Carsten Fischer schrieb:
> when I start a programm in the user space e.g. compiler , ls
> , rm ..... the length grows up to 3
> mircoseconds. That is not hard realtime!
3 microseconds is way better i would have expected it for a
486 class processor! Also, with your example code in your recent
mail, you are not measuring the time for the calculation,
but mostly how long it takes to drive some data from
the cpu via the hostbridge to the super-i/o chip :-)
It would make me wonder, if you are getting much better
results with QNX, VxWorks or any other operating system
that claims to be hard real time capable (on a 486).
In this context: How do yo define "hard real time"?
> Also when I start the servo-module from B. Kuhn the servo
> moves (there are jitters) when the harddisk
> is active or when I write something on the keyboard,etc.
The jitter does not necessaryly derive from bad timing:
The position of the servo is not only dependent to the
length of the high impulse, but also dependend on the
voltage level of the signal:
In idle mode, the power consumption of the computer
is low and the 5 Volt line has, lets say, 5.03 Volt.
When moving the mouse, an interrupt occurs and the CPU
wakes up, processing the event. This will certainly
lead to a much higher power consumption as in Idle mode.
If you have a weak power supply or a badly design PCB,
then the 5 Volt line may drop down to, lets say, 4.74 Volts.
As the voltage level (in high state) of the output pins of
the parallel port are directly affected by the power supply,
the servo connected to the parallelport is also affected.
You can reduce/eleminate this effect with the following tricks
a) Put the CPU on stead high load, i.e. by an endless loop like
main() {for(;;);};
This is certainly not a good solutions, but helps determing
the problem.
b) Buffer the output signal via TTL-chips (i.e. 74LS245)
and a seperate power supply.
But some PC hardware is so poorly designed that nothing helps:
I recently had a Gericom Supersonic 1000 (with 1GHz PIII and
256 MB PC133 memory) that produced extremly bad jitters
with both RTLinux-3.0 and RTAI-24.1.3 (BTW: i returned that
crapy notebook). As positive example, two years ago, i had a
Compaq Armada 1750 (with PII/333) where everything was running
like a charm RTLinux-2.2a - no jitters were recognisable.
hop that helps
Bernhard
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
--
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/
