• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 34 subscribers
  • Views 307 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.

HD3SS3212-Q1: 10GBASE-R Crosstalk

Luke Korte
Prodigy 210 points

Part Number: HD3SS3212-Q1

Tool/software:

Goodday

We have a design where we use the HD3SS3212-Q1 as a MUX to connect a FPGA 10G SERDES interface to either a SFP connector or a Marvell 10GBASE-T1 PHY on the same board. The interface type for each is configured as 10GBASE-R.

I am seeing some errors during packet tests which look like they are caused by crosstalk. Specifically in the test case where the FPGA and Marvell PHY are selected, and at the same time there is either a RJ45 10GBASE-T SFP module in the corresponding SFP connector or some traffic is being sent to the same SFP connector using a direct attach cable.

Below is a block diagram to try explain this:

1. Is this type of crosstalk expected for the HD3SS3212-Q1 despite the low crosstalk figure in the datasheet? Or could this point to a deeper issue?
2. During testing I had a design where both sides of the HD3SS3212-Q1 were AC coupled and there were no termination resistors present to set a common mode voltage. The datasheet says this should be avoided, but in my test case the 10GBASE-R traffic passed with no errors. Does this make sense or is possible?
3. On a related note the datasheet says that the common mode voltage must be from 0 to 2V (Vcm) and the maximum differential voltage is 0 to 1.8Vpp (Vdiff). Also the absolute Differnetial voltage must be between -0.5 and 2.5V. I just want to confirm that given any Vcm and Vdiff the differential voltage swing must always remain between -0.5 and 2.5V?

Thank you

  • 0
    TI__Mastermind 26725 points

    Hello Luke,

    1. How much crosstalk are you seeing? The mux has a rated crosstalk of -28dB. This crosstalk could be coupled by other components in the system. Is there a schematic you can share?

    2.AC coupling both sides is not needed since one side is enough, but it shouldn't cause any issues. Termination resistors are not needed as long as the signals you are passing are within datasheet specifications. If there is common mode voltage higher than the limit then using pull ups are necessary. 

    3. No, absolute maximum ratings shouldn't be followed in an application. These are stress conditions where we cannot guarantee device operation. Vcm and Vdiff have to remain within recommended operation conditions.

    Thanks,

    Nir 

  • 0 in reply to Nir Gilgur
    Prodigy 210 points

    4338.UMAR-10G02.pdf

    Here is the applicable schematic page. 

    I have replaced R256, R258, R331 and R333 with 100nF caps, because the Vcm of U30 is 2.5V.

    This seems to work fine, but now both sides of the HD3SS3212RKSTQ1 are ac coupled, Is this fine to have both sides ac coupled with no termination resistors to some Vcm?

    thank you

  • 0 in reply to Luke Korte
    TI__Mastermind 26725 points

    Hello Luke,

    Thank you for sharing the schematic, I see no issues there.

    If the 2.5V Vcm is seen by the mux, then you are exceeding specs and risking reliability. 

    Its fine if both sides are AC coupled, that is what you need to do if the common mode voltage is above 2V. You also need a VBIAS below 2V on the line.

    More information can be found in the datasheet.  

    Thanks,

    Nir