DP83TC811S-Q1: RGMII in DP83TC811s

Hi team,

I am writing to seek help for RGMII standard, especially the TX_CTRL.

In DP83TC811s datasheet, RGMII interface TX_CTRL true table is shown below. When data comes, TX_CTRL/RX_CTRL should change between 1 and 0. At the negative edge of TX_CLK, the TX_ER should be low.

But from DP83867is datasheet, when in RGMII is used for 1000M mode and no error, RX_CTRL should always high.

Below is the testing waveform, Blue is TX_CTRL and yellow is TX_Dx, it conflict with our datasheet, which causes the IP error.

Could you please analyze why RGMII in two datasheet are different? Is there something misunderstand?

Thanks a lot.

Best regards

Chen

Hi Gerome,

Thanks for your reply.

Now we need to figure out TX_ER signal is active high or negative high in different situation, Audo Phy 100M RGMII, traditional Phy 100M RGMII and 1000M RGMII.

So in my understanding, DP83TC811S also match RGMII standard, and just 100M RGMII is different from 1000M RGMII which causes this misunderstanding, am I right?

Could you please also confirm if there is difference between auto Phy's 100M RGMII and traditional Phy's 100M RGMII interface?

Thanks a lot.

Best regards

Chen

Hello Chen,

You are correct in your understanding that the primary difference is between 1000M RGMII and 100M RGMII. There is no difference regarding the RGMII standard between automotive and traditional ethernet.

Sincerely,

Gerome

Hi Gerome,

This information is very important. Now we are stuck at this point.

Do you happen to have some material about 100M RGMII standard? I can only find 1000M standard online.

Thanks a lot.

Best regards

Please also see the description both in online documents and DP83867is datasheet as below. I am very confused about this message.

Could you please give some solid information about TX_ER polarity? And do you know if there is registers can change TX_CTRL(TX_ER) polarity？

Thanks a lot.

Hello Chen,

The RGMII standard will act the same regardless of the MDI speed in terms of the signaling. However, the biggest difference is during 1000Mbps mode, during the falling edge you will have data present that is vital to the communication as this allows DDR. During 10 and 100Mbps operation, there may be data on the falling edge as well, but it will most likely be a duplicate, and as this is not DDR during these modes, it will not affect the communication.

Regarding TX_ER, this is an output from the MAC so if there were controls for it, it would be there. However, as this is a standard that the PHY must follow to communicate, I do not expect there to be any way to change the polarity.

Sincerely,

Gerome

Hi Gerome,

In my understanding, TX_CTRL control logic in auto phy standard 802.3bw is different from the one in 100M or 1000M RGMII interface, which conflict with the conclusion you mentioned before. Could you please give some more comments and details about this?

I want to make myself more clear below.

From DP83TC811s datasheet, during normal data transition, TX_CTRL on positive edge of CLK should be 1, TX_CTRL on negative edge should be 0.

From DP83867IS datasheet, during normal data transition, however, TX_CTRL on positive edge and negative edge of CLK should be both 1.

In below description, 100M logic is the same with 1000M logic.

So in my understanding, TX_CTRL control logic in auto phy standard 802.3bw is different from the one in 100M or 1000M RGMII interface, am I right?

Could it be possible there is more than one version of RGMII regulation?

Appreciate if you can give some comments.

Thanks.

Best regards

Hi Gerome,

Thanks for the clarification.

Thus DP83TC811S's RGMII logic is not 802.3bw standard neither, am I right?

And we need to change the logic from standard to the one in DP83TC811s datasheet.

Hi Chen,

That's correct. IEEE802.3bw dictates data communication on the MDI line over automotive ethernet at 100Mbps. This is independent of the RGMII standard as that is on the MII portion of the PHY.

Sincerely,

Gerome