DP83867IS: DP83867ISRGZR

JYOTI AGRAWAL

Part Number: DP83867IS

Hi,

We have used TI PHY (DP83867ISRGZR) and other protection circuits in our board to power-up using PoE of 48V through RJ-45 connector.

It is designed to have an option for power-up using both DC supply and PoE with PHY DP83867ISRGZR.

When we were using dc supply to power the board and using normal ethernet, the board was working properly with 1Gbps link speed.

After we started powering-up with PoE, observed failures over time.

Detail testing exercise done is as below:

48V dc power is given at pair C and pair D using PoE adaptor.
Initially it was working properly with successful auto-negotiation completion and link establishment.
When multiple (10 times) plug-in and plug-out of RJ-45 was performed, auto-negotiation has not been able to complete successfully and link didn’t establish.
Then it is observed that there is variation in impedance (measured w.r.t. TPAD pin) at PHY Differential Transmit and Receive Signal Pins (Pin No. 7 and 8, 10 and 11). They are getting low.
We have observed this issue on 4-5 chipsets.

So, Kindly check and provide us the solution.

Please find the detail circuit design that we have used at below link.

/cfs-file/__key/communityserver-discussions-components-files/138/PoE-Problem-Statement.docx

Thanks

Jyoti Agrawal

over 4 years ago

0 Adrian Kam over 4 years ago

TI__Expert 5425 points

Hi Jyoti,

Let me take a look at your design and consult with my team. I will get back to you tomorrow.

After the cable was plugged in and out multiple times and you see the change in impedance, is the PHY still functional? Are you able to get link using the dc power supply again after the device goes through the testing exercise?

Regards,

Adrian Kam

0 JYOTI AGRAWAL over 4 years ago in reply to Adrian Kam

Prodigy 20 points

Hi Adarian,

After the testing exercise PHY is no more functional as impedance of pins are low and also not able establish link using dc supply.

Regards,

Jyoti Agrawal

0 Adrian Kam over 4 years ago in reply to JYOTI AGRAWAL

TI__Expert 5425 points

Hi Jyoti,

For the PoE center tap capacitors, can you try using 10 nF capacitors instead of 1nF?
Can you provide a set of register data for registers 0x00 to 0x1F, 0x2D, 0x2E, and 0x6F?

Regards,

Adrian Kam

0 JYOTI AGRAWAL over 4 years ago in reply to Adrian Kam

Prodigy 20 points

Hi Adrian,

1.I have tried with 10nF Capacitor and below are the results:

Chip-Set	PHY Pins	After rework and before testing (Cap. C418, C420, C422, C424, C426 are changed to 10nF)	After rework and testing (Cap. C418, C420, C422, C424, C426 are changed to 10nF)	Conclusion
3	Pair C : Pin 7 and 8 Pair D : Pin 1 and 2 Pair C : Pin 4 and 5 Pair D : Pin 10 and 11	Pair A: 2.08K Pair B: 2.08K Pair C: 188E instead of 2K Pair D: 1.18K instead of 2K	Pair A: 2.1K Pair B: 5E instead of 2.08K Pair C: 181E instead of 2K Pair D: 1.18K instead of 2K	After multiple (10 times) plug-in and plug-out of RJ-45 was performed Impedance of Pair-B has gone low.

2. Please find attached document for register dump.

/cfs-file/__key/communityserver-components multipleuploadfilemanager/02623835_2D00_46d9_2D00_4be5_2D00_af4b_2D00_5454f1a6c307-389832-complete/Phy_5F00_register_5F00_read_5F00_dump_5F00_e2e_5F00_TI.xlsx

Regards,

Jyoti Agrawal

0 Adrian Kam over 4 years ago in reply to JYOTI AGRAWAL

TI__Expert 5425 points

Hi Jyoti,

Let me take a look at everything you sent me and consult with my team. I will get back to you by Wednesday at the latest.

Regards,

Adrian Kam

0 Adrian Kam over 4 years ago in reply to Adrian Kam

TI__Expert 5425 points

Hi Jyoti,

While I'm going through what you sent, I have a few questions:

Do you see the same issue if you turn the PoE power source on/off instead of plugging the cable in and out?
Is plugging the cable in and out with a live 48V a necessary test for your application?

Regards,

Adrian Kam

0 JYOTI AGRAWAL over 4 years ago in reply to Adrian Kam

Prodigy 20 points

Hi Adrian,

Below is answers for your questions:

1. We have not performed the PoE power source ON/OFF instead of plugging the cable in and out, as our testing requirement is cable plug-in and plug-out.

2. Yes, Cable plug-in and out with 48V is required test for our design..

So. If there is any limitation of PHY(DP83867ISRGZR) for cable plug-in and out with live 48V is there any other PHY switch available that will work fine in required test case.

Regards

Jyoti Agrawal

0 Adrian Kam over 4 years ago in reply to JYOTI AGRAWAL

TI__Expert 5425 points

Hi Jyoti,

Would you be able to perform the PoE power source ON/OFF and see if you get the same issue?

Regards,

Adrian Kam

0 Adrian Kam over 4 years ago in reply to Adrian Kam

TI__Expert 5425 points

Hi Jyoti,

After consulting with my team, it seems the ESD diode array could be the issue. The 1A clamp voltage is typically the energy the PHY will see during the insertion transient. The diode array you are using is a bit higher than 6.6V. I would suggest trying a different part that has a lower 1A clamp voltage like the one linked below, if that is possible.

https://semtech.my.salesforce.com/sfc/p/#E0000000JelG/a/44000000MCod/Fs.eSBQs1U.304Nkc5rxRZ3Lk6_1A4.ZX_npjU1R.C8

In addition, I can suggest adding 5 ohm series resistor to all MDI pairs to slow the current drain through the magnetics ground. It may have an impact on differential voltage output and cable reach, but it should help with cable removal transient events. There is an article from Microchip that covers this information, and I have linked it below for your reference.

http://ww1.microchip.com/downloads/en/Appnotes/00002157B.pdf

Regards,

Adrian Kam

Interface

Interface forum

DP83867IS: DP83867ISRGZR

Chip-Set

PHY Pins

After rework and before testing (Cap. C418, C420, C422, C424, C426 are changed to 10nF)

After rework and testing (Cap. C418, C420, C422, C424, C426 are changed to 10nF)

Conclusion