----- Original Message -----
From: "David Megginson" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Friday, March 28, 2003 8:22 AM
Subject: Re: [Jsbsim-devel] MSFS Aircrafts
> Ron Freimuth writes:
>
> > One thing MS developers have is access to a lot of real pilots.
> > Who can at least give some idea if an AC 'feels right'. All the
> > way to 747's.
>
> That's harder than you might think: how a plane feels to the pilot
> depends as much on control loading as on responsiveness ...
>
> A pilot familiar with that plane is almost certainly going to find it
> very unstable in the pitch axis, and complain that the nose bounces up
> and down too much. In the real plane, the dynamic pressure from the
> relative wind tends to hold the control surfaces in one spot, and it
> takes a bit of effort to move them from where they want to be (a *lot*
> of effort for a big deflection). A home-computer joystick or yoke
> might have a little spring in it, but in general, it's going to be far
> too easy for the computer user to create an elevator deflection, and
> the plane's going to feel unstable.
I flew a Level C Falcon 20 a couple of years ago. It used variable
rate springs on the yoke to give an indication of aerodynamic loading.
Same as the real FA-20.
Other than the fact that the real Yoke took more force than my MS
Sidewinder JS, I didn't find it hard to adapt to. Nor did I feel my
non-FFB Sidewinder was far different. I'm used to the small,
non-varying spring forces, use a light touch and have a horizon
indicator turned on so I can get attitiude dynamics visually and by
force. My main problem is the backlash in my overused Sidewinder.
However, trying to control the FA-20 on the TO roll with the
steering knob, rather than the rudder pedals, was not easy. ;)
OTOH, WWII AC had manually linked control surfaces and required all
the pilot's strength to fully maneuver. Some AC required pitch trim to
be set closely; the pilot couldn't get an AC off the ground otherwise.
No real pilots think FFB is realistic. I bought a 'Wingman FFB' and
it gives the illusion of way too much sensitivity. Even with a lot of
fake 'spring centering' set the forces were weak. Moving the base to 45
degrees resulted in the stick moving and the AC pitching or rolling
also. The forces were too weak for the first 50% deflection, then
increased more rapidly and I could feel cogging. Not at all smooth.
Jerry Beckwidth's "1% SS" sets the control moment vs q tables to
emulate the limits of the pilot's strength. In combination with Cl_p,
etc. this generates CFS AC that roll and turn at the published rates.
> There are two solutions to this problem (other than building a full,
> force-feedback console). One is to exaggerate the coefficients to
> make the nose more stable than it should be, but that sort-of sucks.
> The better solution -- which I stole from Andy Ross -- is to square
> the joystick axes (preserving sign), so that small joystick movements
> are less sensitive than large ones:
PS1, a good 747 simulator mentions 'exponential' response to JS
movements.
Fly! allowed one to change the exponential effect. Possibly it is
misnamed, x^n involves an exponent, perhaps it was 'n' that could be
varied. MSFS2K appears to have changed to some intrinsic non-linear
mapping compared to FS98.
FS/CFS AIR files have 'gearing' tables that let one shape the
control moment sensitivity as desired. I tend to set them at 0.80 at
neutral elevator deflection.
The problem with this is that the control surfaces drop in
moment/force as they increase in deflection (similar to flaps). This is
in the opposite direction. I've set the rudder 'gearing' something like
an inverted "W"; this reduces sensitiivty in the middle and at the
higher deflections. In the end, a flat 'gearing' curve may be a good
compromise in MSFS. Especially since FS2K+ appears to have some
'exponential' effect added implicitly.
Fly!, and MS FS/CFS allow one to change 'null zone' and
'sensitivity' for the JS in the menu. Lower sensitivity adds more low
pass filtering to the JS command, not less ultimate movement. This
mainly affects AC with fast control actuation, such as aerobatic AC. A
nominal setting adds a time constant of perhaps 0.1 second. Which is
about what one might expect from hydraulic actuators on AC that use
them.
> -1.0 => -1.00
> -0.5 => -0.25
> 0.0 => 0.00
> 0.5 => 0.25
> 1.0 => 1.00
>
> Now, the pilot will have to move the joystick 50% of the way just to
> get a 25% elevator deflection, and a 10% joystick deflection will
> result in only a 1% elevator deflection, so the plane will seem more
> stable. This is still far from perfect, but it's better.
I do rapid full forward, full back JS deflections and check that the
G level changes about the same amount + and -. Say 1.0G (level) to
+3.5G full back, and -1.0G full forward. Considering the different up
and down elevator limits the G force should change less going negative
than positive.
This assumes the wing downwash doesn't hit the tail and AoA hasn't
changed a lot. So, IAS has to be high and movements rapid and equal to
get useful information.
Virtually all FS/CFS AIR files drop control deflection as q
increases. This keeps JS sensitivity from getting too high as IAS
increases. Relative Aileron moment in jets typically drops to 0.50 at
q = 300 in the "Aileron moment vs q" table. To 80% for the elevators.
The rudder stays effective at high speeds, perhaps in part because a
pilot can apply over 150 lbs of pedal pressure. One can even make the
controls reverse at high q if he desires. Some effects might better be
done in the Mach related tables. It depends on whether one is modeling
non-linear effects due to mechanical flexing at high q or Mach related
aerodynamic effects.
Heavy Yokes are used by hard core flight simmers. They can cost
over $500. At times home built cockpits are equipped with real yokes
and throttles (salvaged from old AC). Just the same, the Airbus series
use a joy stick in the cockpit. But, with complex automatic control
algorithms to make the jets handle nicely, not as they would with only
force augmentation. Even more important for supersonic fighters,
which may not even be stable without stability augmentation. Feasible
to model in detail JSBSim, but not in MSFS. ;( X-Plane lets one set a
'stability augmentation autopilot' but one has no idea what it is doing.
> > Further, they are good at scanning Flight Manuals that might be
> > hard to obtain otherwise. ;) We also have airline employees with
> > access for photographing cockpits etc. .....
> We're in a different situation. I work from time to time as a
> consultant to a major aircraft manufacturer, helping to design and
> implement the system that produces their manuals (AMM, SRM, FIM/FRM,
> ............
> aerospace companies as well. Obviously, we're not in any position to
> play with data that even gives the *appearance* of having been
> obtained illegally: our information has to be above suspicion.
The Flight Manuals aren't proprietary. However, it's a lot of work
to reverse engineer gauge readings and fuel flows to engineering drags,
SFC, etc. One guy paid Boeing over $1000 for performance data to aid in
modeling a Fly! AC. I thought Boeing only sold such AC Performance
Manuals to owners of the AC.
I'm not sure how the professional Full Motion sim companies get
their data. I suspect a lot is by cut and try. I heard it can take a
year to tweak such a simulator to required specs and 'pilot feeling'.
One MSFS guy spent a lot of time taking analog data from a FM 727
simulator he had access to. That gave some useful data one can't see on
the panel. Such as thrust.
> > I'll point more of the engineers fooling with MSFS to
> > sci.aeronautics.
> I think I might start reading sci.aeronautics. Thanks.
> All the best,
> David
I saw a recent message in sci.aeronautics.simulation. However, it
is virtually dead. Jon B. has posted in sci.aerodynamics and it might
be a good place to share ideas and links.
Ron
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
