Re: [SI-LIST] : 2.5 Gbps across a backplane

About this list Date view Thread view Subject view Author view

From: Jeff Cain (jcain@cisco.com)
Date: Mon Oct 16 2000 - 13:35:50 PDT


Ed, I know you are familiar with some of the work we are doing here, so see
comments inline.

"Dr. Edward P. Sayre" wrote:

> Folks:
>
> A subtle issue came up in the recent T11 (Fibre Channel) meetings. How do
> you spec an interconnect when adaptive equalization parts will be used to
> dig out the signal?

My preference is frequency domain attenuation (see below)

> If you have really good eye-diagrams response then do
> you need equalization?

No

> Conversely, if you need equalization because of
> excessive deterministic jitter (read that insufficient rise time bandwidth)
> or eye closure (read that too much attenuation) or both, then what
> constitutes an acceptable way to specify the interconnect, be it backplane,
> cable or connector?

If you don't have programmable settings of the equalization levels, then all
the links in a connector/backplane channel would have to be of similar
length/attenuation. My personal preference is to specify the frequency domain
loss characteristics of the cable/connector and then I'll figure out how much
equalization I need to recover the signal.

>
>
> Now, the question of cross talk corruption of the signals is another
> matter, especially when equalization is involved. Cross talk can be of no
> consequence or it can be of terrible consequence, especially if there is a
> requirement for byte (a 4 bit nibble) aligned serial data streams. Our
> work indicates that connector cross talk in differential connectors appears
> to be essentially dV/dt noise due to any asymmetries in the geometry of the
> connector or interconnect (vias count here!). Although small, and
> generally in the (1 - 4%) range, it is obviously a problem when many lines
> are switching and especially if the time alignment is not within a small %
> of the rise time between individual differential components. If the signal
> is broken into its even (common mode) and odd (differential) mode
> components, it can be shown that very soon the common mode cross talk
> dominates the cross talk.
>
> So, when is cross talk an equalization issue from a differential eye
> diagram point of view; and can equalization help from an EMI point of view?

We were able to measure xtalk as a function of frequency for some links (the
transfer function actually) and then 'add' it the transfer function of the
channel. The xtalk shows up as a 'bulge' in the eye. No amount of equalization
(at least for the 2 tap system we are using today) helped lower the size of the
'bulge'.

>
> I look forward to folks' ideas on this matter.
>
> ed sayre
>
> +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
> | NORTH EAST SYSTEMS ASSOCIATES, INC. |
> | ------------------------------------- |
> | "High Performance Engineering & Design" |
> +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
> | Dr. Ed Sayre e-mail: esayre@nesa.com |
> | NESA, Inc. http://www.nesa.com/ |
> | 636 Great Road Tel +1.978.897-8787 |
> | Stow, MA 01775 USA Fax +1.978.897-5359 |
> +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
>
> **** To unsubscribe from si-list or si-list-digest: send e-mail to
> majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
> ****

--
Jeff Cain
HW Engr. Mgr.
Cisco Systems
170 W Tasman SJ-G2
Santa Clara, CA 95134
408-527-7754

**** To unsubscribe from si-list or si-list-digest: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu ****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:29:46 PDT