Elliot,

 

Good thinking based on the sampling rate your trying to achieve.  Our control and measurement modules have a processor and a custom ASIC to handle high speed measurements and communications (CSI CR1000).  High speed measurements are used in a variety of applications (Boyen Ratio systems, Flux Networks and Thermocouple applications).  Even with our design there are issues due to using an RTOS with a single processor with respect to IP and Serial communication.  Measurements always have the highest priority but communications can fail from time to time based on the sample rate used.  The other challenge you might face with higher sampling rates is noise (60/50hz can be painful).

 

I want to do something similar and have to develop a low cost control/measurement board with its own processor.  This would allow for high speed measurements and would also buffer the data and pipe data to the gumstix for further processing based on the next available time slice.  The gumstix will act as the communication platform between other devices, algorithms etc.

 

Didn’t there were that many people doing measurements the gumstix, cool!

 

Eric

 

 


From: gumstix-users-bounces@lists.sourceforge.net [mailto:gumstix-users-bounces@lists.sourceforge.net] On Behalf Of EXT-McGhehey, Mark J
Sent: February 15, 2007 12:08 PM
To: General mailing list for gumstix users.
Subject: *TELUS Detected Spam*Re: [Gumstix-users] Critical Real Time on the gumstix

 

Elliot,

  I find that to be a very intriguing concept and I have also begun to persue such an implementation.  I am working on a system that requires sensor sample rates as high as 5000Hz and so far I have not been able to achieve this with only a single 400xm BT Gumstix.

 

  I would be very interested in hearing more from you about this and hopefully maybe I can contribute something also.

 

 Thanks.

 

Mark 

 


From: G Elliott [mailto:gelliott@gmail.com]
Sent: Thursday, February 15, 2007 2:36 AM
To: General mailing list for gumstix users.
Subject: Re: [Gumstix-users] Critical Real Time on the gumstix

can i offer a different way of going about this?  i had a very similar problem, though i'm only reading at 140hz (but very time critical).  

 

my solution was to offload that processing to a pic chip that communicates via serial.  i also offloaded some algorithms onto the pic so that much of the heavy lifting was done in outside of the gumstix, then the pic would send relevant data to the gumstix as needed.  its cheap, robust, and fast.

 

if you're interested in pursuing this type of solution, i'd be happy to let you know how i worked it out.

 

greg.elliott

arts.computation.engineering

university.of.california.irvine

 



 

On Feb 13, 2007, at 6:33 AM, Brendan Maguire wrote:



Hey,

Still looking for help with the below if anyone has any info?

>Does anyone have any simple example C code for using the one of the timer
>interrupts? I've been looking around for ages and I'm not getting anywhere
>with this! I'd be very grateful if someone could just post up the code that
>would enable the interrupt to execute a small isr function.

I also tried changing the HZ value in /gumstix-buildroot/build_arm_nofpu/linux-2.6.18gum/arch/arm/Kconfig to 1000 to see if I could make the scheduler swap tasks faster and therefore run my code often enough but when I compiled the new filesystem and reflashed the device it still seems to be running with the HZ value of 100. Heres the output of this program:

int main()
{
    struct sched_param sp;
    int periods;
    int ms;
    long i;   

    struct timespec ts;
    struct timeval tv;
   
    ts.tv_sec = 0;
    ts.tv_nsec = 10;
   
    sp.sched_priority = sched_get_priority_max( SCHED_FIFO ) - 1;
   
    if( sched_setscheduler( 0, SCHED_FIFO, &sp ) == -1 )
    {
        printf("Cannot set scheduler\n" );
        return(1);
    }

    for(i = 0; i < 10000000; i++)
     {
        nanosleep(&ts, NULL);
        gettimeofday(&tv, NULL);
        printf("%d\n", tv.tv_usec);
    }

    return(0);
}

Output:

870000
880000
890000
900000
910000
920000
930000
940000
950000
960000
970000

As can be seen it is still loosing processor time for 10ms at a time. Any ideas?

Thanks,
Brendan

-------------------------------------------------------------------------

Using Tomcat but need to do more? Need to support web services, security?

Get stuff done quickly with pre-integrated technology to make your job easier.

Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo

http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642_______________________________________________

gumstix-users mailing list

gumstix-users@lists.sourceforge.net

https://lists.sourceforge.net/lists/listinfo/gumstix-users