rec.autos.simulators

I'm no scientist but I understand that your eye-hand reaction time is
typically 0.3 seconds or so. If you don't use FF this is as good as your
reactions are likely to get, and you have to 'predict' the rest.

However, your hand touch reaction time is more like 0.1 to 0.05 seconds, so
you instinctively react to feel more quickly than seeing something, thinking
about it, then acting on it. This makes sense to me.

To test this, get someone to hold a ruler vertically at arms length. Put
your hand around the bottom of the ruler, ready to grip it but not touching
it.
When they let go of the ruler, you catch it, and measure the distance it
travelled. You can work out your eye-hand reaction time from this since the
ruler will fall at 32ft/sec/sec (9.8 metres/sec/sec).

To test your touch reaction time, close your eyes, do the same test, but
this time with one of your fingers just touching the ruler so you can feel
when it starts to move. You should find that you will catch the ruler much
sooner if you can't see it.

Of course if you try this test by yourself (you both drop the ruler & catch
it) you will know when it will fall. That's your prediction time, and it's
essentially zero of course.

There's probably loads of more scientific tests for this, that's just the
only one I can think of right now.

Malc.

The controllers I use are 3-4 year old technology, and so I assume if I
bought a new model, I would need to reduce my latency setting to compensate.

Mine is currently 0.065 seconds, but judging from other posts here, the new
momo might only need a setting of 0.02 seconds for example. That's not _my
reaction time, it's the _controller's reaction time.

Malc.

  with latency THAT high my Momo will spike/jerk hard at every odd texture
border.  Driving like that is muy hard.
dave henrie

As I understand it GPL normally sends only a force and a constant damper
effect to the wheel.  The constant damper effect is like a preset
"heaviness" of the wheel and is not related in any way to the damping
coefficient that you set in core.ini.  This constant damper effect is what
Marc is turning off by setting damping to zero in the control panel.  The
damping coefficient in core.ini OTOH is used in the GPL physics model and is
applied to the force that is sent to the wheel.  GPL apparently calculates
the aligning torque on the wheel due to lateral force, then algebraically
adds a force based on the damping coefficient, then divides and if necessary
clamps the value by the max_steering_torque value.

If you keep the damping coefficient low you are mostly feeling the force
generated by the aligning torque - the "grip" of the tires.  If you raise
the damping coefficient very high the damping force becomes much larger than
the aligning torque, but the total still gets scaled and clamped to the same
maximum value.  So as a result the aligning torque becomes barely noticeable
and you are mostly feeling the wheel's resistance to turning, which is a
function of friction and weight transfer.  The two effects are similar and
both depend on grip and weight, but damping always opposes the *motion* of
the wheel, whereas aligning torque will normally be a function of the
wheel's *position* relative to the current direction of motion.

Because of the difference in the nature of these two effects they seem to
have both different dependencies on latency and different sensitivities to
latency correction.  For the damping it seems to hardly be noticeable if you
use too small a latency correction, but you get huge errors if you use too
much.  The aligning torque OTOH feels very wrong if you don't use enough
latency correction, but does not seem to suffer from over prediction.

So which is right?  Damping gives you feedback on the weight transfer effect
due to acceleration and deceleration.  Aligning torque gives you feedback on
the slip angle and grip level of the tires.  Ideally you would like to be
able to mix both together in realistic proportions.  Personally I think
aligning torque tells you more about what the vehicle is doing, but some
might feel that high damping is more realistic.  If you calibrate your
latency correction using the "weaving down the straight" test and you get a
very small number, say around 50 msec or less, you can use fairly strong
damping and suffer no ill effects.  But if you get a large number, say over
100 msec, you will probably have to choose to either use high damping and no
latency correction or turn the damping coefficient way down and use the
correct latency correction for the aligning torque.  Marc's damping setting
just seems off the scale to me though and at that level I think that any
amount of latency correction will create problems.

With two different wheels, a Logitech Formula Force GP and a Logitech Momo
Racing, I had to use a latency correction of 120 msec under Windows 98.
With that setting almost any amount of damping felt wrong.  For that
situation I set the damping to 1 with the FFGP and I think I shut it off
completely when I went to the Momo.  Since getting XP Pro I have been able
to turn the latency correction down to 35 msec.  Everything else is the
same, all I did was change OS, but apparently this made enough of a
difference in the system to improve the latency dramatically.  With this
lower latency correction I have found that a damping coefficient of 30 feels
about right, with both wheels.  So the two numbers are definitely related
and that, I believe, is why Marc's settings feel great to him and our
totally different settings feel fine to us.

Marc

So tell us your impressions...

Marc