DP83867E: Impossible to make a ping with DP83867E Ethernet

Hi Guillaume, 

Could you please explain your request in this sentence below

"The problem is to find how to the MII loopback....to test the SGMII link between the MACs." Are you wondering how to put the device in MII loopback?

If you refer to table 4 it shows the availability of SGMII in MII loopback however, it does not provide the link status indication.

Cecilia

Hi Cecilia,

I'm looking for a solution or methodoly with Linux commands or other things to test the SGMII link between the PHY MAC and the T1042 MAC.

The first Step consist of setting the PHY DP83867E into MII/Digital loopback mode. I have found in the Troubleshooting guide but not for the SGMII link. Does it just necessary to change the value of the register 0x0032?

For the SGMII link, the AC-coupled capacitors must be place near the receiver or transmitter or at any position?

The register 0x0037 is always at 0x0000. Could you tell me what the reasons of no received sgmii page and no auto neg sgmii completion?

I have check the requirements to establish a SGMII link and i don't find where is the problem....

Could you help me?

Thanks a lot!

Best regards

Guillaume Arrias

Hi Guillaume

Can you share the entire register dump you see with the device?

Thanks!

Cecilia

Hi Cecilia,

Of course!

Have you any command that can i send on the mdio link to have the register's value of all registers with a minimum commmands?

I go to try Post increment access!

Guillaume

Hi,

Do you tell me if the lack of a pull up resistor on the MDIO signal might cause auto negotiation problem on SGMII? 

However, this lack might cause also communication on MII link, no?

Could you explain me the importance to set le LED1 for the Autoneg? It's just for the MDI interface?

Thanks a lot for your help!

I start to get back register values soon :)

Guillaume

Hi Cecilia,

Please find the values of a lot of registers:

address     Data

I have checked the clock electrical and frequency caracteristics and everything works fine!

Please, could you take a look on the register? Maybe there is a problem on 0x0032 and 0x005E? Maybe there is a bad value into the register 0x0170 for the IO_IMPEDANCE_CTRL bit name? FFFF is not to low to set the impedance of MAC IO/s? Normally, it must be set to 50Ohms and not at 35Ohms.... What's the meaning of this "impedance"? Is it for the SGMII link? The default value is set à 35Ohms by this register? Where is the trim indicated in our documentation page 87?

Is it necessary to put a pull resistor of 1.5K on the MDIO signal? The lack of this resistor could create SGMII autenogociation error?

Thanks a lot for your help!

Have a good day!

Guillaume Arrias

Hi Guillaume,

Yes I can take a look at the registers you have sent over. Thank you for recording these values for me. Please allow me 3-5 days to review these items and get back to you with my feedback.

In the meantime you can also look at our troubleshoot guide for the DP83867 on items to confirm some key measurements to ensure proper function.

http://www.ti.com/lit/an/snla246a/snla246a.pdf

Thanks!

Cecilia 

Hi Cecilia,

Have you taken a look at the registers sent almost two weeks ago?

I have already read the troubleshoot guide for the DP83867.

Maybe the problem is the lack of the pull up resistance on the MDIO signal that the SGMII Autonegotiation has not been done! I don't know.

What are the signs that would allow me to recognize signs of unstable MDIO link?

Perhaps the problem is the bad configuration of IO_IMPEDANCE_CTRL in the 0x0170 register. What is the impedance setted by this parameter?  single ended to ground impedance of SGMII link? 

Thanks so much for your fast reply!

Have a good day!

Take care of you!

Guillaume

Hi Guillaume,

Apologies on the delay, due to external circumstances and COVID-19 I have had a delay in responses to some E2E questions. On your question about the unstable MDIO link have you been able to probe the MDIO and MDC lines? You would be able to then see on your oscilloscope whether or not the signals are unstable. 

Based on the register reads for 0x1, it looks like you had a link established and that Auto-negotiation has been enabled. 

Are you still seeing issues on the SGMII registers? 

Thanks,

Cecilia

Hi Cecilia,

The current circumstances require more time to do the same job than ever before!

I have proben the MDC line and the frequency signal was about 500KHz. I think it's rather slow... I haven't proben the MDIO line. I could take a look with my own oscilloscope to get some mease of there signals. Maybe i need to solder a 1.5K pull-up resistor on the MDIO line?

Yes the SGMII registers are still at 0x00....

I would like to try to set the IO_IMPEDANCE_CTRL of the 0x0170 register at a 50 Ohms impedance. Could you tell me if this parameter set the impedance output of the DP83867 SGMII link? Actually this impedance has been set to 35Ohms impedance. However, the actual layout of this SGMII signals have been routed with 100 Ohms differential traces and single ended impedance to ground at 55 Ohms...

Have you another idea that i can to check?

Thanks a lot!

Guillaume Arrias

Hi Guillaume

Yes that is very true. Thanks for your patience, hope you are staying healthy and safe during these times.

Yes you can use register 0x170 to change the impedance control. Please let me know what you are reading and if you are able to adjust the values and see any change.

Have you also added the dummy straps shown in Figure 25?

Thanks,

Cecilia