I don't market T&S products, but I do write real-time
software for cable set top boxes, so I'll put in $.02
Canadian. (Geek talk ahead--delete now if you aren't
interested--don't say I didn't warn you! :)
I agree with Gary, there is no reason why a PC cannot
be made to record T&S with 1 ms accuracy. The key is
to set up a piece of software called an "interrupt
service routine" so that when a signal comes in from
the timing equipment, the PC stops what it's doing,
records the event with time & date, and goes back to
its business. DOS makes it easy to control hardware
directly; Windows doesn't because it insists on
getting its filthy paws into every ISR, and goes about
it pretty slowly. To do this in Windows, you have to
write a more complex code called a VxD ("virtual
device driver" acronym in Redmondese).
To be assured that this *always* happens on time, the
computer's operating system has to be "deterministic,"
i.e. every task it does must have a worst case
execution time. Some of the stuff Windows does is
unbounded in execution time (memory management being
among the worst offenders), so you write a VxD to do
time critical stuff because it runs at a higher
priority than normal applications. With an operating
system designed for real time jobs, the norm is
*microsecond* accuracy. PC's that do real time
control tasks like factory automation usually run an
OS specifically designed for real time operation. T&S
isn't that tough of a job...basically any computer
that can run a sound card can run a *properly
programmed* T&S system. It's just not as simple to
write as "hello world". (If you work out how, it's a
good way to pay for a racing habit!)
Now, the next person who says "can't be done" shall be
bound and gagged and forced to listen to me rambling
about rate monotonic analysis! >;-)
Craig Blome
--- GPSoftware <gpsoftware@icehouse.net> wrote:
> I can't help but to chime in one last time on this
> subject, since I market a
> software product for timing and scoring.
>
> I've heard a lot of talk about how a PC can't be
> used for accurate timing.
> I've devulged a lot of details about how Accucros
> goes about doing this,
> with theory and actual tests as proof. I can only
> say that those that think
> a PC cannot perform this task either do not know of
> what they speak, or they
> may have a hidden agenda to downplay the ability of
> a properly programmed PC
> to perform the task. A PC is made up of hardware
> that can be told what to do
> by software. Precise real-time control is done by
> PC's routinely. There is
> no magic involved. The techniques may not be known
> to the average PC user,
> but are not uncommon or radical to the embedded
> controller programmer. Could
> a PC be programmed for inaccurate timing? Of course!
> Could stand-alone
> timers be designed, assembled, or used improperly to
> produce inaccurate
> results? Of course! I think the stand-alone timers
> are great and are
> appropriate for many situations. However, clubs do
> have a choice of several
> different systems and depending on their situation
> but compromising the
> timing integrity doesn't need to be a factor.
>
> Someone pointed out that PC's, being general purpose
> machines, must not be
> able to perform the timing task as required for an
> Autocross. Certainly
> every machine has it's limits, and maybe if you need
> 1/100,000 resolution a
> PC won't do it. But we only need 1/1000, same as F1
> uses and much better
> that the 1/100 used by downhill skiers (this has
> boggled me for some time).
>
> Gary Poole
> GPSoftware
> P.O. Box 421
> Liberty Lake, WA 99019
> Phone/Fax (509)532-1702
> E-Mail: gpsoftware@icehouse.net
> Web: www.icehouse.net/gpsoftware
>
>
|