• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 34 subscribers
  • Views 825 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DP83865DVH PHY problem on 2 of 4 boards.

Mark Green
Intellectual 660 points
Other Parts Discussed in Thread: DP83865

Hi,

I'm having trouble getting the PHY on the DP83865 to connect to a Gb switch.

I have 3 prototype boards say A, B & C designed for RGMII operation.

A & B the PHY connects fine. PHY on C attempts to connect then fails (but only in RGMII mode!)

This is very repeatable and the same cable and switch port is used on each board in turn, so I'm sure it is a problem with the PCB setup.

.

I'm using the DP83865 in the following configuration:

FPGA - DP83865 - termination & magnetics & RJ45. Then 1m of cat5e into a  Gb switch.

For these tests I'm tapping into the DP83865 Management Debug interface using an off board mcu.

I'm only interested in getting the PHY to connect at this stage, so the FPGA can be ignored.

.

If the straps RGMII_SEL0 and SEL1 are pulled high (i.e. RGMII mode) and a reset applied, then the PHY on boards A & B connect.C repeatedly attempts then fails (switch box lights for ~1s then out for 1s). After a while on board C the 5 status LED's on the DP83865 all flash at ~ 4 times per second.

if the straps RGMII_SEL0 and SEL1 are pulled low (i.e. GMII mode) and a reset applied, then the PHY on all 3 boards connect at 1Gb/s full duplex. 

I am determining connection by the light on Gb switch box lighting continuously and also the value of LINK_AN (addr 0x11) bits 4:1.

.

This is where its gets a bit stranger...

After establishing a good link:

Board A shows 0x0416 in LINK_AN (i.e. MDI cross over mode) in either RGMII or GMII modes

Board C shows 0x0016 in LINK_AN (i.e. MDI straight mode)in GMII mode (PHY won't connect in RGMII mode)

Using the same cable!

.

So my main problems are:

1. Why do the strapping values for RGMII (pins 40, 60) have an effect on the PHY connecting or not? (I was surprised by this since I thought this would only impact the MAC interface.)

2. what is the significance of all 5 status LED's flashing rapidly (I can't see this documented anywhere).

3. After connecting board A shows a cross over on the MDI and board C no crossover on the MDI. Yet it is the same PCB artwork, and the same Cat5e cable and switch  port in both tests. Any ideas?

.

Any help would be much appreciated!!

.

Mark.

  • 0
    TI__Mastermind 31600 points

    Mark,

    Could you provide your schematics for review?  You can email them to me directly at <email address redacted>.

    Patrick

  • 0 in reply to Mark Green
    TI__Mastermind 31600 points

    Mark,

    For the most part, the schematic looks quite good, but some of the component numbers and values are difficult to decipher in the image provided.  Could you please compare the specifications for the transformer against the recommendations in our DP83865 design guide(http://www.ti.com/litv/pdf/snla056c)?

    We also have a troubleshooting guide (http://www.ti.com/litv/pdf/snla065) that may be useful as we dig into this issue further.  I have shared some thoughts regarding this issue with a design engineer for this part and I am awaiting feedback.

    Patrick

  • 0 in reply to Patrick OFarrell
    Prodigy 50 points
    Hello Mark,
    We see that same problem on our design. Unstable Gbit link on some boards when using RGMII, but stable when using GMII. Have you found any solution to the problem?
    Rgds,
    Morten
  • 0 in reply to Morten Reknes
    Intellectual 660 points
    Hi Morten,

    I don't think I'm going to be much help but gut feel was that it was a bad batch of connectors which was causing damage to the chips.

    I only had a handful of PCB's but these were my findings:

    A couple of bad boards had their DP83856 replaced, and were still bad. Connector was replaced, still bad. Chip replaced again, worked ok. (i.e. only after both the chip and connector had been replaced, and the new chip had not been exposed to the suspect bad connector).

    I was lucky enough to have use of an Agilent Ghz scope with active probe and could see no difference in the signals going into the chip on both good and bad boards and throughout the above process.

    After this I replaced both DP83865 and connector at the same time on our other bad PCB's and they sprang into life straight away.

    The chip was designed out on the next phase (not my call).

    Sorry I can't be more help.

    Mark.
  • 0 in reply to Mark Green
    Prodigy 50 points
    Hi Mark.

    Thanks for your reply and for sharing your findings. Did you use a connector with integrated magnetics?

    Rgds,
    Morten
  • 0 in reply to Morten Reknes
    Intellectual 660 points
    Yes we did.
    I would give the part number but I don't have access to those project documents any more.
    From memory I think I replaced with a BEKO part, which I found to be good.

    Good luck!

    Mark.