• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 33 subscribers
  • Views 1456 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.

The unused pins on TLK2711-SP

Liang Li1
Prodigy 80 points
Other Parts Discussed in Thread: TLK2711-SP

I want to use the TLK2711-SP in single operation. In orther words, one TLK2711-SP only  transmit the serial data, and another TLK2711-SP only receive the serial data. In this case, how should I deal with the unused pins? Especially with the RXD, RXCLK, RXN, RXP, TXD, TXN, TXP, would it be better if I pull up or pull down these unused pins?

And, by the way, some engineers told me that  the PLL in the TLK2711 is sensitive to jitter in the range of 300KH to 3MHz. Does it means the GTX_CLK's is acceptable if  its jitter is slower than 40ps in the range of 300KHz to 3MHz?

Thanks for your help.

Liang Li

  • 0
    TI__Mastermind 20820 points

    Hi Liang,

    The unused inputs TXD should be tied either high or low so that they are forced into a know state. The RXD, RXCLK, TXN, and TXP outputs can be left floating or forced into a known state. It may not be a good idea to tie RXP and RXN (serial receive inputs) to VCC or ground since this could violate the common mode input range spec (See Vcmr spec on page 16 of datasheet).  It’s probably better to let these float or to use a resistive divider to set a fixed high or low differential voltage that is within the allowable Vcmr and VID ranges.

    The GTX_CLK is sensitive to jitter in the range of 300kHz to 3MHz so you should try to minimize this as much as possible. Jitter slower than 40pS is required for normal operation of the TLK2711

    Regards,

    Mike

  • 0 in reply to Michael Peffers
    Prodigy 80 points

    Thanks a lot, Mike.  

    I have another question. Theoretically, the ramdom jitter of a  fixed clock is increased with the testing time. And in reality, when I test the GTX_CLK's jitter with Digtal Serial  Analyzer, I find that the clock jitter is increased with the increased test samples. So, in this case, how can I assert that the GTX_CLK is acceptable for the TLK271-SP?

    Expecting your answer

    Liang Li 

  • 0 in reply to Liang Li1
    TI__Guru 62930 points

    Liang,

    I want to add a couple clarifications on biasing.

    The parallel outputs (RXD, RKMSB, RKLSB, RXCLK) should not be connected shorted to high or low, unless device is properly set to transmit only mode.  This is done by placing LCKREFN low.

    Biasing these outputs to known levels is not required.

    With respect to RXP and RXN.  These pins are self biased internally, and can be left floating.   

    I recommend reviewing the datasheet on the web with respect to the usage of LCKREFN.  Essentially, if the device is disabled, then RKMSB will take on new pin function and will not be tristated.  This is detailed on page 2, and the pin function tables.

    Regarding your clock jitter question.  Many scopes capable of measuring jitter have statistical software that can calculate the jitter based on fewer acquired samples.  I do not know if your analyzer is capable of this or not.   Otherwise you will need to obtain at a minimum the jitter to match your intended BER.  The TLK2711-SP was characterized with input conditions aligning to 1E-12 BER.  At full rate (2.5Gbps)  this results in a test time of 400s.

    Regards,

    Wade

  • 0 in reply to Wade Vonbergen
    Prodigy 80 points

    Thank you, Wade. I really have never taken notice of RKMSB in the Loss of Signal (LOS) Detection function before.

    Now, I use one TLK2711-SP woking in transmit-only mode. Does the LOS Detection function still work in this case? If it don't work, I think I can connect the LCKREFN, RXD[0:15], RKLSB and RXCLK to the ground though 10K resistors; leave RXP and RXN floating; and RKMSB is pulled up with 10K resistors, and used as a indication signal in LOS Detection function.

    Another TLK2711-SP is working in receive-only mode, so the LCKREFN, TXD[0:15], TKMSB and TKLSB are connected to VDD through 10K resistors, and TXP and TXN are left floating.

    I want these works will give better reliability to my design. Do you think so?

    As you said, my digital serial analyzer has the statistical software. It can calculate the random jitter based on few hundred milliseconds acquired samples. With the test results, I think I can figure out the random jitter on the condition of 1E-12 BER.

    Regards,

    Liang Li

  • 0 in reply to Liang Li1
    TI__Guru 62930 points

    Liang,

    Your transmit only biasing looks good.   However, you cannot detect LOS on the transmit only side as there is not any signal.

    Also note for the the RKMSB to be used as LOS only applies when device is disabled.  When device is enabled RXD0-15, RKLSB, and RKMSB will assert high for LOS condition.

    This was a special case that allows the device to be disabled and still detect LOS on only RKMSB.

    The receive-only configuration is acceptable.

    Regards,

    Wade

  • 0 in reply to Wade Vonbergen
    Prodigy 80 points

    Thanks, Wade. I get further comprehension about the TLK2711-SP from your reply.

    Best Regards,

    Liang Li