• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 98 subscribers
  • Views 483 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

6713 - High(er) speed timer input?

Joe Bauman
Prodigy 160 points

Good Morning -

Does anyone have any experiences with running the timer input on a 6713 (running at 300MHz) at speeds higher than 10MHz?  The datasheet says it should work for speeds up to around 75MHz.  I'm using a function generator to supply the signal - it is set to about 3V peak-to-peak (0V-3V) - and then comparing that input signal to the 1 PPS output signal from the timer circuit using a frequency counter.  At 10MHz it works perfectly, but at 20MHz or 40MHz it appears to be missing cycles occasionally.

Thoughts?

Thanks much,

Joe.

  • 0
    TI__Guru* 84110 points

    One possible problem would be if it is a signal integrity issue. Can you look at the TINPx input signal at the various speeds and confirm that it is a clean signal from 0-3V with a good rise time and remains a square wave?

  • 0 in reply to RandyP
    Prodigy 160 points

    Hi Randy - Thanks for the response.

    I'm attaching some pictures here - hope they help.  All are scope traces measured at a pad connected straight to the DSP TINP pin.

     

    IMAGE 1: 10MHz OCXO on-board oscillator.  Results are very reliable.

     

    IMAGE 2: 10MHz - function generator.  Results are MOST reliable.

     

    IMAGE 3: 20MHz - function generator.  Results are least reliable.

     

    They look pretty clean to me, although the rise time on the OCXO and the 20MHz plots could be a little better.  Thoughts?

    Thanks much,

    Joe.

  • 0 in reply to Joe Bauman
    TI__Guru* 84110 points

    I do not see ringing on any of these, but the scope resolution is so low that we might not see anything. None of these look like good square waves, and the rise and fall times are pretty slow for something going into a clocking input. Is this with your function generator set to put out a square wave? Are you at its upper limit, or is this due to the scope display or the board loading?

    You are not running too fast for the timer to be able to respond, so the only thing I would suggest for the cause is signal integrity. Could you try putting a buffer on the board near the TINP to sharpen this signal?

  • 0 in reply to RandyP
    Prodigy 160 points

    I put a buffer between our onboard OCXO and the TINP and it did clean up the edges quite a bit.

     

     

    Also with a PLL 2x multiplier inline to get 20MHz.

     

    At this point the 10MHz is rock solid and our timing application is showing good results.  At 20MHz, however, we are still seeing strange behavior.

    One telling (I think) symptom shows when I compare the 20MHz input and the TOUT pin configured in pulse mode.

    It's more clear to watch it move, but triggering on the CH1 pulse we see the CH2 20MHz signal jump back and forth.  Here the pulse lines up with a 20MHz falling edge, and it will move between 20MHz rising and falling edges.  The pulse also appears to vary between 2 full cycles long and 1.5 to 2.5 cycles long.

    I can continue to clean up edges - maybe find a faster buffer or something.  We are working on getting a better scope, so that might help too.

    Thanks again for the comments,

    Joe.

  • 0 in reply to Joe Bauman
    TI__Guru* 84110 points

    The Timer Ref Guide says that the TINP signal will be synchronized. This means that it is sampled by an internal clock and then the timer operations will occur relative to that internally synchronized signal timing. Since the datasheet says that the minimum TINPx pulse duration must be 2 cycles of the DSP clock, this implies that the synchronizing clock is running at the full DSP clock rate; the spec is set to guarantee that the high pulse will be seen by at least one internal clock rising edge.

    It makes sense, then, that as you get higher in frequency there will be more apparent jitter of the clock operation. You should not miss any edges of TINPx up to 75MHz, but any timer event (like TOUTx) could vary because of this jitter.

    This may not fully explain what you are seeing, but it is a start.

  • 0 in reply to RandyP
    Prodigy 160 points

    Hi Randy - 

    Thanks again for the responses.  We are going forward with our design using 10MHz and getting good results.  We might try to increase the frequency in the future, and if we do I will keep your suggestions in mind.

    Much appreciated,

    Joe.