Re: [rtl] Power System's Control under RTL

[Michael Barabanov <baraban@fsmlabs.com>]

Yes, RTLinux should be up to the task.
Using a faster machine will help of course.

Some jitter is unavoidable because of device interrupts and caching.
One of the most radical ways to reduce it to sub-microsecond level
is to schedule your tasks to run slightly in advance and then busy-waiting
(with interrupts disabled) until the exact time.
Please see examples/measurements/rt_process ("advance" parameter
handling) for an example.

Michael.

Ramon Costa i Castell? (costa@ioc.upc.es) wrote:
> Hi,
> 
> In our research group we are working in designing
> control algorithms for Power System.
> By now we've using DSP based systems, we are interested
> in moving to a PC based system, so we decided to study
> if RTLinux is a good option for as.
> 
> Our task are periodic with T=20us-50us depending on the
> specific system. We have developed a very simple periodic task
> in order to obtain a square wave over the parallel port.
> In this experiment we notice that there's a jitter which
> varies for 3-5us. This would reduce over algorithm performance.
> 
> 
> I have the following questions :
> 
> -Do you think RTLinux is fast enough for this kind of task ?
>  Does anyone have experience in using RTLinux with this kind of systems
> ?
> 
> 
> -Could we reduce the jitter through the variation of the scheduler timer
> tick size ?
>  (Which is the jitter source ?)
> 
> - Using a faster machine would reduce the jitter ?
> 
> -If we put our control algorithm in the ISR of a periodic interruption,
> which kind
> of jitter would we have ?
> 
> 
> 
> Best Regards,
> 
> R. Costa-Castello
> 
> 
> ----- End of forwarded message from owner-rtl@fsmlabs.com -----
> -- [rtl] ---
> To unsubscribe:
> echo "unsubscribe rtl" | mail majordomo@rtlinux.org OR
> echo "unsubscribe rtl <Your_email>" | mail majordomo@rtlinux.org
> --
> For more information on Real-Time Linux see:
> http://www.rtlinux.org/