| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Steve and Brad Steve correctly explain what I meant. I do agree that TP1 and TP2 specifications have been relaxed for the 40GBase-SR4/100GBase-SR10 compare to SFP+ (SFF-8431)/CL52 but the receiver specifications for TP4 is the same as what we have in SFP+ 0.7 UI TJ. The SR4/SR10 SerDes RX is harder by keeping the value the same as SFP+ but we are only doing 100m. Assuming Jack Jewel proposal for tighter transmitter is practical and since the receiver is already as hard as SFP+ I don't see why we can't do 300 m OM3 per CL52! This approach would not add cost to the baseline proposal and we are not adding a new PMD. Thanks, Ali Swanson, Steven E wrote: Brad, I think what Ali is saying is that if one had the same transceiver specifications that we had in 802.3ae, one could support 300m on OM3 parallel fibers. The problem is that most VCSEL manufacturers cannot maintain those same VCSEL specifications on a 4x arrays or a 10x arrays. My problem is that they have gone too far in the relaxation of the 802.3 ae specs so that we now cannot support anything longer than 100m on OM3 and even if we were to provide a higher performance fiber, we can't get additional length because we are dispersion limited. I don't think Ali's proposal to ref Cl 52 is really practical nor do I think the current 802.3ba baseline is really practical; the practical answer lies somewhere in between the transmitter spec in 802.3ae and the baseline proposal in 802.3ba. Cheers, Steve -----Original Message----- From: Brad Booth [mailto:bbooth@xxxxxxxx] Sent: Wednesday, July 23, 2008 2:46 PM To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx Subject: Re: [802.3BA] MMF Extended Reach conference call reminder Ali, Hopefully you can help me understand something related to what you said. I agree it would be possible for XR to use the existing specifications in Clause 52 as a baseline for doing 300m on OM3. What I don't understand, and I hope you can clarify, is that the 300m on OM3 in Clause 52 was based upon a single wavelength on two strands of MMF; are you implying the re-use of that 300m of OM3 fiber currently being used for 10GBASE-SR? A WDM approach would permit re-use of the existing installation, but so far the task force approach being taken is parallel fiber. As far as the MMF reach objectives are concerned, there is no legacy cabling infrastructure. I apologize if I've missed some of the discussion related to this point, but it seems that extended reach requires a new cabling infrastructure and therefore needs to meet the 5 Criteria if the task force is going to put any serious effort into it. If it was a case of saying "add EDC and get an extra x meters" or "the next generation MMF can go x more meters", then that seems like a simpler effort which can be handled by an annex. If this is about creating a whole new PMD, then the task force needs buy into it with an objective. Just my thoughts on this. Thanks, Brad -----Original Message----- From: Ali Ghiasi [mailto:aghiasi@xxxxxxxxxxxx] Sent: Wednesday, July 23, 2008 12:39 PM To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx Subject: Re: [802.3BA] MMF Extended Reach conference call reminder Alessandro/John Looking at some of the email trails related to XR several people suggested for application longer than 100 m just use SMF. I absolutely agree with the approach of using SMF for >100m if you could shove the SMF module into the MMF QSFP cage. We had several constructive discussion toward a resoluAlessandro/John Looking at some of the email trails related to XR several people suggested for application longer than 100 m just use SMF. I absolutely agree with the approach of using SMF for >100m if you could shove the SMF module into the MMF QSFP cage. We had several constructive discussion toward a resolution for MMF XR last week with 5 option on the table: - Tighter transmitter specifications - CDR outside the module - CDR inside the module - Simple EDC (2 tap DFE) ghiasi_01_0108 was replaced with dudek_01_0708 - Optional FEC - FEC always encode (John P withdraw this proposal) At the meeting it was stated CDR in the module will be least invasive and would not require defining a 2nd PMD. There is no guarantee if you put CDR in the module the TX or RX jitter will improve, we have to define a new PMD with full set of TP2 and TP3 specifications. If we do not want to define a new PMD for XR the obvious solution would be to reference 802.3 CL 52 single optical specifications for SR4/SR10, it already defines 300 m on OM3 fiber! Thanks, Ali Alessandro Barbieri (abarbier) wrote: |
begin:vcard fn:Ali Ghiasi n:Ghiasi;Ali org:Broadcom;HSIP adr;dom:;;3151 Zanker Road;San Jose;CA;95014 email;internet:aghiasi@xxxxxxxxxxxx title:Chief Architect tel;work:(408)922-7423 tel;cell:(949)290-8103 version:2.1 end:vcard