On Mar 5, 9:27

And now, for something entirely different!

What you don't know about, I have to measure for a living. You fly
code that is barely close to a PCATD. Thats a joke approval wise. You
answer about 10 questions in a 5-10 page document, and you promise to
get reapproval if you make hardware or flight model changes to the
unit. A busy guy from the local FSDO comes out and test flys it for
about 30 minutes to sign you off. Its a formality at best. There are
small minutae in the rules for the quality of the visuals, and how the
flight model has to handle, but no specs on how tight the timing loops
in the software must match the real aircraft.

IN MS, you are always about 150-200 mS , ie one tenth to two tenths
of a second or more behind the flight model on the visuals, and your
flight model is updating every 16.76 ms, and that is tied to the
monitor vertical refresh rate, which is one of the few accurate
timing sources on a PC platform under windows. In fact window's
timing varies all over the place. As a result, they have to do a lot
of fudges in the code to smooth things out. From the 1978 Skyhawk
manual, 65 KIAS is the recommended speed for just about everything.
One knot is 1.68 feet per second. So at 65 knots your doing 109
feet per second. A 200 mS error at 109 feet per second is 22 feet
on the runway distance or roughly 5 feet on a 3 degree glide slope.
So your +/- 10 feet visually off or below the runway in software at
any given time right as your wheels hit.

Lets recalculate that for a F18. FAS.org claims 134 knots
approach, your moving 226 feet per second, and on a 3' glide slope
thats 11 feet. Since your always at least 200 mS behind, you have a
potential error of +/- 22 feet at that airspeed and that update rate.
Moral of the story, FS is helping you land with a software fudge,
because the granularity of your visual corrections is very rough as
you get close to the ground. In a real aircraft, there are cables and
servosystems, they react nearly instantly, but there is no software
fudge on a manual landing. Yet in MSFS, even a 10 year old kid can
hit the edge of the stripes on the runway time after time, because
the visual update rates must be slowed close to the ground. The math
says it has to be, or you would miss the deck by some multiple of 22
feet. IN a real aircraft, there is no 22 foot vertical fudge. Plus
that doesn't include the average human's .1 second reaction rate.

How do we measure that 200 mS delay?, there are a couple of ways.
You can command a max roll on the stick and measure the time to the
aircrafts maximum roll rate analysing the video on a tape frame by
frame. Or you can have a switch on the stick, a special pixel
programmed to flip from black to white, and a digital timer with a
photosensor. I know for a fact the delay on FS04 is roughly 150 mS
minimum from max command input to a visual change. That is a measured
value. FS10 will be no faster, its a hardware thing.

So if we put you in a real aircraft, you've been trained to
compensate for a time lag on all your reactions. That is a Bad thing.

Next up, lets talk about your feel for the controls.

You invariably have a 8 bit control resolution on your joystick,
ie your stick position is one possible position in 256 available.
So lets say your joystick swings +/- 20 degrees, that means you need
to move your stick 7 degrees before the software even notices it. You
move a real yoke 7 degrees and your gonna really be rolling or
pitching. So your already trained to overcontrol if we plop you in
that 747 cockpit. Oh and it gets worse. To smooth things out, you'll
find that in slow flight MS and other simulators condition you to
jerk the controls quite a bit to get motion started in the above
mentioned landing mode. Its control law smoothing, and its done to
make you feel comfortable as you approach the ground, otherwise with
that 200 mS lag, you'd come darn close grinding a wing tip into the
ground from the roll error every time. The real aircraft, other then
control system friction and cable stretching, has unlimited
resolution, and the flight response shows it.

MS does a lot of fudging to make your flight smooth and landings easy.
The real thing doesn't. Heck, with those errors how do you think
people can catch the wire in carrier landings on consumer sim
software, time after time? The answer is the flight model code is
doing it for you. Especially when landing.

Still think you'd make it if we stuck you a in a real plane alone?
Odds are you'd lead or lag in time so much on the landing you'd
break the gear or cartwheel. Your conditioned to do it by the code and
the system timing and control resolution.

I'm curious, is your sim set up so one PC runs the flight model and
controls and the other does the visuals? Or do you just have one
machine and one graphics card? . Because we need one dedicated flight
model machine just to get rid of the lag. And we use 10 bit controls,
and I'm being asked for more resolution.

Oh, dont get me started on flight models. Look at a real Bell 206.
It does have horizontal stabilizers, does it not? If you look closely
the airfoil is inverted on them. Well, most flight models put them in
as flat planes. This causes the aircraft to oscillate in pitch as the
airspeed increases. It took a long while to figure out that the guy
who did the flight model was wrong on that.

In one very popular sim, the rotorcraft model is a cessna 172 body
with a rotor added and the wing removed, and the tail rotor and main
rotors are defined as a thrusters. Because their is no rotor disk in
the form of a wing , Ground effect does not show up on approach.

Your basic sim software is very unlike the real thing.

Steve