• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 33 subscribers
  • Views 313 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

Naveenchandra Veerabhadra Swamy
Prodigy 170 points
Part Number: DS280DF810

Hello Team,

My self Naveenchandra from Tejas Networks ltd. I am working on behalf of Signal Integrity team. We are performing the SI analysis of a Chip to Optical transceiver Module interface for one of our high speed products. We are using your Retimer device ( i.e DS280DF810 ) for our product.

Presently we are doing correlating the SI results with lab measurements. But we are not getting the same electrical specs measurements in the lab which we are getting in the SI analysis.

Please see our design inputs for more understanding. Below are the SI Analysis inputs considered during serdes channel simulation:

Data rate 28Gbps with NRZ mode, Stimulus used PRBS 31 pattern.

QSFP28 Optical Module as Transmitter and TI Retimer (DS280DF810) as Receiver.

Our end to end board Channel losses are around just IL(Insertion loss) as 7dB and RL(Return loss) as 20dB.

QSFP source amplitude ranges starting from 600mV to 1070mV. Presently its fixed with between 600mV and 800mV. Also, as of now we are considering no de-emphasis at QSFP transmitter side.

See below are the Results variations from SI to Lab measurements (i.e TI GUI tool).

We kept Auto CTLE and AUTO DFE enabled, VGA as 0 and EQ gain as 1. and we are could get eye densisty is around 600mV. But, we could see only 200mV to 250mV as max amplitude in the TI GUI tool measurement. So, its clearly understood that, there is 350mV to 400mV deviations in the results.

As per our understanding, EQ gain settings are playing crucial role in our simulations. but not in the lab measurements.

We would appreciate you if any one get us the valuable support on this. Also, please let us know if you need any other SI setups and Lab setup are required for your reference.

Thanks,

Naveen

  • 0
    TI__Mastermind 19235 points

    Hi,

    Related to: "We kept Auto CTLE and AUTO DFE enabled, VGA as 0 and EQ gain as 1. and we are could get eye densisty is around 600mV. But, we could see only 200mV to 250mV as max amplitude in the TI GUI tool measurement. So, its clearly understood that, there is 350mV to 400mV deviations in the results."

    • The "retimer_internal" test point for the TI IBIS-AMI is not equal or have full 1:1 correlation with the retimer eye opening monitor readbacks per channel registers 0x27 and 0x28.
      • The voltage amplitude levels will be different for these two separate measurements. But the relative improvement in eye height with vs without EQ applied should be very similar
      • The eye width values however should be more or less the same for retimer HEO readings vs simulation "Retimer_internal" test point
    • In terms of retimer VEO readings observed during actual device evaluation, values in the range of 200mV to 350mV are quite normal and typical for 28Gbps operation. The exact magnitude depends on the input channel characteristics

    TI has established the minimum thresholds below for retimer HEO and VEO values (per 0x27 and 0x28) based on retimer test validation for ensuring error free operation for BER 1E-12 test. During actual system evaluation, the host system should target to ensure that these minimum retimer input eye opening criteria are met, with some margin if possible 

    • HEO > 0.375UI
    • VEO > 196.875mV

    Debug question: For your "Chip to optical transceiver module" case in question could you provide what values you are observing for the retimer channel registers below? These values will allow me to assess the retimer Rx EQ adaption optimality. If using the TI GUI, the low-level page may be used to read these channel register hex values

    • 0x27, 0x28, 0x71, 0x72, 0x73, 0x74, 0x75, 0x8F

    Cordially,

    Rodrigo Natal

    HSSC Applications Engineer

      •  
  • 0 in reply to Rodrigo Natal
    Prodigy 170 points

    Hello Rodrigo Natal,

    Thanks for your reply. We are going through your suggestions.

    Could you please provide us your official mail ID, we have a follow up queries.

    Best regards,

    Naveen

     

  • 0 in reply to Naveenchandra Veerabhadra Swamy
    TI__Mastermind 19235 points

    Hi Naveenchandra,

    It looks like we had communicated via direct email in the past once before, so I already have your email address. I will send you direct message.

    Cordially,

    Rodrigo Natal

    HSSC Applications Engineer