DP83869HM: Jumbo frame in Media converter

Hi Sir,

We got a jumbo frame issue.
The DP83869HM is designed in a media converter (RJ45 to fiber optical) .
We use a PC-A ping a PC-B as below:
PC-A <==> media converter 1 <==> fiber cable <==> media converter 2 <==>PC-B

We added a 1M ohm resistor and a 4700pF capacitor between erth and phy gnd by TI's recommend.
The jumbo frame problem still exists.
Attached file are schematic and register values.
Please help check it again.

Thanks.

1107027300-PCBG-CV1-M4-4B24K-V0-GP2.pdf

Hi Ronald,

Jumbo Frames shall be supported for 100M and 1G speeds. We will be able to review the schematic and provide feedback by Wednesday of this week. 

For now, can we try some loopback tests to isolate the source of the issue. You may try using PC-A to generate packets to Media-Convertor 1, and using an external loopback to try to loop back the Fiber Outputs of Media Convertor 1 to the Fiber Inputs of Media Convertor 1. If no errors are observed, we can conclude that this data path is function. Similarly we can perform an external loopback between Media Convertor 2 and PC-B. This shall confirm the data paths of each individual Media Convertor is functional. 

Thank you,

Nikhil

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and it is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml).

Hi Nikhil,

I use 2*DP83869EVM to do jumbo frames test for 100M and 1G  speeds.

The setup is the following.
100 Mbps Media Converter:

Test result: Jumbo frames Fail

1G bps Media Converter:

Test result: Jumbo frames PASS

Could you help to confirm if DP83869HM can support Jumbo Frames for 100Mbps speeds ?

Thank you,

Ronald

Hi Ronald,

DP83869 supports jumbo frames for 1G and 100M speeds. It looks like between the two setups, only the SFP connector changed. Is it possible to use the 1.25G SFP for the 100M case? If you use the 1.25G SFP for the 100M case do you still see failures?

Thank you,

Nikhil

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and it is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml).

Hi Nikhil,

I use 1.25G SFP Fiber Transceiver on 100M case, the jumbo frames test still failures.

Test result: jumbo frames Fail

Do you have any advice for jumbo frames issue on 100M case ?

Thank you,

Ronald

Hi Ronald,

For the 100M case we may try some loopback tests to help identify the interface causing the issue. Is it possible to try an external loopback on the fiber side of EVM1? This would be connecting the TX to RX on the fiber side of EVM1, so the configuration would be PC-A <-> EVM 1 <-> PC-A. Does PC-A observe errors in this case?

Thank you,

Nikhil

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and it is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml).

Hi Nikhil

As far as I know,it need two PC to do jumbo frames test.When I use a PC to ping its own IP without any connection, it still able to ping itself.So I think loopback test is not help for jumbo frames test.I'm not sure if what I said is correct, please let me know if it is wrong.

Thank you,

Ronald

Hi Ronald,

Given your block diagram, it looks like the end to end communication is between two PC's. I would assume one PC is generating packets to be received by other and vice versa. My understanding is this can be "simulated" by doing a loopback test in between the two PC's, sending the jumbo frames sent from one PC through one PHY and back to itself. Can this experiment be tried as described in my previous post?

Thank you,

Nikhil

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and it is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml).

Hi Nikhil,

I have no idea how to use one PC to do jumbo frames test with loopback. Do you recommend any testing tool ?In addition ,when I do the jumbo frames test on 100M case,the maximum packet can only be set to 1970.If the packet exceeds 1970,it will fail. Is it correct?

Thank you,

Ronald

Hi Ronald,

The PC must have the capability to both send and receive packets as you are showing in your screenshot. For EVM 1, can you connect the cable in such a way that the RX terminals are connected to the TX terminals of the EVM, thus looping back the packets sent out from EVM1 back to EVM so they get sent back to PC A? The PC should be able to count packets sent and received, and then you can at least check to see if EVM 1 is functioning as expected.

Thank you,

Nikhil

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and it is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml).

HI Nikhil

Thanks for your suggestion, we will try to setup the test environment for the platform certification.

But I have some questions hope to get your reply.

1.How big MTU size  is the 100Mbps media converter mode  In your internal test?

2. Which register can observe the jumbo frame function enabled?

3. The jumbo frame issue of 100Mbps media converter mode is really interfered us, or 1518bytes is limited on 100Mbps using.

    If the 100Mbps media converter mode wants to support up 9k bytes, the MAC is necessary?

    But 1000Mbps media converter is no more probole on over 10k bytes frame applying.

    So, I am really confused.

Thanks for your review again.

Arthur

Hi Arthur,

To answer your questions:

1. What does MTU refer to? I'm not familiar with this term?

2. There is no register for jumbo frame function enable. Jumbo Frame support is automatically and always enabled.

3. I'm not quite following this question. The inter-packet gap for 100Mbps is standard defined by IEEE, along with jumbo frame support. 

Have you had any success setting up any of the loopback tests?

Thank you,

Nikhil

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and it is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml).

Hello Arthur

Do you have some update.

Regards,

Sreenivasa

Hello Arthur

Do you have some update.

Regards,

Sreenivasa

Hello Arthur

Thank you for the message. 

Regards,

Sreenivasa