• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 33 subscribers
  • Views 818 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.

DP 83640 Link Time

Damon Pritchett
Prodigy 80 points
Other Parts Discussed in Thread: DP83640

Hello,

We have a DP83640 Precision PHYTER implemented on a card and we are having issues with variable link times.  The link times seem to vary from card to card and over time and appears to be bimodal in that the link times are either about 248 ms or about 320 ms.  So far no other link times have been observed.  We are strapped for 100BASE-TX, Full-Duplex, Auto-negotiate disabled.  This behavior is causing system level timing issues.  What connections to the PHY or other conditions could cause this type of behavior?  Could one of these link times be considered nominal and, if so, which one?  We have seen cases where a card will power up with the faster link time and then slow down on subsequent power cycles, but we have also seen the reverse.  Anyone have any ideas on what might be going on here?

Many thanks,

Damon Pritchett

  • 0
    TI__Mastermind 31600 points
    Could you provide some additional details on the test setup? What is the link partner? What length cable is used?

    Patrick
  • 0 in reply to Patrick OFarrell
    Prodigy 80 points

    The link partner is another DP83640 and the cable length is approximately 2 feet.  The link partner resides on a test fixture which is used to test these cards.

    Damon

  • 0 in reply to Damon Pritchett
    TI__Mastermind 31600 points
    Damon,

    This is due to the process of adapting and equalizing the analog front end. At very short cables, the PHY make take a little longer to complete the process.

    Patrick
  • 0 in reply to Patrick OFarrell
    Prodigy 80 points

     

    Serial Number and Link Time (msec)

    Ethernet Cable Length (ft)

    0001

    2005

    2011

    GIC Link

    ETF Link

    GIC Link

    ETF Link

    GIC Link

    ETF Link

    14'

    248.437

    249.894

    248.094

    249.159

    248.414

    249.863

    10'

    248.524

    249.925

    248.215

    249.861

    248.630

    250.066

    8'

    319.825

    249.862

    248.275

    249.864

    248.514

    250.066

    7'

    320.127

    249.863

    248.207

    249.863

    248.538

    250.064

    6'

    320.175

    322.064

    248.045

    320.892

    248.494

    321.861

    5'

    320.142

    321.244

    248.395

    249.865

    248.536

    250.066

    4'

    320.100

    322.064

    248.191

    249.286

    248.657

    250.068

    3.5'

    319.848

    321.865

    248.093

    320.832

    248.450

    321.861

    1'

    320.028

    321.864

    248.094

    320.832

    248.386

    321.863

    8"

    320.021

    249.863

    248.138

    249.251

    248.513

    249.862

     

    Hi Patrick,

    Thanks for the reply.  We performed some experiments with different cable lengths and three different serial numbers (see data above).  It seems that the timing must be dependent on something other than just cable length.  Our assumption is that cable impedance or component variability or both are coming into play as well.  We’re also curious as to why the link time is either one time or another and not anything in between.  Can you give us a little more details as to what’s going on without getting into proprietary design information?

    Thanks,

    Damon