[802.3ae_Serial] 850nm eye mask and draft link model
Tomorrow, we hope to discuss an alternative eye mask for 850 nm serial.
Here is a straw man to focus the discussion in tomorrow's serial PMD call:
Please consider the scaled OC-12 mask which is a little more demanding than
the 10GE mask but not as much as the 1 GE mask.
| X1 | X2 | Y1 |
10GE |0.3 |0.4 |0.25 |
OC-12 |0.25 |0.4 |0.2 |
1GE |0.22 |0.375|0.2 |
The significant difference is that the OC-12 mask is 60% of the height of
the eye instead of 50%. This forces faster risetime and/or lower jitter at
the transmitter, to allow for distortion from the multimode fibre. (The
OC-1/3 mask is intermediate between OC-12 and 1GE.)
I have been updating the link model with a view to assisting the choice of
an eye mask for 850 nm serial. The current file is too big to go through
the reflector. There is a temporary file at
http://www.ieee802.org/3/ae/public/adhoc/serial_pmd/documents/10GPBud_tmp010
622a.xls but it only has a 1310 nm page. I have a draft 850 nm page which I
can mail tomorrow to anyone attending tomorrow's meeting, or one can input
the 850 nm parameters to this temporary file.
Meeting coordinates at the bottom of the message below:
Piers
> -----Original Message-----
> From: DAWE,PIERS (A-England,ex1) [mailto:piers_dawe@xxxxxxxxxxx]
> Sent: 21 June 2001 16:30
> To: '802.3ae Serial'
> Subject: [802.3ae_Serial] From Serial PMD 19 June: 850nm eye mask, ORL
> spec, 1310serial ma x Tx power, ...
>
> Eye mask for 850 nm
> -------------------
> 850 nm proponents, please notice that the serial ad hoc is
> charged with
> producing a proposal for a 850-specific eye mask which would
> preserve the
> transmitter integrity requirements of the current draft
> standard but remove
> the need for rise time specification and measurement. This
> might be like a
> scaled OC-12 or 1GigE mask. Please email and join next week's call if
> interested.
>
> ORL
> ---
> We understand that ORL is total Optical Return Loss measured by
> disconnecting a transmitter from a link and measuring the link with a
> reflectometer. The far end receiver remains attached.
> Historically this
> has been a requirement of SONET but not of Ethernet. It seems that it
> guards against multiple reflection effects (parasitic
> etalons) in single
> mode links, where reflections that are individually within
> spec combine to
> create a problem.
>
> The following is a table (E&OE):
>
> | G.691 | 10GE |
> Discrete | 27 | 27 | 26 | 26 |
> Receiver | 14 | 24 | 12 | 26 |
> ORL | 14 | 24 | - | - |
> RIN test | - | - | 12 | 21 |
>
> If the ORL requirement is there to protect the transmitter from back
> reflection, 10G Ethernet could reasonably use the same
> numbers for ORL spec
> as it uses for RIN test. Input from the fibre and cabling
> community on the
> merit of an ORL spec at 1310 and/or 1550 nm is needed: please
> email or join
> next week's call!
>
> Receiver overload
> -----------------
> Suggested re-calculating the Tx max mean power from current
> likely Tx min.
> mean power + 5 dB for tolerances (keeping the 5 dB range we
> started with).
>
> How is stressed eye ISI value derived?
> --------------------------------------
> Thought to be calculated using a high receiver bandwidth
> rather than the
> usual 7.5 GHz, and with TP3 DCD (6 ps) rather than TP4 (8
> ps). If this is
> right, to be recorded in "Notes" page of link model at next revision.
>
> Jitter
> ------
> 0.015 UI sigma_RJ thought to be too small. Obviously too
> much causes an
> error floor, but maybe 0.025 UI would be in the right ball
> park? But we
> don't know what would cause Tx RJ apart from RIN, because
> most oscillator
> phase noise is expected to be <4 MHz and tracked out by Rx PLL.
>
> We have accounted for some DJ in the model with the DCD
> factor. We are
> still thinking in the range 0.3 to 0.35 UI.
>
> Next meeting
> ------------
> Agenda items to include:
>
> 850 nm mask proposals
> ORL spec
> Receiver overload
> Refining jitter DJ and RJ spec numbers
>
> Usual coordinates:
>
> 15:15 GMT = 4:15 pm BST = 17:15 CET = 11:15 am EDT =
> 8:15 am PDT,
> Tuesday
> +1(816)650-0631 Access code 39209
>
> Piers
>