Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: PAM-5 4-WDM deskewing




Rich,

I do not need the 8b/10b coder. I do not want
either to be forced to add a 3.125 GHz clock
to my proposal, because it would make the
package of the electrical transceiver, the
manufacturing of the PCBs and Copper backplanes
and the compliance with FCC EMI requirements
more expensive. The maximum clock in my 10 GbE
proposal is only 1.25 GHz and the maximum
symbol rate on the Copper traces and in the
fiber is only 1.25 Gbaud.

The architecture of my 10 GbE system is
identical to the one described in:

   "10Gig MII update"
   by Howard Frazier
   Kauai, Nov 99

This architecture has four independent
lanes, each one consisting of one PCS and
SERDES, running in parallel.

In my proposal the PCS is the 1000BASE-T
PCS, clocked at 312.5 MHz. This PCS outputs
four PAM-5 symbols every 3.2 nsec.

The SERDES takes the four PAM-5 symbols
generated by the 1000BASE-T PCS and delivers
them serially to the PMD using a 1.25 GHz clock.

The symbol rate in the fiber and in the PMD
is 1.25 Gbaud.

Byte synchronization at the receiver is
performed using the IDLE properties of the
1000BASE-T. Lane alignment is trivial once
you performed byte sync on the individual lanes.

I do not need equalization since I use the low symbol
rate of  1.25 Gbaud and the eye is wide open at the
input of the receiver.

This architecture is a piece of cake. Literally.

There is no need to complicate it  by adding the 8b/10b
coder and a 3.125 GHz clock. They are completely
unnecessary.

Jaime

Jaime E. Kardontchik
Micro Linear
San Jose, CA 95131
email: kardontchik.jaime@xxxxxxxxxxx


Rich Taborek wrote on March 18, 2000:

> Jaime,
>
> In the original note below you say that you manage deskew in "exactly the same
> way as its sister proposal 8b/10b 4-WDM at 3.125 Gbaud". The latest update to
> the WWDM proposal presented in Albuquerque by Mr. Del Hanson:
> http://grouper.ieee.org/groups/802/3/ae/public/mar00/hanson_1_0300.pdf, slide 9
> ,shows that deskew is handled by "Hari". Hari has very recently been renamed as
> the XAUI/XGXS. Therefore, my conclusion is that your PAM-5 4-WDM uses XAUI/XGXS
> for deskew.
>
> I'm trying to figure out how many interface proposals for attaching to the
> Reconciliation Sublayer we're up to.
>
> If your presentation uses a totally different interface back to the RS, then
> just say do. Don't claim that it's the same when it's different.
>
> Best Regards,
> Rich
>
> --
>
> Jaime Kardontchik wrote:
> >
> > Hello 10G'ers,
> >
> > At the end of my presentation in Albuquerque,
> >
> >    "850nm-4WDM-1.25Gbaud transceiver
> >     over Multimode Fiber for 10 GbE"
> >
> > one of the members in the audience asked me
> > about how we manage the deskewing since the
> > group velocities of the four streams of data
> > are different for different wavelengths.
> >
> > PAM-5 4-WDM at 1.25 Gbaud manages the deskewing
> > in exactly the same way as its sister proposal
> > 8b/10b 4-WDM at 3.125 Gbaud.
> >
> > In fact, PAM-5 4-WDM at 1.25 Gbaud shares a lot
> > of common with 8b/10b 4-WDM at 3.125 Gbaud,
> > because both are 4-WDM systems.
> >
> > Hence, encoding, serializing, deserializing
> > and decoding of both systems is done exactly in
> > the same way, as described in the presentation:
> >
> >    "10Gig MII update"
> >    by Howard Frazier
> >    Kauai, Nov 99
> >
> > the only differences being:
> >
> > 1) one uses the 1000BASE-X PCS (8b/10b)
> > and the other uses the 1000BASE-T PCS
> > (scrambling, convolutional and PAM-5 encoding).
> >
> > 2) one uses a SERDES at 3.125 GHz and the
> > other uses a SERDES at 1.25 GHz
> >
> > 3) one uses a baud rate of 3.125 Gbaud
> > in the fiber and the other uses a baud rate
> > of 1.25 Gbaud in the fiber.
> >
> > 4) one uses the K-R characters of the 8b/10b
> > to perform byte synchronization at the receiver
> > and the other uses the IDLE properties of the
> > 1000BASE-T PCS to do the same task.
> >
> > Deskewing in PAM-5 4-WDM is handled in exactly
> > the same way as deskewing in 8b/10b 4-WDM,
> > by realigning the octets of data of the four lanes
> > after decoding in the receiver, using the
> > low frequency 312.5 MHz clock.
> >
> > Also Open Fiber Control is dealt identically
> > in both systems, since both are 4-WDM.
> >
> > Jaime
> >
> > Jaime E. Kardontchik
> > Micro Linear
> > San Jose, CA 95131
> > email: kardontchik.jaime@xxxxxxxxxxx
>
> -------------------------------------------------------
> Richard Taborek Sr.                 Phone: 408-845-6102
> Chief Technology Officer             Cell: 408-832-3957
> nSerial Corporation                   Fax: 408-845-6114
> 2500-5 Augustine Dr.        mailto:rtaborek@xxxxxxxxxxx
> Santa Clara, CA 95054            http://www.nSerial.com