Re: Link Status thoughts
Ed and Ben,
I think that the bias is necessary for interfaces like XAUI, where we
want to support a pluggable interface. If the transceiver ain't plugged
in, you have to have a bias or the receiver inputs at the XGXS will chatter.
The reason we couldn't rely on a flat line condition with 1000BASE-X
was because the PMD outputs weren't required to squelch. The crosstalk
problem came from the chattering at the PMD receive outputs in the
absence of sufficient optical input signal. Note that many designers
of GBICs figured this out eventually, and added the squelch.
For the non-pluggable serial receivers, I can live with squelch alone,
without the bias on the serial receive inputs, but for the XGXS,
the bias will be needed. Yeah, this decreases the sensitivity, and
I think that the trade off has to be made in favor of a higher transmit
amplitude. More power, yes, but also better signal integrity.
Howard
Ed Grivna wrote:
>
> Hi Howard,
>
> part of what you suggest here makes sense, but I dissagree with
> a different part.
>
> The part I don't like is the idea that the PCS receive should
> have a bias to prevent chatter in an AC-coupled connection.
> I don't particularly like offset-biased differential connections
> because they cause duty cycle distortion of the signal.
>
> I think you suggested the best solution in your first paragraph;
> i.e., that being to have the PCS receive also have a signal
> detect (peak differential-amplitude detector) and to squelch the
> line receiver if there is insufficient amplitude.
>
> Regards,
>
> Ed Grivna
> Cypress Semiconductor
>
> >
> > General comment:
> >
> > Pins = Bad
> >
> > I believe that the limiting amp stage should have a signal detect
> > function built into it, and should squelch its output when the
> > incoming signal is below the noise floor.
> >
> > For AC coupled systems, this implies the addition of a bias to
> > the PCS receive input so that it doesn't chatter when the
> > PMD receive output is squelched. The "flat line" state observed
> > by the PCS receiver will be a sufficient indication of a dead link.
> >
> > We can't rely on MDIO for signal detect reporting between the
> > PMD and the PCS. We can rely on MDIO for Link Status (a superset
> > of signal detect) reporting between the PCS and Station Management.
> >
> > Howard Frazier
> > DomiNet Systems, Inc.
> >
> > Ben Brown wrote:
> > >
> > > Hey,
> > >
> > > After a discussion some of us had during the editorial
> > > meeting yesterday, I thought I'd put some of our thoughts
> > > down on paper. There are still quite a few questions
> > > buried in these slides. Anyone with comments, please
> > > feel free to share them. I'd like to have these comments
> > > before the November meeting so we can put a proposal on
> > > the table.
> > >
> > > Thanks,
> > > Ben
> > >
> > > --
> > > -----------------------------------------
> > > Benjamin Brown
> > > AMCC
> > > 2 Commerce Park West
> > > Suite 104
> > > Bedford NH 03110
> > > 603-641-9837 - Work
> > > 603-491-0296 - Cell
> > > 603-647-2291 - Fax
> > > 603-798-4115 - Home Office
> > > bbrown@xxxxxxxx
> > > -----------------------------------------
> > >
> > >