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.

TLK2711 loopback

Hi

I have the TLK2711 SerDes on my board. I wanted to test it with the loopback feature.

At the moment I don't have a real data to send, so I took the GTX clock (coming from the FPGA) at 100Mhz

and started to divide it by two to generate clocks as data (meaning that D0 is a 100Mhz/2=50Mhz clock, D1 is 100M/2^2=25Mhz clock and so on up to D15 is 100M/ 2^16 = 1525Hz clock).

I shifted the derived clocks to verify the GTX 100Mhz clock will rise at the middle of the "data" clocks.

I activated the Enable/ Rx_EN/ LPEN pins, verified the MSB/ LSB K/ D bits are low (Data bytes) and I have noticed that signals started to appear on my FPGA input bus pins (and no more signal on the TX serial pins).

Problem is that the received signals, after the loopback, where not the transmitted fixed clocks, but rather they looked like real data, meaning all variations of pulses width.

Is it the expected result from such "data"?

Why don't I get the fixed clocks as my Rx data?

Thanks

Amnon

  • Hi Ammon,

    Have you tried using PRBS (Pseudo Random Bit Stream) Generator feature? Combining PRBS with loopback path.

    Regards!

    Luis Omar Morán Serna
    HIgh Speed Interface Group
    SWAT Team
  • Ammon,
    Did you supply Comma character on LSB for byte alignment?
    If you did not provide this, the serial stream will be randomly interpreting byte alignment and this will yield incorrect transmission.

    Regards,
    Wade
  • Hi Wade,

    What exact value in HEX should I send?

    What should be its frquency? Is this frequency of the Comma character measered in time units or in K-Byte data ?

    I guess when I send the Comma character I should also activate the TxLSB pin , true?

    Thanks

    Amnon

  • Amnon,

    This app note explains some of the key aspects.

    Regards,

    Wade

  • Hello

    We have read the appnote.

    I would like to ask:

    1. What the D5.6 has to do with the Comma?

    2. In the block diagram at the datasheet it presents two separate blocks for Comma detect. On for LSB and the other for MSB. Why appnote suggests sending Comma only over the LSB?

    3. In the appnote it says that one toggle over the TKLSB pin is enough to reach sync. How frequent should we send the Comma?

    Thanks

    Amnon

  • Hopefully the app note answered question 1 sufficiently.
    The comma on LSB is by specification on the TLK2711.

    The frequency of sending comma is not really defined.
    A bit error could cause erroneous byte alignment, so periodic sending of comma would be recommended.
    This would depend on your system, and tolerance of errors and duration.

    If a bit error caused an erroneous byte alignment, the data would be wrong until it was resent.

    Regards,
    Wade
  • Hi Wade

    I would like to start and thank you for your help.

    We found a small mismatch between FPGA and SerDes databus pinout, and now

    the internal loopback works fine.

    My question on the D5.6 is trying to understand what is so special about 110.00101 that it was selected as the data combination that will (if I understand correctly) increase the probability of the comma 0011111 to appear at the receiver.

    Thanks

    Amnon

  • The inverting idle k28.5/d5.6 causes the running disparity to flip. So two consecutive inverting idles will guarantee that the proper comma will be present on one of them.
    Regards,
    Wade
  • Hello

    I would like to know how is the TLK2711 affected in case there is no signal at its input?

    Is there a risk that random signals starts to appear on its output bus?

    If so, is there a way you recommend to prevent such random output (pullup/ down, etc.) ?

    If so, I will appreciate a detailed description of such solution.

    Thanks

    Amnon

  • Amnon, it would be best to start new post for new questions. It is easy to loose track of new questions to old posts.
    However, there is a threshold detection for LOS (los of signal) that will cause the RX bus to be all 1's. This threshold is ~200mv, and is a gross detection. It will detect if cables not connected. It does not indicate poor signal quality.
    BTW, this question has been answered prior.
    You can search for TLK2711 and LOS in E2E and find more related information.
    Regards,
    Wade

    Please mark posts with Verify Answer