Quote :

2 - Imagine you can move your seat horizontally, along the longitudinal axis, because it is set on a base running over trails. These trails can be linear ball bushing, ok? So you have one base running forward and back, and this is moved by a linear actuator, which can be a pneumactic or electric cilinder, or a ratchet gear, or cables, whatever.

that's actually not a bad way of solving the issue of lateral(or is it linear, i forget) g forces, instead of using some sort of tilt mechanism.

edit: instead of the seat though, it'd probably be easier to move the whole thing in that fashion.

Quote from Speed Soro :

1 - My idea is way to expensive, so I'll just put it here, but I have no intention to develop it, at least not for while.

I'm David from Force Dynamics. First, this is a pretty cool thread, and it's gratifying to know that a bunch of people out there appreciate what we're doing!

Second, to Soro: Your idea isn't a bad one at all. After all, if you consider the best possible simulator, what would it do? It'd be able to move, laterally and longitudinally, exactly like a car. Benefit? Perfect simulation. Negative? Needs 4 square miles to operate and a 900hp servo system.

But you can do a pretty good job even with far less, just like you describe. But it's still very expensive - that's why we didn't do things that way. There's a system out there called the National Advanced Driving Simulator (yes... NADS... I know) that has a whole Stewart platform on a huge X/Y table. It needs its own building and cost $15m plus, if I recall.

So, what do you do in the absense of that? Try to get as close as possible. There's a whole range, from most expensive to least:

1: Real car
2: Simulator with same motion range as real car
3: Simulator with large motion range; tilts to do 'constant forces' as it runs out of XY travel
4: Simulator with very high center of rotation (above your head); your butt is actually moving laterally during rotations, so it's basically like 3
5: Simulator with reasonably high center of rotation (like the 301); not as good as 4 but probably half the cost or less for the same dynamic performance
6: Simulator with low center of rotation (under your butt). This is awful, since your entire body moves the OPPOSITE way it should for any given force input

The one thing nobody's mentioned here is yaw (rotation; spin). That, to us, is the single most important force you can have with a ground vehicle simulator of any kind. Very few, if any, companies out there have any yaw, and the ones that do have only small amounts (+/- 15 degrees maybe), which cripples its usefulness. A large amount of correctly-done yaw trumps almost anything else (aside from really bad errors like sub-chassis centers of rotation and inverted forces!).

HI David, really glad to see you here.

I've sent a email to your [email protected] sooner, before read you here.

There I gave these two suggestions.

David, I understand perfectly what you said, and believe, I'm not a begginer in machines development, so I have perfect idea that to achiece the ideal simulator, there is just one way: buy a car and go get a track!

But, I think I can explain better what I'm proposing here, that is just a way to get more immersion, not perfect, but better, and less "mechanical bull" than Stewart plataforms way.

As I said before, I see St. good for flight simulations, but for cars, with realy fast accelerations (longitudinal, lateral, vertical), it does not look the best approach IMHO.

So, my expensive idea is a short trail that permits this movement along with the inclination. The "kick" would be made linear, no more rotational, but the rotation would still be there, to assure the pressure of the inertial forces after the short trail course finishing.

I know that is far way to be perfect, but it looks to be hugely better than just rotation.

I'd like to see a weathy company like yours (how is going within this crisis??? hope the best), leading this idea towards a better sim.

By my side, poor side , I'll try a simpler system, just thinking in small inclinations, and change of the distance between controls and seat.

Wait you back here soon. bye

ps: sorry my bad english. Some words maybe are wrong, others wrong placed, others I think I just imagine. Sometimes I have time to correct it with google translator, but not always. So execuse many mistakes.

I have the weird feeling you do not read at all the answers. I am not dreaming.... you are trying to explain FD how they should do their sim?

Quote from Juls :

I have the weird feeling you do not read at all the answers. I am not dreaming.... you are trying to explain FD how they should do their sim?

???

Quote from Speed Soro :

the transitions of a plane is always slower than a car

Quote from Speed Soro :

What I want here is to make a brain storm with people who know what a simulator is,

I think this would be cool. Driver could be housed in a sphere for total immersion. Add some shocks for vertical movement/drops (bumps)


Now all we have to do is figure out how to make it for under 2k

sure it will cost much more!

It looks to be consensus that the centre of spin must be near the chest to the neck, for minimize the perception of rolling and gives some balance between the displacement of your head and your legs.

I'm thinking in a system with a two circular rails crossed under the seat, each one with its own guides and a sector of gear with the same centre of the circular rails.

A little spur gear, actioned by a AC motor, rotates the big sector and controls the inclination.

As soon as we have a drawing I'll put it here for you figure the idea. I think this is not a original idea, but I can't find a video or photo with it in the Internet.

Quote from NightShift :

Could you please explain what did you mean? I really didn't understand your intentions

That bit of yours "The transitions of a plane are always slower than a car" doesn't quite sound right to me.

I think that in a plane, it spends more time to change the direction of the forces of inertia, than a car. Because that I think is easier to built a motion cockpit for Flight Simulator than LFS, even why there is no bump on a plane (in fact there is, but you can dismiss, while in a car it is present all the time).
Even in the fighters, in maneuvers of avoidance, the time to change the direction of a plane, are much larger than the time a car spends to make a closed curve, I think.
What I call a transition, is the slope of the rising curve of acceleration, either lateral or longitudinal.
For example, imagine at the end of a straight, with zero lateral and longitudinal accelerations.
You brake suddenly, and longitudinal acceleration rises almost instantaneously to its maximum value, generating a maximum force of inertia, and then begins to decrease slowly, until the turning point of the curve, then when you release the brake.
At that point you start to curve. The lateral acceleration rises from zero to its maximum value virtually instantaneous. If you maintain a constant speed throughout the trajectory of the curve, the centripetal acceleration will be constant until, at the end of the curve, slowly declined to zero, with the return of the steering wheel to the center.
The point where the accelerations range from zero to its maximum value is where I'm focusing on.
It is to meet the transition between 0 and maximum that I proposed a linear system. I believe that this linear system can be short and works well, because the ramp would be given to increase the acceleration curve, not the complete curve.
What David said is a system composed of a platform Stweart and a mobile base, and the platform would be responsible only for the inclination of the vehicle and bump. But this is not my idea. Again, the linear system would be responsible only for the ramp, just the part that gives the kick.
In parallel with this linear motion, the platform should be inclined, as with any FD does today, adding the inertia effects. Upon completion of the course line, the inclination to keep "weight", thus giving the impression of being in the middle of the curve.
Even being in the middle of the curve, while the inclination is maintained, the linear system must be slowly returning to the center, so that when the ramp goes back to be flat, the linear base is again in its central position.
The superposition of effects, of course will ask a fairly complex algorithm.
This is the kind of work for a company with enough know-how, as the Force Dynamics, not me.
I have the concept, but not the mathmatical knowlegde and the money to the develop something like this.

You guys should read what Soro's saying a bit more carefully before you dismiss him - he's obviously thought this through and makes a lot of good points.

Soro, you misunderstood my post - I meant exactly what you say, actually: Transitioning between 'real' lateral/longitudinal acceleration and 'faked' (via rotation) lateral/longitudinal acceleration.

The reason this is a Good Idea is that - again, Soro, good job - vehicles have very short transition times. A car can go from accelerating at 1g to decelerating at 3gs in basically zero time - but a motion platform that relies on rotational position to simulate that feel takes quite a long time (relatively - for us, maximum transition time is about 1 second, for others usually longer). During that period of transition, you get incorrect forces. Center of rotation can reduce that negative perceptually, by moving your butt (or, if the CoR is very high, your whole body) the right way during that transition. Basically, a high center of rotation starts to 'automatically' accomplish what Soro suggests.

But you need a very high CoR indeed to get good enough for a truly correct feel during bad transients.

It should be pointed out that, even with problems during transients, a driving simulator can be an excellent training or simulation tool. If you're driving well, you're already driving with a minimum of load transients (except in certain situations like jumping HARD on the brakes in a formula car). It's if you start to get into a tank-slapper that the transients get really bad. So if you're driving reasonably well, you don't 'bump up against the end-stops' of the platform's capability that much; most of the times you really want to feel what the car is doing (and you're not about to fling it into the scenery) aren't transient situations. Of course, the faster and lighter your car, the faster the transitions are. Motion simulators are bad at fast formula cars and downright awful at go-karts, unless you turn down the maximum *level* of force to increase the speed of transition between different loads.

I hope that was somewhat clear.

Quote from Speed Soro :

Even in the fighters, in maneuvers of avoidance, the time to change the direction of a plane, are much larger than the time a car spends to make a closed curve, I think.

So if I understand this well, you're saying the rate of change of acceleration is higher on a race car than a fighter plane?

I can't see a reason why it should be that way. As long as the car is driven on a track (and not crashed against a concrete wall), the RoC of acceleration should be limited by the total stiffness (tyres + suspension + chassis) of the car.

Every car has some kind of damping setup, and I see no reason to state the flexibility of the materials and the plane structure should be lower than the figure you'd get from even a purebred racing car.

AFAICT planes pilots are bound to wear special equipment and engage in specific training to minimize the chance of them fainting. Even in F1 while pilots are trained to be able and withstand the physical effort, there's no special equipment involved except those needed for safety.

So IMO simulating fighter planes(*) may be as hard and likely harder, on the motion simulator, than cars.

(*) and that assuming air pockets aren't able to generate intense acceleration pulses on passenger planes too.

Quote from Speed Soro :

Again, the linear system would be responsible only for the ramp, just the part that gives the kick.
In parallel with this linear motion, the platform should be inclined, as with any FD does today, adding the inertia effects. Upon completion of the course line, the inclination to keep "weight", thus giving the impression of being in the middle of the curve.
Even being in the middle of the curve, while the inclination is maintained, the linear system must be slowly returning to the center, so that when the ramp goes back to be flat, the linear base is again in its central position.
The superposition of effects, of course will ask a fairly complex algorithm.
This is the kind of work for a company with enough know-how, as the Force Dynamics, not me.
I have the concept, but not the mathmatical knowlegde and the money to the develop something like this.

Do not forget that stewart platforms do translation. This does not appear on the wikipedia page, but they can translate. And if you recline the actuators of the stewart platform, then you increase it's translation range and decrease the rotation range...

With a large enough stewart platform you can do everything...translation for the ramp, then tilt while the platform translates back to the center. Expensive simulators using stewart platforms do exactly that. They translate to kick real acceleration, then they tilt to use gravity while they translate back. This concept mixing tilt and translation has been used since the 70's. And the algorithms to mix the translation with tilt are called "washout algorithms", you can find documents if you google it.

Force Dynamics and others use other concepts (almost rotation only) because it is less expensive to build.

Fascinating stuff here..

My understanding is that very high or low center of rotation demands a lot of torque from the motors. Wouldn't it be best to have CoR at the middle of the driver? What are the downsides to this approach?

Quote from NightShift :

So if I understand this well, you're saying the rate of change of acceleration is higher on a race car than a fighter plane?

Yes.

Quote :

I can't see a reason why it should be that way. As long as the car is driven on a track (and not crashed against a concrete wall), the RoC of acceleration should be limited by the total stiffness (tyres + suspension + chassis) of the car.

Every car has some kind of damping setup, and I see no reason to state the flexibility of the materials and the plane structure should be lower than the figure you'd get from even a purebred racing car.

AFAICT planes pilots are bound to wear special equipment and engage in specific training to minimize the chance of them fainting. Even in F1 while pilots are trained to be able and withstand the physical effort, there's no special equipment involved except those needed for safety.

You're confusing maximum accelerations with *jerk* - the rate of change of acceleration. Even a fighter plane can't go (effectively) instantly from +1 to -5gs like an F1 car can. Or at least could.

Think about F.Alonso warming up his tires - that's hard enough with a 1000lb vehicle stuck to the ground with soft tires and downforce; it'd be flat-out impossible with a 40,000+ vehicle sliding around through the air. A fighter plane can probably generate some fast transients, in specific scenarios, but not *faster* than a race car, which can generate *instant* transients quite easily, and does so on a fairly regular basis.

Quote :

So IMO simulating fighter planes(*) may be as hard and likely harder, on the motion simulator, than cars.

If you want to get full-scale accelerations, it obviously is. But in general, flying is a pretty nebulous thing (I've done it, though not in a jet!). The feel you have isn't anything like the feel in a race car, and you don't USE the feel in the same way.

They're two very different disciplines, but my opinion is that it's a lot easier to screw up motion while simulating a race car than a plane.