• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 32 subscribers
  • Views 171 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.

TUSB1002A: TUSB1002A placement suggestion

Vincent Chiu
Prodigy 190 points
Part Number: TUSB1002A
Other Parts Discussed in Thread: TUSB522P

Tool/software:

Hi TI Expert

Our placement is below, total trace length from Jeston Orin (host) to type-c connector is around 12.3 inch, which is over Jeston Orin's layout guidance (maximum trace length is 6.2", loss budget is -3.78dB@2.5G).

Our application is USB 3.2 Gen1 (5Gbps), should we modify TUSB1002A location closed to HS3SS3212 or keep current placement?

BTW, Jeston Orin's loss budget is based on EM370 (middle loss) and our board is base on TU863+(low loss)

Regards,

Vincent

  • 0
    TI__Mastermind 24880 points

    Hi Vincent,

    Do you know the loss of your trace material in dB/inch at 2.5GHz? If you have the Dk/Df constants I can help estimate this.

    For a better understanding of how the TUSB1002A compensates for loss, please see this E2E FAQ.

    Best,

    Shane

  • 0 in reply to Shane Hauser
    Prodigy 190 points

    Hi Shane

    The loss is -0.4dB/Inch@2.5GHz. Please help to advice to keep current location or we are considering to move 2.5 inch close to  HS3SS3212 (4.6 to 2.1")?

    Regards,

    Vincent

  • 0 in reply to Vincent Chiu
    TI__Mastermind 24880 points

    Hi Vincent,

    Here are the loss values for your system based on that trace loss:

    Just to confirm, the USB host can only compensate for 3.78dB on its receivers correct? If so, you may not want to move the TUSB1002A further from the host. The TUSB1002A will not compensate for the RX path between the re-driver and the host, so moving the re-driver further from the host will require the host to compensate more loss on its receivers:

    Have you considered using a limited re-driver like the TUSB522P? Since you're only doing 5Gbps this could be an alternative. Limited re-drivers are able to compensate on both receivers and transmitters:

    You can see the difference between the 1002A and TUSB522P in this E2E FAQ.

    Best,

    Shane

  • 0 in reply to Shane Hauser
    Prodigy 190 points

    Hi Shane

    Due to the host is Nvidia Orin module, and according to the layout guidance only 3.7dB reserved for PCB trace lost (connector loss is included, but we need to consider other loss source ie. via, MUX. CMC)

    For the host RX concern, we want to keep the current location, but for the redriver TX to connector? will the current placement lead to the signal integrity issue?

    Regards,

    Vincent 

  • 0 in reply to Vincent Chiu
    TI__Mastermind 24880 points

    Hi Vincent,

    For the TX path the USB3 spec has a defined loss budget of 10dB on the host side of any 5Gbps connection:

    Since you are using a low loss material with only 3.04dB between the re-driver and connector, your TX loss should be well within the budget.

    Best,

    Shane

  • 0 in reply to Shane Hauser
    Prodigy 190 points

    HI Shane

    Thanks for your explanation, due to the linear re-driver will not compensate the TX (re-driver output to connector) and we check the application note that the re-driver should be "closed to" connector and 2 inch will be the best location.

    However, from you explanation, the driving of linear re-driver output could provide 10dB lost capability ? In spec of TUSB1002A output voltage swing linear range and receiver equalization DC gain is only 2 dB.

    Regards,

    Vincent

  • 0 in reply to Vincent Chiu
    TI__Mastermind 24880 points

    Hi Vincent,

    Vincent Chiu said:
    However, from you explanation, the driving of linear re-driver output could provide 10dB lost capability ?

    The TUSB1002A can provide up to 10.7dB of loss compensation at 2.5GHz (5Gbps). This is controlled by the EQ settings, not VOD/DC gain.

    For a description of VOD, DC gain, and EQ please see this E2E thread. 

    Since your system is relatively low loss, I suggest using a lower EQ setting to avoid over-equalizing the signal. What I referenced in my last response was the USB3 loss-budget spec. This states that a USB3 system can have up to 10dB of loss between the host and the Standard-A receptacle that does not need to be compensated. This allows the USB3 signal path to work even if our re-driver cannot compensate all of your loss (3.04dB to be exact).

    Vincent Chiu said:
    Thanks for your explanation, due to the linear re-driver will not compensate the TX (re-driver output to connector) and we check the application note that the re-driver should be "closed to" connector and 2 inch will be the best location.

    Yes in practice it is generally better to keep the TUSB1002A closer to the connector to compensate the TX path, and let the USB host compensate the RX path. The issue is (my understanding is) that your USB host can only compensate for 3.78dB loss on the RX path, which is not much. This means the re-driver will need to be closer to the USB host to help compensate the RX path, or your host may not be able to recover the incoming USB signal.

    Best,

    Shane