DP83848-EP: Incorrect preamble and not received data

Rick Purcell

Part Number: DP83848-EP

Hi,

I am having ethernet connectivity issues. I have two identically built subsystems, both using the DP83848 for downlinking data. The first links to our Linux server and downlinks data flawlessly. The second unit does not. I have captured a number of scope plots and confirmed both units send data out, however the preamble of the non-working unit looks wrong. The working units preamble is 1010101... (as expected), the non-working units preamble looks like 10110... I'm trying to figure out why the 'bad' unit isn't working.

I have run the BIST on the PHY chip (both internal loop back and external loopback) and both units pass.

My suspicion on this issue is:

1. Master clock into PHY is bad (skew, jitter)

2. Damaged magnetics

Since the BIST passed it appears that everything should work. However seeing the incorrect preamble got me thinking why the preamble is bad.

Any thoughts or ideas on this failure?

Rick

over 7 years ago

0 Wade Vonbergen over 7 years ago

TI__Guru 62930 points

Rick,
I am going to move this to the commercial forum. They will have more insight to potential causes for this behavior.
Thanks,
Wade

0 Rob Rodrigues over 7 years ago in reply to Wade Vonbergen

TI__Mastermind 19780 points

Hi Rick,

Can you provide the PHY section of the schematics as well as the specification for the magnetics? If you could provide scope shots of the pre-amble, those would be helpful.

Best Regards,

0 Rick Purcell over 7 years ago in reply to Rob Rodrigues

Prodigy 40 points

Here is the information see attached pdf.

The magnetics is PN 100B-1003/X specification is also attached.M101 (4).pdf

Rick

schematics and plots.pdf

0 Rick Purcell over 7 years ago in reply to Rick Purcell

Prodigy 40 points

Based on the testing I did today the clock rate looks OK. I don't have direct access to the 25MHz clock but the transmit is 10MHz (as expected). I also verified waveforms between the good and bad unit, they appear the same. I don't suspect the magnetics or clock at this point.

We did a custom build of firmware that send all 0's for ethernet data. The output of the PHY for the bad unit was wrong, the good unit had the expected output. My suspicion is the PHY at this point as the preamble is generated internally. Are there more detailed block diagrams of this device that show how the preamble is generated?

0 Rob Rodrigues over 7 years ago in reply to Rick Purcell

TI__Mastermind 19780 points

Hi Rick,

Moving this conversation to email to expedite the solution.

Best Regards,

Interface

Interface forum

DP83848-EP: Incorrect preamble and not received data