Macintosh/FlightGear - Graupner/JR MC-19

Well, I looked into this a couple years ago under Linux, and decided one would need to write a Joystick driver. I've got an interface cable that converts a Hitec PWM signal to an RS-232 data stream, and I started to look into writing a joystick driver that would work with that.

But, FlightGear didn't have a "standing on the ground" view, so it wouldn't have been all that useful. Is there a fixed-point view available now ("tower view" or some such")?

Are there any "model" models available?

I even found a cable that converts to USB:

switch to english, search for product #85153). This should work with all Multiplex, Graupner JR and Airtronics transmitters plus some Futaba, Lexors). Similar cable for HITEC see This will run FMS,, Realflight Classic, Realflight Deluxe,, Realflight G2 without Interlink, RC Plane Master, X-Plane, RCF-,Sim, ClearView, Microsoft Flight Simulators, Cockpit Master, PRE-Flight version 1,95,  Aerofly Pro (as 2nd interface), EasyFly, AeroFly, PicoFly according to this website.

Unfortunately most if not all of these run under Windows only - and FlightGear looks very nice to me apart from being free and extensible.

Yes, it does: It has the following views: Helicopter, chase, tower, tower view look from ?, chase view w/o yaw, cockpit

Also it runs on Windows, Linux, Solaris (sparc/x86), sgi, Mac OS X, FreeBSD and has the full world available as terrain, > 20'000 airports etc.

Not that I know of, but I thought of using one of the available models (motor, sailplanes, helicopters, jets available)

As I currently run a sailplane (Callistic Electro, I thought of using the asw20 model. (For a motor plane there is a piper etc)

As it is geared towards testing planes it should be fairly easy to model a model airplane, there's a lot of documentation available.

(3D looks) (flight dynamics)

Unfortunately I do not have the necessary skills.

Thanks

Marc

I've attempted something very similar. My interface was based on the old FMS method in which the PWM was fed into one pin of the parallel port. I wrote a kernel device driver that would decode the timing from the para port and return it to user land via the kernel as if it were a joystick device. In this way, I could use a RC Tx with any Linux program that expected fully proportional joystick. (BTW, kernel 2.4.x). As an input device for simulation of full scale aircraft, it worked real nice.

I didn't find this to be a big issue. I used the "look at" method of setting the view point. As you anticipated: very similar to the tower view.

Now, this is where my project went off the rails and caused me to give up. My approach was to use the default full scale aircraft and "shrink it".

First problem was the power plant. Flight Gear doesn't just simulate an aircraft, it models it in software. The piston engine model wants parameters such as displacement and manifold vacuum. I could not fiddle these to make a very accurate representation of a glow engine. The best I could do was to limit the maximum power to about one horse.

The second problem was to reduce the physical properties of the aircraft. I started by reducing the mass to just a few kilos. The result was that even before setting on the tarmac, the simulation would jump all over the place and report a crash. It appears that FG sets the aircraft a few inches above the tarmac and allows the suspension to settle down. If you are observant, whenever you start FG, you may notice that the image bobs up and down a bit in a well dampened response. Reduce the mass to almost nothing and the simulated spring constant throws the CG all over the place. Solution was to gradually reduce weight, spring constants and damping until some sort of balance was achieved.

Next issue concerned aerodynamic forces. The combination of low power and low mass was still not enough to get off the ground. I attempted starting the simulation with the aircraft in the air, the result was nothing more than a glider. I couldn't figure out what parameters to tweak and in which order.

Other problems remained to be solved. The terrain surface textures look fake from ground level. The aircraft visual dimension need to be shrunk to match the reduction in simulated properties. In order to make the product fit for mass consumption, the modelled RC world should be allocated its own root directory so that the full scale folks don't complain about contamination of real world scenery and aircraft with scaled material.

Reluctantly I had to give up. Even though I put in dozens of hours learning to write kernel drivers and working through the FG source code, there was no end point in sight. However if an special interest group on this topic were established, I'd be very interested.

At one point I asked on the FG mailing list about models for scale models, and the response was that it was possible in theory, but would be a lot of work.

[...]

IIRC, there wasn't any way to simulate electrics at all...

I didn't get into it as far as you did, but I concluded that it would probably be faster and easier to write something from scratch. Despite what one would think at first, I don't think there is very much overlap between a full-scale general aviation cockpit-POV simulator and an RC model simulator.