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.

SN75DP130: SN75DP130 Link Training Issue

Part Number: SN75DP130

Hello TI,

Our motherboard + expansion board which use the DP130 somehow uses HBR instead of HBR2.

Our GPU is capable of HBR2 and we're able to confirm that the sink we use for testing supports HBR2 as well. The sink is not using any redriver.

The motherboard uses the DP130 on one port and connects to an expansion board that also uses the DP130.

It seems like the redriver is causing the lowering of the link rate because on the same motherboard without the redriver, the DisplayPort can successfully use HBR2 during link training.

Is there any setting that needs to be changed in order for the HBR2 to be selected during link training? (This is provided that cables, sink, and source support HBR2).

Or is cascading 2 DP130's causing the problem? (Reference here:  https://e2e.ti.com/support/interface/f/138/t/556372?tisearch=e2e-sitesearch&keymatch=sn75dp130%2520dpcd )

We are using the DP130's as source + expansion (not as sink). The sink connects through the expansion.

Looking forward to your response.

Thank you very much.

  • The DP130 will snoop the AUX bus and then set the data rate based on the AUX snooping information, it will not force the data rate. Depending on how the link training is implemented in the Source, it is possible that the link training failed at HBR2, the Source downgrades the data rate to HBR, and then able to pass link training at HBR.

    The DP130 is a limited re-driver which means it has its own SWING and Pre-emphasis level. This SWING and Pre-emphasis level is different from the Source SWING and Pre-emphasis level. During the link training, the Sink would request Pre-emphasis and voltage SWING level based upon the output of the DP130, not the DP Source. Both DP130 and the Source would adjust its EQ, Pre-emphasis and voltage SWING accordingly based upon the request from the Sink. But for the signal path between the Source and the DP130, the Pre-emphasis and voltage SWING may not ideal. potentially cause the signal to be over- or under- equalize, and the problem gets worse with multiple limited re-driver in the same path.

    You can try to disable DP130 link training and manually set the EQ, Pre-emphasis, and SWING level based on the loss of each signal path or switch to DP141 or DP142 linear re-driver.

    Thanks

    David

  • Hi David,

    Thanks for your reply.

    If we disable link training in the DP130, does that mean, we need to configure the parameters via I2C?

    Does that mean we could also manually set the LINK_BW_SET to HBR2? (default is HBR)

    If that is the case, how will the source and sink perform the link training now?

    How can the DP130 that will be manually set, affect the link training between the source and sink?

    Also, why would there be a problem with multiple limited re-drivers in the same path?

    Looking forward to your response.

    Thank you very much.

    Best regards,

    ice

  • Ice

    Please see the block diagram. The DP130 is a limited re-driver which means it has its own SWING and Pre-emphasis level. The actual DP130 SWING voltage and Pre-emphasis is different from the Source SWING voltage and Pre-emphasis. During the link training, the Sink would request Pre-emphasis and voltage SWING level based upon the output of the DP130 #2 (Link 3), not the DP Source (Link 1). Both DP130 #1, DP130 #2 and the Source would adjust its EQ, Pre-emphasis and voltage SWING accordingly based upon this request from the Sink. While the DP130 #2 SWING and Pre-emphasis level may be ideal for Link 3, the same SWING and Pre-emphasis level may not be ideal for Link 1 and 2 since Link 1, 2, and 3 will have different loss.

    To overcome this, you need to disable link training in both DP130 #1 and #2. The EQ of DP130 #1 will be manually set to compensate the loss of Link 1. The DP130 #1 SWING/Pre-emphasis and DP130 #2 EQ will be manually set to compensate the loss of Link 2. The DP130 #2 SWING/Pre-emphasis will be manually set to compensate the loss of Link 3.

    With Link Training being disabled, the link bandwidth and lane count must be set. The link bandwidth can be set to 5.4G, but it will still support lower data rate. 

    Thanks

    David

  • Hi David,

    Thanks for your response. The process is a bit clearer now.

    We do have a few more questions.

    If we set lane count to 4 lanes, will it also support lower lane count (e.g. 2 lanes), or should we set it based on the source lane count?

    How do we choose the best EQ, SWING and Pre-emphasis manual settings for each DP130?

    Is there a known set of procedures to identify it? Do we have to manually measure the loss? 

    Thank you very much.

    ice

  • Ice

    With DP130 link training disabled, please set the lane count to 4, and it will still work with lower lane count. The max supported lane count will then be negotiated between the source and sink. 

    The EQ, SWING, and Pre-emphasis level need to be chosen base on the insertion loss of each link. 

    You can measure the insertion loss of each link, or you can estimate the loss/in of each link using this formula.

    You can also have the source sending the HBR2 compliance pattern, having a scope connected to the output at Link 3, and measure the eye diagram while tune the EQ/SWING/Pre-emphasis level.

    If you have a AUX link training monitor box, you can also tune the EQ/SWING/Pre-emphasis level while monitor the result of equalizer training portion of the link training.

    Thanks

    David

  • Thanks David!

    We were able to implement HBR2 based on your instructions.