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

Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference 12/18/2006




Here are the updated tables; the 2x50G DML row (not a possible
alternative) is replaced with the 1x100G TDM alternative.


-----Original Message-----
From: Dove, Dan [mailto:dan.dove@HP.COM] 
Sent: Wednesday, January 17, 2007 4:21 PM
To: STDS-802-3-HSSG@listserv.ieee.org
Subject: Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference
12/18/2006

Folks,

It sounds like an additional row needs to be added to the table.

Also, I would like to propose that future presentations highlight which
table elements their proposal address just for clarity's sake.

It occurs to me that as time goes on, it would be nice to have the table
show the number of technical and economic feasibility presentations have
been made for each row.

Just an idea.

Dan 


------------ Previous Message Below ------------

-----Original Message-----
From: Ali Ghiasi [mailto:aghiasi@BROADCOM.COM] 
Sent: Tuesday, January 16, 2007 3:01 PM
To: STDS-802-3-HSSG@listserv.ieee.org
Subject: Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference
12/18/2006

Rami

I do want to 2nd your proposal to add External Cavity DML to the list of
potential solutions. As you stated ECL can operate at 10Gb/s over 120 Km
now. With adiabatic chirp negligible and dynamic chirp low enough to
operate over 120 Km at 10Gb/s this makes them excellent candidate for
the 5x20 or 4x25 Gb/s per lane.

Thanks,
Ali

Rami Kanama wrote:
>
> Ali,
>
> Since last year's OFC, there have been other successful EDC 
> demonstrations extending the reach of an 80km 10Gbps 1550nm ECL to
120km.
>
> Standard DFB DML are dominated by transient or dynamic chirp as well 
> as adiabatic chirp, however in external cavity (ECL) DML, the 
> adiabatic chirp is negligible and the transient chirp has both 
> negative and positive components which can be optimized to extend the 
> reach over fiber. In addition to that, due to the round trip of 
> external cavity resonance, the modulation bandwidth can be up to 20GHz

> or more.
>
> For this reason, 1550nm cooled external cavity based DML should be 
> added to the list of potential solutions for 20-25Gb/s per lane.
>
> Regards,
>
> Rami Kanama
>
> Redfern Integrated Optics (RIO), Inc.
>
> Mobile: (408) 386-3191
>
> -----Original Message-----
> *From:* Ali Ghiasi [mailto:aghiasi@BROADCOM.COM]
> *Sent:* Monday, January 15, 2007 10:50 AM
> *To:* STDS-802-3-HSSG@listserv.ieee.org
> *Subject:* Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference
> 12/18/2006
>
> Frank
>
> For the last year OFC we had a demo using and external cavity DML 
> operating over 80 Km in a SFP+ like form factor, the device had very 
> little adiabatic chirp and the chirp was dominated by dynamic chirp. 
> Our finding was that EDC had very little benefit in this case outside 
> compensating for the FR4 trace.
>
> Standard DML are dominated by dynamic chirp which EDC will not be op 
> much help.
>
> Thanks,
> Ali
>
> Frank Chang wrote:
>
> One datapoint is current EDC seems working very well for the type of 
> positive chirp coming from short-reach EMLs. I agree its interesting 
> to investigate the impact of DML type of positive chirp.
>
> Obviously the tweaking of EDC circuit should be done with the change 
> in its link/media characteristics. Also EDC could be implemented in 
> several ways such as at the Rx, pre-distortion at the TX or E-field 
> domain. So it should expect to play roles for 20/25G etc resulting 
> from its ease of use and attractive economics.
>
> -Frank
>
> -----Original Message-----
> *From:* Mike Dudek [mailto:mike.dudek@picolight.com]
> *Sent:* Thursday, December 28, 2006 9:54 AM
> *To:* STDS-802-3-HSSG@listserv.ieee.org 
> <mailto:STDS-802-3-HSSG@listserv.ieee.org>
> *Subject:* Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference
> 12/18/2006
>
> All,
>
> The EDC assumptions used for LRM (10G) development was that the 
> dispersion effect is linear. (ie every one or zero is identical and 
> the eye closure is created because the bit energy is not confined 
> within one bit period and smears out over multiple pulses such that 
> with superposition of the bits the eye closes. ie it's linear ISI that

> is being compensated.). It also assumes that the ones and zeros are 
> inverts of each other. I think the problem for 1550nm DML is likely to

> be chirp. Does anyone know if chirp has this deterministic linear 
> characterisitic and whether the ones and zeros are inverts of each 
> other. (I don't think they are).
>
> I'm not suggesting that EDC can't help. I am suggesting that a more 
> complicated version of EDC might be needed than that assumed for LRM, 
> and the benefits achieved might not be as great as initially expected.
>
> Regards,
>
> Mike Dudek
> Director Transceiver Engineering
> Picolight Inc
> 1480 Arthur Avenue
> Louisville
> CO 80027
> Tel 303 530 3189 x7533.
> mike.dudek@picolight.com <mailto:mike.dudek@picolight.com>
>
> ----------------------------------------------------------------------
> --
>
> *From:* Roger Merel [mailto:rmerel@GMAIL.COM]
> *Sent:* Friday, December 22, 2006 1:16 PM
> *To:* STDS-802-3-HSSG@listserv.ieee.org 
> <mailto:STDS-802-3-HSSG@listserv.ieee.org>
> *Subject:* Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference
> 12/18/2006
>
>
> All,
>
> To the extent that dispersion penalties adversely affect the link 
> budget (even if the link works) for 1550nm 10G (or 20G/25G) using 
> DMLs, one can consider using EDC to reclaim some of those penalties.
> EDC at 10G is readily available, but yet not presently available at 
> 20/25G.
>
> -Roger
>
> ----------------------------------------------------------------------
> --
>
> *From:* Tsumura, Eddie [mailto:etsumura@EXCELIGHT.COM]
> *Sent:* Friday, December 22, 2006 1:57 PM
> *To:* STDS-802-3-HSSG@listserv.ieee.org 
> <mailto:STDS-802-3-HSSG@listserv.ieee.org>
> *Subject:* Re: [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference
> 12/18/2006
>
> Hi everyone,
>
> I would like to propose the change of Reach (Technical) Feasibility 
> Table what Chris created.
>
> Please see attached file.
>
> We confirmed that 1550nm DML is possible to support 10G-10km/40km.
>
> Also, it might be possible to support 20G/25G-10km if only dispersion 
> penalty is a bottleneck.
>
> My proposed change is as follows.
>
> 1) 10km, 1550nm, 10G DML: change to "yes" from "maybe"
>
> 2) 40km, 1550nm, 10G DML: change to "yes" or "maybe" from "no"
>
> 3) 10km, 1550nm, 20G/25G DML: change to "maybe" from "no"
>
> Any comments are welcomed for me.
>
> Have a happy holiday!
>
> Eddie
>
> **Eddie Tsumura**
>
> Vice president of Engineering and Marketing
>
> **Excelight Communications** (A Sumitomo Electric Company)
>
> 4021 Stirrup Creek Drive
>
> Suite 200
>
> Durham, NC 27703
>
> Phone 919-361-1634
>
> Fax 919-361-1619
>
> etsumura@excelight.com <mailto:etsumura@excelight.com>
>
> ----------------------------------------------------------------------
> --
>
> *From:* Dove, Dan [mailto:dan.dove@HP.COM]
> *Sent:* Monday, December 18, 2006 5:46 PM
> *To:* STDS-802-3-HSSG@listserv.ieee.org 
> <mailto:STDS-802-3-HSSG@listserv.ieee.org>
> *Subject:* [HSSG] <HSSG-FO> Fiber Optic Ad Hoc Teleconference 
> 12/18/2006
>
> HSSG Members,
>
> We held our teleconference this morning but unfortunately, the bridge 
> supplier I use had changed their PIN system and I was forced to send 
> out a new bridge number at the last minute. I realize some people were

> unable to attend due to this occurrance, and I give you my apologies.
>
> I am going to provide the meeting notes below. If you see any need for

> correction to the notes, please send an email to dan.dove@hp.com 
> <mailto:dan.dove@hp.com> and I will update the minutes before posting 
> them on the HSSG-FOAH website. This will prevent multiple messages 
> with corrections being sent onto the reflector, OK?
>
> Also, for all who are interested, I am calling for presentations at 
> the January Interim. Those presentations should be specific to proving

> technical and economic feasibility for a single-mode 10Km link, and 
> for a 100m multi-mode link. It is important to recognize that we are 
> not trying to select a specific proposal at this time, but to 
> demonstrate to > 75% of the HSSG and IEEE 802.3 that we have proven 
> these things.
>
> Today's work was focused on single-mode only, so we should be sure to 
> get multi-mode presentations on the table as well.
>
> Regards,
>
> Dan Dove
>
> Chairman, HSSG FO Adhoc
>
> ==============================================
>
> Meeting Notes:
>
> Attendees:
>
>
> Last
>
> First
>
> 	
>
> Employer
>
> 	
>
> Affiliation
>
> 	
>
> 12/18/2006
>
> Anslow
>
> 	
>
> Pete
>
> 	
>
> Nortel Networks
>
> 	
>
> Nortel Networks
>
> 	
>
> Y
>
> Clairardin
>
> 	
>
> Xavier
>
> 	
>
> Kotura
>
> 	
>
> Kotura
>
> 	
>
> Y
>
> Cole
>
> 	
>
> Chris
>
> 	
>
> Finisar
>
> 	
>
> Finisar
>
> 	
>
> Y
>
> Dove
>
> 	
>
> Daniel
>
> 	
>
> Dove Networking
>
> 	
>
> ProCurve Networking by HP
>
> 	
>
> Y
>
> Dudek
>
> 	
>
> Mike
>
> 	
>
> Picolight
>
> 	
>
> Picolight
>
> 	
>
> Y
>
> Fischer
>
> 	
>
> Thomas
>
> 	
>
> Siemens
>
> 	
>
> Siemens
>
> 	
>
> Y
>
> Jaeger
>
> 	
>
> John
>
> 	
>
> Infinera
>
> 	
>
> Infinera
>
> 	
>
> Y
>
> Jiang
>
> 	
>
> Wenbin
>
> 	
>
> JDSU
>
> 	
>
> JDSU
>
> 	
>
> Y
>
> Patel
>
> 	
>
> Sashi
>
> 	
>
> Foundry Networks
>
> 	
>
> Foundry Networks
>
> 	
>
> Y
>
> Pepeljugoski
>
> 	
>
> Petar
>
> 	
>
> IBM
>
> 	
>
> IBM
>
> 	
>
> Y
>
> Song
>
> 	
>
> Steve
>
> 	
>
> Exelight
>
> 	
>
> Exelight
>
> 	
>
> Y
>
> Tatah
>
> 	
>
> Karim
>
> 	
>
> Cray
>
> 	
>
> Cray
>
> 	
>
> Y
>
> Tsumura
>
> 	
>
> Eddie
>
> 	
>
> Exelight
>
> 	
>
> Exelight
>
> 	
>
> Y
>
> Discussed the new table additions and corrections made since last 
> teleconference;
>
> Discussed Chris Cole Presentation;
>
> 10G Rows:
>
> Matt: Discrete devices available, extrapolation not necessarily valid
>
> For example, EMLs tend to be larger, take up more wafer space
>
> Chris: Another thing not captured, 10G DML @ 1550 not considered in 
> green but maybe we should take it off the study list because hard to 
> build as monolithic array
>
> Xaviar: Would like to avoid taking things off at this point. Would 
> like to present work showing it's a viable alternative.
>
> Chris: DML at 1550 gonna work at 40k?
>
>
> Xaviar: Objective is 10K, we need to keep. Not captured, whether CWDM 
> or DWDM in that column.
>
> Mike D: CWDM being considered for all or just 1310?
>
> Chris: Does not address this distinction. Slide 4 addresses some of 
> this, but slide 3 is trying to capture wavelength and transmitter 
> type. This table does not capture "optimum" or "implementation 
> complexity".
>
> 20G Rows:
>
> Chris discusses his perspective. Nobody argued with his position that 
> 20G 10K DML not possible.
>
> 50G Rows:
>
> General: Would be nice to have 40K and 10K leverage common approach. 
> Green applied where it appears to be possible.
>
> Marc Lucent: Main objective regarding DQPSK at 50 / 1310 is
dispersion?
>
> Chris: Yes, the implementation for 50G 1550 looks large and therefore 
> does not seem LAN oriented. At 1310 makes better sense.
>
> Marc: For 40Km, 1550, might make better sense for this longer reach.
>
> Chris: Yes, the breakpoints for 10G are 1310 DML and 1550 EML, so this

> is consistent. For us, its possible to set the breakpoint at 40-80Km.
> Question is, do we really want to add the cost of this technology for 
> 40Km.
>
> Marc: Chromatic dispersion your primary concern?
>
> Chris: Looking forward to presentations on this subject.
>
> Robert Lingle: Main Point to have a pluggable in a small form factor 
> and dispersion compensation may prohibit this.
>
> Chris, yes.
>
> Peter: Do you see an activity to standardize a form factor for a
module?
>
> Chris: yes. Likely done outside IEEE.
>
> Peter: Would larger group oppose having different form factors?
>
> Matt: With regard to 20/25G DML, been looking at 1310...wonder, at 
> 10K, what were the dispersion numbers that led you to your conclusion?
>
> Dan: My notes missed some of the content on this part..it was pretty 
> dense..sorry;
>
> Chris: Is there a breakpoint between 20G and 25G on DML?
>
> Chris: Could provide an EML spec as long as it was possible to reduce 
> cost in future.
>
> Frank: We are dealing with tech feasibility, we need to consider cost 
> too. For example, with 25G, you cannot do arrays, it creates a big 
> cost issue.
>
> Chris: yes, this is accurate. We need an economic feasibility table.
>
> Discussion of page 4:
>
> Copying conclusions from page 3 and modified format to make things 
> more visible;
>
> Why did 1550 get eliminated? Concluded DMLs in the timeframe not 
> feasible due to dispersion.
>
> This conclusion has been challenged and presentations may come in to 
> address that.
>
> For EML 20G 40Km is not leveragable.
>
> Matt: Agrees with earlier comment to allow for 1st gen to focus on EML

> with longterm DML targets... maybe able to collapse rows 2,3 and 4,5 
> together.
>
> Chris: Would be good to come up with an approach that allows this to 
> happen.
>
> Chris: We can add to the format additional proposals per email and 
> discussion.
>
> Mike: Is the intent of the cooling and grid columns to be 1 for 1
across?
>
> Chris: If no cooling, drift will be larger and thus they should be 
> cited independently.
>
> Chris: Semi-cooling is less precise than cooling, but offers a lower 
> cost means..for example heating to ensure minimum temps are
eliminated.
>
> No more comments on presentation. Some discussion on the upcoming 
> meeting and it was stated by the chair that we should build 
> presentations for January's interim rather than divide our attention 
> on another phone conference.
>
> Teleconference closed.
>

HSSGFO_alternatives_011707.pdf