they made a fuss about was that it had three DOF!), for around 3000UKP each, and a complete simulator would have come in at just over 10000UKP per unit including computer systems and electronics. Over 2500UKP of this was the projector, as well.
What can I say? It pissed me off then, and it still does a bit. Anyone want a design for a cheap simulator?
Back to the beginning. Home