• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 3 answers
  • Subscribers 33 subscribers
  • Views 356 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.

DS280DF810: DS280DF810 application

Penn Zhang
Intellectual 1180 points
Part Number: DS280DF810

Hello,

1. I have seen the datasheet as below, retimer could only acceprt the AC signal higher than 196mVppd, I just want to know what's your test condition, what 's the amplitude of 25.78125Gbps with PRBS7 pattern signal? after 20dB it must higher than 196mV right?

2. If my signal only 1.225V, after 20dB loss, it only ~122mV, it means retimer can't detect this low signal, so it means I must put the retimer closer to the start, right?

  • 0
    TI__Mastermind 19235 points

    Hi, see my inputs below.

    • As per the datasheet "test condition," the measurement point for this amplitude parameter is at the retimer device input pins. So yes, the assumption is that the Tx launch amplitude before the 20dB test channel must be significantly above 196mVppd in order to allow observation of 196mVppd after the test channel.
    • The 20dB insertion loss value listed on the datasheet is for the 25.78125Gbps Nyquist frequency. The lower frequency component of the signal will be attenuated by less than 20dB. So, if the launch amplitude before the channel is 1.225Vppd with 25G PRBS7 then the measured signal outer amplitude after the 20dB channel would exceed 122mVppd, and the retimer would indicate signal detect asserted
    • 1.225vppd is actually a rather large launch amplitude. You may use that amplitude level for a transmission channel of loss as high as 35dB at 25Gbps.

    Cordially,

    Rodrigo Natal

    HSSC Applications Engineer.

  • 0 in reply to Rodrigo Natal
    TI__Intellectual 1180 points

    Hi Rodrigo,

    i'm still confusing.

    1. I want to confirm Vsdat=196mVppd first, this means if the signal at the input pin is lower than 196mVppd, the retimer can't accept this signal and recover this signal right?

    2. Based on this, if the 1.225V signal at begining, after 20dB loss, it will be 122mV, this is lower than 196mV, why the retimer would indicat signal detect asserted?

    3. 1.225V after 35dB loss will be ~20mV, it is far away from the 196mV, why it is still work? How to calculate it?

    Thank you.

    Penn 

  • 0 in reply to Penn Zhang
    TI__Mastermind 19235 points
    1. Correct, if the signal outer amplitude at the signal pins is lower than 196mVpp then signal detect may be de-asserted
    2. The retimer CTLE + DFE are able to compensate for the 35dB loss, as they can provide 35dB of high frequency boost

    Cordially,

    Rodrigo Natal

    HSSC Applications Engineer

  • 0 in reply to Rodrigo Natal
    TI__Intellectual 1180 points

    Hi Rodrigo,

    1. If the signal below 196mVpp will be de-asserted, the 1.225V after 20dB loss is 122mV, 122mV < 196mV, I think the retimer will de-asset this signal, and this could not work? Right?

    2. On the other hand, if you want to use the signal after 20dB loss, it should be at leasr 196mV*10=1.96V, am I right?

    Looking forward to your reply.

    Thank you.

    Penn

  • 0 in reply to Penn Zhang
    TI__Mastermind 19235 points

    If the signal below 196mVpp will be de-asserted, the 1.225V after 20dB loss is 122mV, 122mV < 196mV, I think the retimer will de-asset this signal, and this could not work? Right?

    That is not correct. Only the high-frequency component of the signal will be attenuated by 20dB. The lower frequency component will be attenuated by less than 20dB, allowing an outer amplitude greater than 196mV to be observed at the retimer input pins. When either PRBS or standard encoded data is transmitted there is both low frequency and high frequency component to the signal. See attached document with a figure illustrating this concept Figure_eye_diagram_freq_dependent_loss.docx

    On the other hand, if you want to use the signal after 20dB loss, it should be at leasr 196mV*10=1.96V, am I right?

    Not correct. Assuming standard encoded data, a Tx launch amplitude in the range of 600mVppd to 1.2mVppd is ok for transmission over 20dB channel

  • 0 in reply to Rodrigo Natal
    TI__Intellectual 1180 points

    Hi Rodrigo,

    Thanks a lot. Your explain correct my understanding.

    3 more questions:

    1. The data signal frequency is based on the data, it could be change very fast or very slow, right?

    2. If there is a part of data is all high frequency, so this loss will be very big, will this data lost? Only received the lower frequency data? How to get the high frequency data at this situation?

    3. Let's think about the worst case, 25Gbps data, the data is changing every moment, it means the frequency is always at the high frequency, it means there is no data could be received after loss, right?

    4. Normally, when we design the trace length of retimer, should we think about the data? because of the frequency.

    Thank you very much.

    Penn

  • 0 in reply to Penn Zhang
    TI__Mastermind 19235 points

    Hi, my inputs below.

    1. The data signal frequency is based on the data, it could be change very fast or very slow, right?

    • The data rate is defined by the 1-bit pulse duration. However, the data encoding type (e.g. 8B/10B vs 64b/66B) will affect the signal integrity

    2. If there is a part of data is all high frequency, so this loss will be very big, will this data lost? Only received the lower frequency data? How to get the high frequency data at this situation?

    • In the hypothetical scenario where data is strictly 1010 pattern you may force signal detect to be always asserted on the retimer by setting channel register 0x14[7]=1

    3. Let's think about the worst case, 25Gbps data, the data is changing every moment, it means the frequency is always at the high frequency, it means there is no data could be received after loss, right?

    Yes, if data is strictly 1010 pattern at 25G and the loss for transmission channel is high signal detect may be asserted. obviously in a real application you would not run into this scenario as data would be encoded packets

    4. Normally, when we design the trace length of retimer, should we think about the data? because of the frequency.

    • The first order effect to consider is the insertion loss at the Nyquist frequency. When we design and test our retimer however TI always make sure to evaluate the worst case data stress pattern specified by the standards

    Cordially,

    Rodrigo Natal

    HSSC Applications Engineer