Recently ADI was kind enough to allow the Clockworks A2B® module to be placed on the same test fixture as used in developing the guidance contained in ADI's EE-369: Eye Diagram Measurement for AD242x A2B® Transceiver-Based Systems. (note this document is only available to ADI A2B NDA holders)

The A2B module was tested by sending data to the ADI test fixture with both short and long cables and in general the results were good.  There was one apparent pattern dependent mask error noted; the violation was less then a few percent and occurred at a rate of about 1 in 2000 bits. No actual receive errors were noted and the rate happened regardless of cable length.

It's more or less impossible to see in the eye diagram, there's 4 red pixels on the far right side of the eye, literally on the edge of the mask.  This same type of violation occurred on two different production boards, and it seemed to be independent of cable length. What was interesting was even a rev 1 prototype, which had a far from ideal layout, showed about the same pattern of eye diagram violations.  No actual bit errors have ever been observed using the AD2428W built in bit error test features.

We plan to return again in the future to analyze this issue further to see what might be the underlying cause of the observed eye diagram.  It does not seem to have any impact on operation.

Anonymous