• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 3 answers
  • Subscribers 103 subscribers
  • Views 754 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS320C6748: Fewest components - bare bones board configurations and hookups for these DSPs for simple tasks?

Bob Richardson
Prodigy 90 points
Part Number: TMS320C6748
Other Parts Discussed in Thread: OMAP-L138, TMS320C6678

Hi guys.  So i'm considering a more to a more powerful solution that I have implemented in a Spartan3AN1400 ATM.  I am deciding between FPGA and Ti DSP for my new target.  The design is DSP intensive but is a simple design that requires little more than a reference audio clock, and codec channels for signals In/Out.  My Spartan design consists of the FPGA chip(50MHz.),  codec, some control switches, status LEDs, audio clock input.  Memory and boot flash are inside the FPGA.  Size is an important factor.  How small (few components) and simple can I expect to be able to make a Ti DSP design based on the 320C6748 or other c6000, or other Ti DSP?  Minimum chip count design?  Other better floating point choice?  Are there documented examples?  Thanks for your comments and recommendations.  

Cheers,

  • 0
    TI__Guru**** 161600 points
    Hi Bob,

    I am not awara of document examples about this. Regarding how small can you make a design, that depends on how many peripherals do you need. Here is a list with the available OMAP-L1x boards:
    processors.wiki.ti.com/.../OMAP-L1x_Boards

    Other options regarding a processor choice are:
    OMAP-L138 -> same as the C6748, but has an arm core as well.
    TMS320C6678 -> this is a powerful multicore DSP (up to 8 cores).

    Best Regards,
    Yordan
  • 0 in reply to Yordan Kovachev
    Prodigy 90 points

    Thanks for that, Yordan. I found a treasure trove of little cards with Ti processors in the Launchpad section.
    It looks like they are all fixed point.
    I'm looking for floating point, so I have searching for a while for a C6000 style Launchpad, if you will, or example circuits, to get an idea of what the minimum requirements are to get a C6000 DSP running. Still on that search. Ti always has great engineering notes and example circuits.

    However, I also found the following:

    IQmath™ Fixed Point Math Library: A virtual floating-point engine. This library is a collection of highly optimized mathematical functions enabling programmers to develop with floating-point like math on fixed-point devices. IQMath also enables code to be seamlessly ported between fixed- and floating-point devices for ultimate code scalability. The IQMath functions facilitate execution speeds considerably faster than equivalent code written in ANSI C on fixed-point MCUs, while eliminating the burden of dealing with fixed-point scaling. Just write C floating-point code and let the compiler take care of the rest.

    What is this all about? Does this open up floating point to all Ti fixed point processors? Is this part of the free tools? Is it effective?  What about rounding error and accumulated errors?

    So it looks like it's for TMS320C64x only, but still a fantastic tool.

    Thanks and cheers,

  • 0 in reply to Bob Richardson
    TI__Guru**** 161600 points
    Hi,

    I've notified the RTOS team to elaborate on the IQmath library, because I am not familiar with it.

    As for the minimum requirements for the C6000 DSP, see the schematic checklist:
    processors.wiki.ti.com/.../_AM1x_Schematic_Review_Checklist

    The Critical Connections section is a must (minmum requirements to get the processor running). After that you can add whichever peripherals you would need from the peripheral section.

    Best Regards,
    Yordan
  • 0 in reply to Bob Richardson
    TI__Guru** 116000 points
    Bob,

    IQMATH as you guessed was a library created for TI fixed point DSP parts, for providing higher performance with floating point operations using the Q format that is described here:
    www.allaboutcircuits.com/.../

    Some of the functions in the library handle rounding and saturation like multiplication operations. There is an equivalent IQMATH library for C2000 devices but the one that you are refering to is probably applicable to C64x and C64x+ fixed point DSPs.

    Regards,
    Rahul