[FAQ] DP83867IR: Packet Loss and CRC Errors with DP83867 using Short Cables

Expert 8920 points

Replies: 2

Views: 95

Part Number: DP83867IR

Hi All,

If you are encountering an issue with packet loss or CRC errors while using the DP83867, please consider some of these items for debug when using short cables.

Short cables at 1m or less in length for your device may experience signal quality issues. One reason could be that the digital signal processing internally may take too long to converge or may even converge incorrectly at shorter lengths which could result to a bad SNR - Signal to Noise Ratio. This then creates link dropping or potential packet losses which may require you to reset your device before beginning packet transfer again. 

We have a script below that will be updated in the next revision of the DP83867 datasheet that should help eliminate any packet loss errors due to short cabling. This scripts allows for a change in the timing bandwidths to ensure the DSP converges correctly. 

ADDR

DATA

Description

0x001F

0x8000

Hard Reset

0x0053

0x2054

Threshold for consecutive amount of Idle symbols for Viterbi Idle detector to assert Idle Mode set to 5

0x00EF

0x3840

CAGC DC Compensation Disable

0x0102

0x7477

Master Training Timers - increasing time in different training states

0x0103

0x7777

Master Training Timers - increasing time in different training states

0x0104

0x4577

Master Training Timers - increasing time in different training states

0x010C

0x7777

Timing Loop Bandwidth

0x01C2

0x7FDE

Timing Loop Bandwidth

0x0115

0x5555

Slave Timers - increasing time in different training states

0x0118

0x0771

Slave Timers - increasing time in different training states

0x011D

0x6DB2

Timing Loop Bandwidth

0x011E

0x3FFB

Timing Loop Bandwidth

0x01C3

0xFFC6

Timing Loop Bandwidth

0x01C4

0x0FC2

Timing Loop Bandwidth

0x01C5

0x0FF0

Timing Loop Bandwidth

0x012C

0x0E81

FFE Fix

0x001F

0x4000

Soft Reset

If you have further questions please start a new E2E thread and our Ethernet team can clarify these items further for you.

Thanks,

Cecilia

2 Replies

  • Hi Cecilia, 

    I am the engineer who raised the above issue.

    Thank you for organizing the above issues in a credible form.

    I have an additional question, is the SNR issue for 1m applied equally to all communication frequency bands?

    Or could the length vary depending on the frequency?

    I ask a question because the wave length will vary depending on the frequency and it seems to be related to the part where the SNR becomes weak.

  • In reply to seokyeon kim1:

    Hello,

    Thanks for reading! Let's discuss on a separate thread as this seems to be an issue we can debug further.

    Thanks,

    Cecilia