Designing and Building Simulators

Rob Jonkers's picture

by Rob Jonkers

Ever wondered about the usability of a flight simulator as being close enough to resemble real flight? Well, with computing power and flight control technologies available today, the ability to achieve the realism is getting ever closer. My own dabbling in simulators has been around since Microsoft released their very first MSFS back in the late 80s with only Meig’s field in Chicago as the airport available to fly from using a green flat earth terrain mesh.

Progress has made phenomenal strides since then with photo-realistic scenery being available these days, and even Google Earth terrain mapping. Flying real aeroplanes is getting more and more expensive in this day and age from the time my flying started at R40/hr in 1980 on a Cherokee, and issues of currency, checklists and procedures being front and centre in our world, and thus the decision made to build a simplified cockpit panel to practice various in-flight scenarios from the comfort of the lounge. Previously memorizing and pattering checklists you would do in your head and imagining the switches and controls to move, this being less than ideal where mostly you would tend to skip things, those pilots reading this would probably relate to this. Then of course you can sit on the ground inside the aircraft and talk to yourself, gesticulating all the actions as it would be needed for flight scenarios, with the benefit of seeing and touching the relevant controls and switches, the problem there is that there is no pressure of the actual flight and with no action, plenty of time to ‘do’ vital actions.

My first panel I designed to be compatible to fly twin piston props, and included magnetos, fuel tank selectors, fuel pump, alternator, light switches, etc. I also included a radio/avionics stack with Nav/Com/Xpnder as part of the set-up, and some basic Autopilot functions. This panel worked with Microsoft FSX and had one screen, the lower half of the screen showed instruments, the upper half the outside visuals. This set-up worked very well and as an option for anybody wanting to have a more interactive simulator without having to remember keyboard commands for functions a good way to go.

The biggest limiting factor for VFR type flying and especially practicing in-flight emergencies is the visual field of view being so limiting with one screen. Having tried a head tracking system, which although workable is somewhat disorientating, hence the decision to add more screens to achieve a surround view, and in my desktop sim set-up have 180 deg FOV and a dedicated instrument panel screen. This allows one to do circuits very easily without ever losing sight of the runway. Further realism with instrument panel overlay templates was added to be able to simulate more accuracy in real aircraft instrument panel layouts, where without much fuss you can within minutes change from flying a B1900 to a B737. At around this point, I moved onto using X-Plane 9 as the software of choice, having higher frame rates and more realistic flight models, this of course is a great debate along the lines of the Cessna vs Piper of which is better... enough said! Further than X-Plane, most of the instrument and control software is bespoke needing some knowledge of the C language to write control and display functions.

Around 2009, having had some experience under my belt on building a simple sim, I thought why not try out a full-on full scale airliner as a specialist project, having seen such projects in work through Internet information, although these were mostly based on chopped up airliner cockpits being rescued for such applications. There are so many airliners being sent to the boneyard, the availability of airframes has skyrocketed, sadly not many in SA, and the cost prohibitive to import/transport a cockpit.

My decision was to base a full scale build project on a B737-800, now why the 737 one may ask, in my opinion it is still a pilot’s airplane, with traditional conventional controls and manual control ability. I knew of course it would take some years to build, and a reasonable budget, but seeing what was available as on-line information, decided to take up the challenge, this of course also making it resemble the actual cockpit as accurate as possible. Now you may ask why such a complex project - for fun of course, which is as good a reason as any other.

So, where does one begin? Well, being a system engineer by training, a system architecture diagram is a good place to start, and from there you can start deriving all the required functions and interfaces and from that a picture of what hardware is necessary emerges. Then it is a matter of designing in your favourite CAD software tool the details of the design, taking into account things like in how many sections will the cockpit be split into (logistics of moving it around probably important for getting it in and out of a room), what to make the structure and parts from, and how to assemble/install subsystems to each other.

The decision was made to manufacture all of the structure from Supawood in 12, 6, 3 mm formats, this type of wood is easily cut and shaped. The panels would be made of plastic sheet with some in Perspex for means to display backlighting, and most of the switches being of specific shape and function not easily available off the shelf would also need to be manufactured, mostly involving turning metal bits on a lathe. In this day and age of CAD/CAM being more available and affordable, most of the parts were manufactured using panel router and waterjet equipment. With 3D printing now becoming available, consideration to make parts, especially control knobs will be the next jump in technology in making bespoke parts.

This project is a long term one, with many weekends not taken up by real world flying with design and construction, with the project now at around half way and at least flyable in all axes with the power quadrant also active. The most challenging so far has been the development of the surround visuals layout, having taken around 6 months to perfect, with consideration to field of view, design eye point, image distance, merging of 3 projectors through a single computer, applying anti-warping and image knitting, this mostly having been prototyped on suspended bedsheets until a good result was obtained, the overall visual presentation is arguably very realistic even fooling the brain with pseudo motion cues.

To give an idea of the hardware requirements, it will be running with 7 networked PCs, 7 LCD screens, 3 projectors, 25 control boards controlling around 400 switch connections, 20 led displays, 15 servo controlled instruments/controls. The project will also include a sound system simulating cockpit noise, warnings and a low frequency shaker system to emulate ground operation and engine rumble. All in all a daunting project, but good engineering fun!