DP83TC811R-Q1: Unstable PHY Link

rif

Part Number: DP83TC811R-Q1
Other Parts Discussed in Thread: DP83TC811

Hello,

We have two DP83TC811R-Q1 PHYs connected between a switch and processor. These PHYs have a transformer-less connection and are on the same board.

The processor is running linux. On board power up if i query the link status using ethtool it says link not detected. All the status LEDs on both PHYs are OFF.

If i take a long piece of wire or an oscilloscope probe and touch C201 (or any other capacitor on the link) the link state immediately goes to link detected and the

LEDs light up. Sometimes this is not required on power up and the link automatically goes up. The behavior is therefore unpredictable.

However, once the phys startup no issues are observed and the connection is stable for continuous operation.

Has anyone faced this issue before ?

The circuit and the PCB routing is shown below. The traces are differential with 100 Ohm impedance. The left PHY is the switch phy and the right one is connected to the processor.

Thanks & Regards

over 4 years ago

0 rif over 4 years ago

Prodigy 50 points

Updating images

0 Justin Lazaruk over 4 years ago in reply to rif

TI__Expert 7040 points

Hello,

The DP83TC811 has not been tested using capactive coupling, since the IEEE 802.3bw standard includes the CMC specification as part of the channel specification.

However, the schematic you've included also needs to be corrected. R368 & R369 should be 1kOhm, R370 should be 100kOhm, and C203 should be 4700pF. The net between R368 and C203 should be shorted to the net between R369 and R370.

REgards,

Justin

0 rif over 4 years ago in reply to Justin Lazaruk

Prodigy 50 points

Hi Justin,

We have fixed the schematic related issues which includes the component values and the link. The issue still occurs.

We had asked this info to ti long time back. At that time the topic of cmc did not come up as recommended in design.

PHY to PHY connection without CMC for shorter inter board communication

Regards

0 Justin Lazaruk over 4 years ago in reply to rif

TI__Expert 7040 points

Hello,

I understand. The problem may be related to the very short trace lengths between the two PHYs. Generally, the PHY expects to drive a longer cable so short traces can results in errors. Can you add additional parasitic capacitance to the lines?

Regards,
Justin

0 Justin Lazaruk over 4 years ago in reply to Justin Lazaruk

TI__Expert 7040 points

Hello,

For short cable links, please try to following registers to help establish a stable link.

0x0479 = 0x0442

0x0466 = 0x8000

0x0107 = 0x2605

0x0106 = 0xb8bb

Regards,

Justin

0 rif over 4 years ago in reply to Justin Lazaruk

Prodigy 50 points

Hi Justin,

We cannot add extra capacitors as there is no board space or pads.

I shall however try out the register settings and update you on the outcome.

Regards

0 Justin Lazaruk over 4 years ago in reply to rif

TI__Expert 7040 points

Hello,

I understand. Please let me if issue is resolved through register settings above.

Regards,
Justin

0 rif over 4 years ago in reply to rif

Prodigy 50 points

Hi Justin,

I changed the register settings but the issue still occurred. So i reverted back to the old configuration. We did observe one thing though, the issue was happening when the board

power supply was disconnected and reconnected. The issue was not happening on a soft-reset / hard reset using reset button. We therefore figured that the issue was in the DC-DC

converter design and increased the soft start time of the converter and added a delay via code to power up the PHY first then the host after 100ms.

With this fix we have not observed the issue and the link goes up on every kind of power cycle soft/hard resets or power disconnect..

Regards

- Rishi

Interface

Interface forum

DP83TC811R-Q1: Unstable PHY Link