I have seen presentations from connector vendors showing acceptable results with de-embedded in other venues. Its seemed to me to offer a more agonistic method to compare parts. Perhaps some these folks could
share.
… Richard Mellitz,
Intel Corporation
From: Liav Ben Artsi [mailto:liav@xxxxxxxxxxx]
Sent: Thursday, July 19, 2012 3:04 AM
To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: [802.3_100GCU] Suggested return loss Vs. test fixure
Hi colleagues,
I could not sleep because of too many thoughts about the Tx/Rx return loss (and the jet-lag does not help either
J).
I took the suggested return loss of comments 85/86 and put them against the possible test fixtures I have provided in former E-mails (please see picture) below.
I will be suggesting to keep the return loss as TBD for now and build up a consensus suggestion towards Geneva.
The current remedy for comments 85/86 (the way I see it) is suggesting no de-embedding of the test fixture, but is not aligned with test fixture reflections.
Therefore I suggest building up a consensus regarding the following issues:
1.
Measurement methodology (do we de-embed the test fixture from return loss measurements or not) – De-embedding accuracy and capabilities presentation would be very helpful here in convincing to go one way or the other.
2.
Provide a consensus return loss equation taking into account the decision of point #1.
Please see below the current suggested Tx/Rx return loss (Comment 86/85) (hoping I got it right).
Yellow – note 85/86 equation
Green - -12dB test fixture return loss
Orange – Via with a 4mil stub connected to a u-via
Grey – Short via è trace
è another short via – both have 15mil stub
One can see that both extracted test fixtures are already quite close to the suggested remedy for Tx/Rx return loss (and this takes into account an ideal return loss package – this is just the test fixture return loss…).
Best regards,
Liav