DP83867IR: Gigabit Network Rate issue

Hello Cherry Zhou

Thank you for the message.

As  understand from the above message. the communication works for 10/100 but not for 1000M.

Any idea on the length of the GMII signals to the host and the MDI signals ( distance between magnetics and the PHY).

I do have a few early observations that i can provide after i receive the above inputs.

Regards,

Sreenivasa

Hi Sreenivasa,

Thanks for your reply!

Kallikuppa Sreenivasa said:

Any idea on the length of the GMII signals to the host and the MDI signals ( distance between magnetics and the PHY).

Below are info the customer provided:

Thanks and expecting your reply!

Best Regards,

Cherry

Hello Cherry Zhou

Thank you for the inputs. 

The GMII signals seems long. I would assume the signals are matched. 

Can you please check if the TX signals have series resistor near to the host.

I also assume maser means the host on the bard - please comment.

Regard,

Sreenivasa

Hello Cherry Zhou

I also assume master means the host on the board - please comment.

Regards,

Sreenivasa

Hi Sreenivasa,

Thanks for your reply!

I have updated your reply to the customer, if there is anything new I would follow up asap.

Thanks again!

Best Regards,

Cherry

HR911130C - magnetics has center tap shorted.

Can you check if customer has a magnetics without center tap shorted that he could test. Also, can you please ask customer to add additional caps along with C49.

Power supply decoupling is less  - Each pin needs 100nF + 1uF and each power supply rail type additionally requires 10nF + 10uF. Please refer to the device datasheet for more details. Please request customer to add the same and test.

Regards,

Sreenvasa

Hi Sreenvasa,

Thanks for your help!

The customer has checked and got the following info:

Kallikuppa Sreenivasa said:

Can you please check if the TX signals have series resistor near to the host.

No, there is no series resistor near to the host.

Kallikuppa Sreenivasa said:

I also assume maser means the host on the bard

Yes, it is.

Kallikuppa Sreenivasa said:

Can you check if customer has a magnetics without center tap shorted that he could test.

Do you mean to replace the RJ45, change it to an RJ45 magnetics without a center tap shorting? Is there a recommended model?

Kallikuppa Sreenivasa said:

Also, can you please ask customer to add additional caps along with C49.

C49 capacitance is on HR911130C, what do you mean by "add additional caps along with C49"?

Also, The DP83867 related pins and rails have changed the capacitance according to the chip manual (new board in production).

Thanks again!

Best Regards,

Cherry

Hello Cherry Zhou

Please refer the below link for the magnetics.

https://e2e.ti.com/support/interface-group/interface/f/interface-forum/616947/dp83867ir-pulse-transformer-of-dp83867irpap

Also, can you please ask customer to add additional caps along with C49.

Hi Sreenivasa,

Thanks for your reply!

Please check the latest schematics:

https://e2echina.ti.com/cfs-file/__key/communityserver-discussions-components-files/59/GMII.pdf

Thanks!

Best Regards,

Cherry

Hello Cherry Zhou

Thank you. I will review and provide my inputs.

Regards,

Sreenivasa

Hi Sreenivasa,

May I know is there any updates? How about inputs?

Thanks and expecting your reply!

Regards,

Cherry

Hello Cherry Zhou

Thanks for checking and i will review the new schematics.

In the meantime can you please check if customer will be able to perform some loopback tests per 8.4.4 Loopback Mode.

Regards,

Sreenivasa

Hello Cherry Zhou

I reviewed the schematics, The critical section is the RJ45 magnetics type and the GMII trace length. 

Did you have a chance to check if customer can perform some loopback tests.

Here are few additional suggestions that customer could try

Reset is not controlled by a GPIO. Please check if there is a GPIO customer can use to control the Reset. The Reset by the GPIO can be released after the power is stabilized.  Otherwise Please increase C1309 to 10uf or 22uF and test.

R4983 33R resistor value for LED current limit is too low. Please increase to 220R

MII_TX_CLK R4981 - please change the value to 0R or 10R and test.

Regards,

Sreenivasa

Hi Sreenivasa,

Thanks for your review and reply!

I have updated your suggestions to the customer. If there is any follow-up info I will update to you asap.

Thanks again!

Best Regards,

Cherry

Hello Cherry Zhou, 

Noted and thank you.

Regards,

Sreenivasa

Hi Sreenivasa,

RJ45 magnetic type has been replaced with no center tap shorted (model 7499111221A).

And for the loopback tests and the suggestions you given, the customer will test on the new board in mid-November,  after the customer feedback to me the results, I will reach you asap.

https://e2echina.ti.com/cfs-file/__key/communityserver-discussions-components-files/59/7499111221A.pdf

Thank you so much!

Best Regards,

Cherry

Hello Cherry Zhou, 

Thank you for the inputs.

If customer is working on a new board, please suggest them to add series resistors on all the TX signals from the MAC to the PHY.

Also, please suggest them to optimize the trace length to less than 6000 mils.

If you do not have any additional questions, Do you want to close this thread for now and open a new thread when customer has the new PCB. If yes, please click the resolved button to close.

Thank you.

Regards,

Sreenivasa

Hello Cherry Zhou, 

If you do not have additional questions, please close the thread by clicking the resolved button.

Regard,

Sreenivasa

Hello Cherry Zhou, 

If you do not have additional questions, please close the thread by clicking the resolved button.

Regard,

Sreenivasa