| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
All, First let me say Joel's reference to Tyco's improved launch SMA is not the line card used on the old XAUI test set up, but the new line card design that was introduced earlier this year. A couple of people have contacted me already on this issue, so I wanted to make sure all are aware. For those chip vendors who don't already have any new data with the improved launch please contact me.
Next, I still see people getting hung up on terminology. When it is said that everything to the left of TP1 is the transmitter and everything to the right of TP4 is the receiver, this is terminology based on the agreed upon diagram we are using as a reference. If we do not use this diagram reference, then saying these things are part of the channel is correct. Going back to the diagram reference, if we put these effects in the channel, i.e between tp1 and tp4, we will not get good measurements for these points.
Next, Richard's presentation is a step in the direction that I propose. He has begun to develop a model for this section of the channel (not using the reference diagram), hence our accounting for the AC coupling cap has begun. I propose Richard's work be built on, and that the signaling adhoc include this model in their methodology, i.e. hook up this block between the channel and the receiver device. That way all simulations will account for this area with the same model.
So this will allow testing and comparison to the specification. One could test a device for compliance on an evm directly via the sma, or one could test at the ball level and then cascase it with the model.
Richard, the only things that I think need further inclusion is some trace length between the via on the cap and the via for the BGA. Also perhaps some distance before the first ac cap via near TP4.
Also, one thing that I will point out is that we have no channel data if we have to include the AC cap between tp1 and tp4. This would become a serious schedule issue, as new line cards would have to be built, but not before agreement on what the structure looks like. If this course of action were suggested, I would propose the same thing at that point, i.e. develop a model to cascade with measurements between tp1 / tp4.
john
-----Original
Message-----
I think after thinking about this for while I could support an AC cap after TP4 if the 'clean launch" concept is enforced. See attached slides. I did the modeling, sims, and ppt real quick, so pardon my typos and brevity, ... Richard Mellitz, Intel
From:
owner-stds-802-3-blade@listserv.ieee.org
[mailto:owner-stds-802-3-blade@listserv.ieee.org] On Behalf Of Joel Goergen
My apologies to
all. I need to cancel tomorrows call. I will attempt to discuss
across the web. |