A VERY ROUGH SKETCH OF THE OVERALL PLAN

I try to keep these pages interesting to anyone who may happen into them, but coming to the subject of how you go about converting a scrap 737 cockpit into a flight simulator, the language will, by necessity, take a sharp turn South -- there's no way of avoiding technical terminology with lots of acronyms and other jargon that will numb the minds and eyeballs of some readers.  So, I admit that this page will not be to everyone's taste; it'll probably be most interesting to other sim builders.

 If you have the courage to dive in anyway, welcome, and I invite anyone to make suggestions on how I might better approach any of these subjects.


The cockpit is a B737-322, and I'm determined to keep it a -300 series even though doing that introduces several problems.  There are two at the top of the list that are about tied for first.  To understand, you first have to realize that most everyone else involved in similar projects are doing 737 NGs (-600, -700, -800 and -900 models), and that's if they're doing a 737 at all -- many are doing Airbuses; A320s seem to be especially popular.  A few others I'm aware of are working on B757 / 767 simulator projects.  There is limited commercial help for those, starting with the unequaled LDS 767 model.  These projects have the potential for spectacular outcomes.

Doing glass cockpit aircraft simulators (like 737 NGs and A320s) is a decidedly easier approach to take.  There's no denying that computer screens are just the right things for simulating computer screens – you know, the ones that dominate the panels of these modern electronic marvels.  So with most everyone else doing newer models, the top problem is that there's almost no commercial support for the -300, -400, -500 series.  Closely related to this is that there are very few who have tackled the earlier models, thus there’s not nearly as much support from the online community for doing a –300 as there is for doing  -700s, etc.  I have, however, found one who has immersed himself in a –300 project (successfully) and has been very helpful by passing his wisdom on to me -- I sorely need it.  Another, an expert from the UK in working with the OpenCockpits scripting language, has pledged his assistance in helping me to create my own scripts when I get to the point of needing to write them.  He has already written reviews of several OpenCockpits items that are posted at www.mutleyshangar.com.  These have already been helpful; these "reviews" would more accurately be called tutorials.

It's fascinating to see how everyone working on a home simulator project approaches it in their own way -- there is no norm.  It seems that each individual tackles the job in his own very unique way, although many have common threads.  The path each chooses is based on his/her own skill set, available space, experience, budget, and his  hopes & expectations for the project.

So, just to outline the basics for my project, the core simulator program will be Microsoft FS2004, or FS9 as we call it.  It will be at the center of everything else.  I'm sure some will question my judgment in not using the more up-to-date FSX (or its close kin, Prepar3D), but for my purposes it's clear that FSX would introduce its own set of problems, more than offsetting any likely benefit it would offer. FSX’s main claim to any advantage is in its scenery, which offers better detail and looks better in some locations, but the benefit is most apparent at low altitude.  From the flight levels (above 18,000 in the U.S.), no advantage.  FSX would be more attractive if I were pursuing a Cessna 172 sim, since you might do a lot of your flying below 5000 feet in that sort of airplane.  Whole different subject.  Granted, jets need to land and take off, but during those relatively few moments close to the ground, I doubt that closely inspecting fields, barns, automobiles, houses, trees and other ground details will be a priority.  The underlying reason for my choosing FS9 is its excellent (by comparison) frame rates, which will enable me to use fewer computers to run the whole project.  I can't prove this, but intuitively I believe that the number of problems that arise in a multi-PC environment varies exponentially with the number of PCs involved -- especially when they have to be in absolute synchronization with one another.   A secondary consideration is stability -- in my desktop arrangement, FS9 as been rock-solid for many years.  FSX users just seem to have more problems, based on my own limited observations.  For some examples of what the world looks like through FS9, see the Flight Simulator page of this site.

The basic aircraft model will be the Wilco/Feelthere B737-300/400/500 Pilot in Command.  It's not a widely-acclaimed product by any means (the driveling reviews of simulation magazines excepted), and its bugs and shortcomings are well known.  But if you want to do a 737-300, there just aren't many (any?) practical options.

These mid-generation 737s were originally delivered with two very different cockpits.  Both featured FMCs, but one version had full-analog instrumentation and the other was full EFIS, with glass ADIs, nav displays (which replaced the "steam" ADIs) and also electronic engine instruments.  Piedmont, America West, US Air, and Southwest were four airlines that opted for the analog cockpit.  I know that, at least in Southwest's case, they did it for fleet commonality with their huge 737-200 fleet (read: lower training costs.)  This simulator will be full-EFIS, as originally delivered to United.

The display system I'm shooting for will use three (maybe BenQ) short-throw projectors and a curved screen of hopefully 200 degrees FOV, 16 feet in diameter and having an image height of about 6' 3".  Probably (maybe?) all three projectors will end up being driven by one video port through a Matrox Triple Head to Go.  It's obvious from the beginning that a very high-end video card will be needed for that arrangement -- high frame rates -- under almost any conditions -- is one of my design goals.

Image-warping software will also be needed as part of the display/projection system.  The main issue is that the optics in most any projector (or camera, for that matter) have a flat depth of field and are designed to project onto a flat, not curved surface.  Uncorrected, the image would be distorted on a curved screen.  Further, the “seam” where the images of two projectors coincide needs to be blended; it is not possible to simply position the projectors in such a way that there will not be a line where the two images meet.  There are at least three programs to handle these problems.

At least to begin with, it looks like the captain’s and F/O’s instruments will have to be computer-rendered and displayed on LCD monitors mounted behind the panels with holes in exactly the right places and of correct sizes.  This approach has the distinct advantages of being cheap, accurate, and relatively easy to set up.  Drawbacks are that the instruments will emit their own light, rather than reflecting light, so realism will suffer to some degree.  Of course the monitors are ideal for displaying the 737-300’s ND and ADI, which are glass anyway.  Also, since LCD monitors are rectangular, there will be a rectangle on each panel behind which no switches, lights or anything else can be located, unless they are of very shallow profile and can be surface-mounted.  This precludes using the real panels that I already have.  In the future, hopefully actual (although converted to servo-driven) flight instruments can be installed (in their real panels), and realism stepped up a notch.

I’m not sure yet about the center instrument panel.  Hopefully I can get enough engine, flap, hydraulic, and standby flight instruments working with servo motors that I’ll be able to use that approach from the beginning.   One area in which realism will not be fully satisfied is the engine instruments, at least for this particular aircraft.  In this version of the cockpit, they should be the electronic type, but another variation of panel uses analog engine instruments in the EFIS-type panel, so the compromise will be very minor.

The overhead panel with all its switches, lights, Korrys, etc. will be real, but I’m not sure yet about the gauges on the overhead.  Real is my first choice – using as many real components as I can is another design goal.  As one online acquaintance told me, "nothing that's a replica feels real."

The (simulated) flight instruments will most likely come from the very similar (undocumented) 757/767 panel that is part of the Project Magenta package, unless the stock Wilco/Feelthere articles prove to be suitable for the role – unlikely.  Although the 757/767 EFIS displays differ somewhat from the 737-300/400/500 layout, customer options on both types blur the distinction between the two, so authenticity won't be set back as far as using displays from a different aircraft may sound.

Likewise, software support for the MCP, CDUs and EFIS control heads will be Project Magenta.  Software for driving the aircraft systems will be via the Prosim 737 Systems package.  Although the software is designed specifically for 737 NGs, the systems on the NGs and 737 Classics are nearly identical, aside from the numeric display of some parameters (mainly electrical) on the overhead.  Presumably those digital values can be converted to analog.  Using these NG systems will not result in any appreciable compromise with reality.   A complication of using the Prosim software is that it is known to be somewhat incompatible with SIOC (see below.)  But one mastermind has found his way through the mess and has written a 37-page paper covering all the details.

The hardware interface between the PCs and the physical controls and displays in the cockpit will be entirely OpenCockpits interface cards through USB ports.  OC make a wide variety of cards for interfacing a wide variety of input and output types.  There are cards for DC motors, USB inputs/outputs, servos, potentiometer axes, stepper motors, rotary switch encoders, and 7-segment display drivers, just to mention a few that come immediately to mind.  They also make replicas of MCPs, CDUs, radio and transponder control heads, and so on.  As I mentioned, the software to drive the cards and control their outputs will be with OC’s own scripting language, SIOC.  Using Project Magenta has another advantage in that there seems to be some alliance between them and OpenCockpits, and as a result a lot of the SIOC scripting has already been done for PM, though project-specific tailoring will naturally be needed.  Further, since SIOC communicates with the FS9 program itself through FSUIPC, it’s a benefit that there’s already a block of FSUIPC offsets reserved specifically for Project Magenta data.

The throttle quadrant/control stand will be a real one.  Four functions need to be motorized – spoilers, autothrottle, trim, and trim position indicator.  I may try to seek out an affordable, professionally-modified (motorized) conversion.  "Affordable" and "professionally- modified" may prove to be contradictory terms, however.

Electrical power for 28 volt DC aircraft components will probably come from surplus PC power supplies, 12 volts connected two in series for 24V, which should work well enough.  That takes care of the DC requirements at least – I’ll need to buy or rig up an inverter to make 24V AC.  Car battery chargers (again, two in series) or a homemade 24/28V power supply are other easy options.  Interesting that on the airplane, three transformer-rectifier units (TRUs) are installed to convert AC to DC to supply the DC buses, but in the simulator the problem is reversed.  A byproduct of using PC power supplies is the 5V DC component that’s needed for the OpenCockpits cards, servos, and some minor low-voltage DC requirements of the airplane itself.  I’m hoping to avoid the need completely for 125V/400 cycle AC.  Most of us just don’t have 400 Hz power available.  I read one article on how it can be made, however – reassuring in case it becomes a necessity.  Maybe I can use some of the original circuit breakers.  That also goes for wiring, although I’m not sure yet how practical that will be -- it will be a question of finding matching Cannon connectors.

With the simulator having no need for real JET-A, I look forward to the fuel system being easier to handle!  Same for the hydraulic, pressurization, flight control and pneumatic bleed air systems. 

As the project slowly advances, it will be interesting to see how much of this preliminary plan stands up to real-world realities.  Flexibility is bound to be an asset!
 


Original page, October, 2013