Does anyone know what setting I can adjust to make an aircraft stable using 16x acceleration?

I don't think there is one.

 

Running at 16x means that FS has to do 16 times the amount of processing in the same time period. If there isn't sufficient "headroom" on your PC then something will have to give. The more complex the addon the more likely this effect. Even in default FSX, the autopilot doesn't work at more than 16x

Running at 16x means that FS has to do 16 times the amount of processing in the same time period.

 

 

Actually not. The entire code is executed at standard rate, but on each update cycle those internal variables related to time (ie ac position,ground speed, etc) are multiplied by the sim rate factor (2 in 2X, ... 16 in 16X, etc). User variables (Lvars,Cvars,GVars) are actualized at standard rate always (like the code) so this is something one have to take into account when writing gauge code, for example.

 

Tom

A faster computer can handle higher factors and remain stable because the instability arises from the per frame update "delta time" becoming too big for the flight model to step correctly. Suppose some flight model equation at some state breaks down if dt is much greater than 1/2 a second. A slow computer running the game at 10 fps (dt = 1/10) is accelerated to 8x, the dt per frame increases to 4/5 sec which is greater than 1/2 so the flight model goes unstable.

 

A faster computer running at 40 fps could handle 16x because the dt would increase only to 16/40 or 2/5 second, still under 1/2 second.

 

These numbers are just examples, I don't know where any particular flight model or simulation would break down. This does mean though, that you can increase stability at 16x by getting a faster computer. (not the answer you wanted I know)

 

Cheers

This does mean though, that you can increase stability at 16x by getting a faster computer. (not the answer you wanted I know)

 

Cheers

 

A faster computer would make no difference at all. Frame rates are not affected by current SIM RATE (0.5-64X); they are a result of a combination of processor-Video hardware rendering efficiency. With sim rate set too high (8X-up) problems arise on all those systems that use PID calculations -autopilot is a typical- because the Deviation Error on each cycle becomes too big to be properly handled by its algorithm. As a result, you get an increasing overcorrection that turns the entire system unusable.

 

Tom

Apologies if I'm hijacking your thread Dillion.

 

Tom, I don't understand what a "PID calculation" is. But I didn't say SIM RATE affected viewer frame rates. I meant it affected *dt per frame* as seen by the simulation. The frames per second stay the same, but inside the algorithms, the simulation dt per frame is higher. In all simulations the size of dt in a simulation tick is very much related to algorithm stability, is it not? That is what yields large Deviation Error (your term).

Clark,

 

PID comes from Proportional-Integral-Derivative. You can read about this here:

 

http://en.wikipedia.org/wiki/PID_controller

 

And as I stated before, SIM RATE and FRAME RATES have no relationship at all. What affects a contoller's algorithm is a dt per code cycle, and again this has nohing to do with frame rates. At higher SIMRATE, dt per code cycle is higher, yes, and that's what is related to algorithm stability.

 

Tom

Thanks

 

Yes, frame rate has nothing to do with with it. It is all about dt per code cycle and a faster computer has a smaller dt per code cycle base rate, thus smaller dt at any compression multiplier as well. Thus a faster computer has faster SIM RATE, a smaller dt, and a more stable algorithm output. A faster computer may very well run 16x where a slower computer cant. Make sense?

internal variables related to time (ie ac position,ground speed, etc) are multiplied by the sim rate factor (2 in 2X, ... 16 in 16X, etc).

 

Does that mean that FS multiplies the accelerations by 2 4...also multiply etc and then determines speeds and positions for each time step - surely it can't multiply speeds and positions as well? FS uses a simple numerical integration algorithm according to Zykowski's paper. This itself could also beome unstable at high multiplication factors.

 

Can you clarify the relationship between the frequency of the fundamental simulation loop which varies according to Zykowski says and the fixed 18Hz gauge rate - I've never really understood it?

Make sense?

 

 

Uh...no :(

 

You're still mixing software logic with hardware logic.

FS panel programs run at about 18 hz on any computer, from the slowest to the fastest, what means one cycle last the same on both machines. This is software logic, and Sim Rate is a flag within this logic. A faster computer will render fast (higher Frame Rates) but will not make gauge calculations faster than a slower machine. All aircraft systems are based on gauge calculations. Get the point?

 

 

Tom

FS panel programs run at about 18 hz on any computer

 

Did not know that. Now it makes sense why a faster computer can't help if there is an internal fixed speed. I would not have guessed this because all software simulations I have worked on and programmed in my twenty years (government and game industry) never fixed dt. They scaled with dt and warned if the dt was too low for a given required stability. They never tossed out extra cpu if it was available. Thanks Tom.

As an aside, sometimes increasing the various stability factors in the "flight_tuning" section of the aircraft.cfg will dampen out the wild gyrations experienced at higher sim multipliers, but it also can make the airplane sluggish when flown in real time. I had a little bit of success finding a compromise in FS9 with, IIRC, the PSS B777 to make it behave at high (8-16X) sim rates.

 

Regards