MPC5777C-416DS FEC RMII Problems with 100Mbps

取消
显示结果 
显示  仅  | 搜索替代 
您的意思是: 
已解决

MPC5777C-416DS FEC RMII Problems with 100Mbps

跳至解决方案
1,559 次查看
martinstanzel1
Contributor II

Dear NXP,

 

I am using the MPC5777C EVB in connection with the MPC5777C-416DS. Using the FEC in RMII mode I want to send some data to the onboard PHY (DP83848C) and receive the same data back on the MPC5777C. I have therefore configured the PHY registers for RMII and loopback.

If the FEC is configured to 10Mbps operation then the setup works as supposed, the MPC5777C receives the exact same data back that it sends. If the FEC is configured to 100Mbps then only about 20% of the received data is correct, the rest of the data is corrupted (several bits change their values).

Using the FEC in MII mode there are no issues with 100Mbps operation.

Should RMII in 100Mbps operation be working with this setup? Do you have an idea on the cause of this issue or any other hint? 

Thanks,

Stanzel

标记 (3)
1 解答
1,188 次查看
martin_kovar
NXP Employee
NXP Employee

Hello Martin,

this is information I received from application team:

A colleague mentioned that another customer had to enable the hysteresis on the clock signal and so I did that and now RMII 10 Mbps seems to be working as flawlessly as MII 100 Mbps.

If you would like to get the example I received from application team, please create case via NXP website and I will assign it and send you the example. Unfortunately, I cannot share the example here because of some legal reasons.

Regards,

Martin

在原帖中查看解决方案

6 回复数
1,188 次查看
divyesh_kumar
Contributor I

Hello Martin Stanzel,

Even I had faced similar issue and was able to solve it by adding a resistor (limiting current) between pin 2 and 3 of J63 on mother board MPC57xx rev D. I was able to send and receive with no loss at 100Mbps of RMII mode and also at 10Mbps speed.

Hope this solves your issue too.

For RMII mode jumper settings :

1. On daughter card (MPC5777C-416DS) connect jumper on following pins

J510 : 2-3

J524 : 1-2 (checked with DSO)

J24 : 1-2

J25 : 1-2

2. On Motherboard (MPC57xx rev D)

J63 : 1K Ohm resistor between pins 2-3

J64 : open

Regards,

Divyesh

1,188 次查看
lukezhang
Contributor III

Hello divyesh.kumar

I tried to use you way (put a 1k ohm resistor across J63 2-3 and leave J64 open), and I can get 10Mbps mode work with no data loss but I couldn't get 100Mbps work. My receive interrupt is triggered but the receive data is always 0 under 100Mbps. I'm using MPC57xx rev D and MPC5777c 512DS. Any suggestion? Thanks.

Regards,

Luke

0 项奖励
1,188 次查看
martin_kovar
NXP Employee
NXP Employee

Hello Martin,

we solved this problem some time ago with one of our customer and also with application engineer and the result is we were able to run FEC 10Mbps speed via RMII mode.

Using MII mode, there is no problem with 100Mbps speed.

Application engineer told me: Suspecting similar board layout issues that limit the speed to 10 Mbps max to get RMII working.

Regards,

Martin

1,188 次查看
martinstanzel1
Contributor II

Hello Martin,

thank you for your reply. Did you/your customer try out 10Mbps RMII on the EvalBoard connecting an external device? In my setup there are no problems using the PHY in loopback mode, but connecting an Ethernet cable and disabling loopback I still observe significant data loss. Do you think the board layout may also be the cause for this?

Thanks,

Stanzel

0 项奖励
1,189 次查看
martin_kovar
NXP Employee
NXP Employee

Hello Martin,

this is information I received from application team:

A colleague mentioned that another customer had to enable the hysteresis on the clock signal and so I did that and now RMII 10 Mbps seems to be working as flawlessly as MII 100 Mbps.

If you would like to get the example I received from application team, please create case via NXP website and I will assign it and send you the example. Unfortunately, I cannot share the example here because of some legal reasons.

Regards,

Martin

1,188 次查看
martinstanzel1
Contributor II

Hello Martin,

thank you again for your reply. Enabling the input hysteresis on the 50Mhz input clock signal seems to have solved my problem!

Best wishes,

Stanzel