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

Re: [8023-10GEPON] Revised 10G budget



Dear Dr. Effenberger,

I am afraid I have a little different understanding about FEC.
For the MAC chip suppliers, FEC code selection seems urgent, 
which directly affects the chip design, size and power.

Even with a small sensitivity difference up to 1dB between 
RS(255,239) and RS(255,223), it is still critical for the 
crucial PR30-Upstream power budget.  Otherwise, there could be 
no E-FEC discussion at all.  
If FEC code is different, budget baseline or 'global average' 
should also be different.

At least, we definitely need to choose FEC code, whether 
RS(255,239) or RS(255,223).

Best regards,
Hiroshi Hamano

%% Frank Effenberger <feffenberger@HUAWEI.COM>
%% Re: [8023-10GEPON] Revised 10G budget
%% Thu, 18 Oct 2007 10:42:48 -0400

> Dear All, 
> 
> Regarding the FEC selection - I agree that the FEC aspect is more of a
> target than a fixed value.  However, we also can consider that the bare Rx
> sensitivity is a 'soft value'.  It depends on a wide range of practical
> concerns, such as yield, cost-effectiveness, technical innovation, etc. So,
> in the end, all of these numbers are based on our best judgment at the
> present time, integrating over all the variables in our minds.  
> 
> So, for the purposes of this baseline, I'd like the group to consider these
> values as a 'global average,' and that they don't individually isolate each
> element of technical risk.  Instead, by averaging, we can reduce the total
> risk since it is unlikely that all the risks will line up against us.  
> 
> Also keep in mind that we can change them in future, if we really come up
> against a hard problem.  The purpose of a baseline is to have a default
> answer that everybody can really concentrate on, and effectively 'try to
> shoot down'.  If it survives the onslaught, well then it must be good.  
> 
> For example, once we get this optical baseline set, the group's work will
> turn towards the study of FEC codes and their realistic *optical* gain.
> That will take some time, and in the end we will get our answer of how many
> dB's we get.  We will likely then need to come back to the optical budget
> and make some small adjustments based on our better understanding.  
> 
> On the issue of the PR20-D, I raised the maximum output power of the Tx
> primarily to make the overload of the PR20 and PR10 receivers line up.  The
> fact that this loosens the Tx spec seems harmless.  If manufacturers want to
> make their Tx with better controls, then that is fine.  But, for purposes of
> specification, this set of numbers just comes out with fewer 'loose ends'.  
> 
> If folks want to change that back to a 3dB range, I would not object too
> much.  However, (in the spreadsheet) we will have to modify the Rx overload
> to maintain commonality between the PR10 and PR20 classes of ONU (which I do
> want to maintain).  Alternatively, we can decrease the minimum path loss in
> the PR20 case by 1dB, which would also bring things in line.  
> 
> Sincerely,
> Frank Effenberger
> 
> 
> -----Original Message-----
> From: Motoyuki TAKIZAWA [mailto:mtaki@ACCESS.FUJITSU.COM] 
> Sent: Thursday, October 18, 2007 7:15 AM
> To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
> Subject: Re: [8023-10GEPON] Revised 10G budget
> 
> Dear Frank,
> 
> Thank you for the modification.
> On the whole, it seems very sensible proposal except for some
> items such as FEC selection etc.
> Some questions on the PR20-D;
> 
> Could you explain why you've widened the launch power range from
> 3dB to 4dB only for PR20-D?
> I just guess the reason like the following assumption.
> - Margin
> If we assume a cooled EML, 3dB would be safe enough.
> Or maybe is it a margin for additional SOA variation?
> - uncooled EML
> Since EML+AMP is supposed to be the PR20-D basic transmitter,
> EDFA output stabilization feedback can be utilized to reduce
> variation range.
> It is not clear whether 4dB range is sufficient or not for
> an uncooled-EML with SOA.  And I am not sure the combination
> of cooled SOA and uncooled EML makes sense.
> 
> Having 4dB range will be more comfortable but if there's no
> special intention for that, 3dB range in line with other classes
> may make sense.
> 
> 
> Best regards,
> Motoyuki Takizawa
> 
> 
> On Mon, 8 Oct 2007 16:47:43 -0400
> Frank Effenberger <feffenberger@HUAWEI.COM> wrote:
> 
> > Dear All, 
> > 
> >  
> > 
> > I have taken some of the comments I received on my last Email, and
> modified
> > the slides to come to the attached version.  
> > 
> >  
> > 
> > What I¨ve done is: 
> > 
> > 1.	Reduce the PR10 OLT downstream transmitter Max and Min by 1 dB.
> > (This reverts to the values presented in Takizawa¨s slides in September)
> > 2.	Reduce the PR10 ONU Rx overload number by 1 dB. (Following the Tx
> > change)
> > 3.	Increate the PR20 OLT max power by 1 dB (Makes the Tx range 4 dB,
> > which is more comfortable)
> > 4.	Increase the PR20 ONU Rx overload number by 1 dB.  (Following the Tx
> > change)
> > 
> >  
> > 
> > Taken together, these changes then make the PR10 and PR20 ONUs identical
> in
> > every respect.  One less PMD! 
> > 
> > This is re-capped on the last slide. 
> > 
> >  
> > 
> > Sincerely,
> > 
> > Dr. Frank J. Effenberger      献声針 娃繋鴬鯉
> > 
> > Huawei Technologies USA
> > 
> > 1700 Alma Drive, Plano TX 75075
> > 
> > Office (732) 625 3002
> > 
> > Cell (908) 670 3889
> 
> 




---
---------------------------------------------
HIROSHI HAMANO         Network Systems Labs.
FUJITSU Labs. Ltd., Kawasaki, 211-8588 JAPAN
TEL: +81-44-754-2641  FAX: +81-44-754-2640
E-mail: hhlsi@flab.fujitsu.co.jp
---------------------------------------------