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

Re: [STDS-802-3-400G] Sept Interim Meeting Material Posted



I think that being careful means testing the assumption that everything can be cost reduced through integration, and that it can be further achieved while meeting ‘consensus’ values. I think it would be prudent to look at a buildup of the different solutions that are being discussed (inclusive of cost, power, size comparisons), as well as how those solutions could evolve over time (and where they would run out of steam).

 

Brian

 

From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx]
Sent: Wednesday, September 17, 2014 10:59 AM
To: Brian Welch; STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: RE: Sept Interim Meeting Material Posted

 

Hi Brian,

 

The example is from a published paper, which is properly cited, so that is a good place to get more details. It is also a research device so has many limitations. It is not the basis of a specific proposal or product, but rather an example of what is possible. Despite the limitations, this is a real physical PIC with actual measurements, in contrast to simulations of PICs that have been used to generate many results in .bs.

 

With respect to viability of approaches other than yours, we need to be careful in dismissing them. The only real SiP PICs to date have low optical output power. So we should be open to other approaches which can improve the performance.  

 

Chris

 

From: Brian Welch [mailto:bwelch@xxxxxxxxxxx]
Sent: Wednesday, September 17, 2014 10:44 AM
To: Chris Cole; STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: RE: Sept Interim Meeting Material Posted

 

Chris,

Thanks for sending. Is there cost modeling done using the technique on page 20? Are there losses for that structure? I see a chart on slide 21 that seems to suggest the OMAs are somewhere between -10 to -20 dBm…. Is there more info on this?

 

How do the driver/receive I/Cs figure into it? Would be good to see how a whole module could come together using this approach.

 

I don’t believe the alternate integrative approaches shown in 802.3bm, at least those that were shown consistently, are representative of solutions that can be effectively deployed in volume. Many of them made no mention of how coupling/assembly was performed, variability/control over process and temperature corners, or how polarization diversity was accommodated on the receiver.

 

Thanks,

Brian

 

 

From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx]
Sent: Wednesday, September 17, 2014 10:30 AM
To: Brian Welch; STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: RE: Sept Interim Meeting Material Posted

 

Hi Brian,

 

You make good points that we have seen proposals and measurements with high grade components, such as used for long haul telecom applications. The high cost and power is then dismissed as not representative of integrated implementations, without detailed discussion of what the implementations actually are.

 

We do need to be cautious about using your approach to benchmark implementations. The SiP PIC grating coupler which you have presented is bandwidth limited so obviously has high loss for WDM applications. However, there are other approaches, including for SiP PIC like some presented in .bm, which do not have this limitation.

So I agree with you that if when you present your technology it is not appropriate to use the proposed lower Mux/DeMux loss numbers. However, we will keep in mind that it’s not good approach to WDM.

 

With respect to your beliefs, I hate to shake them, but here is at least one real example of 8x50G integrated solution:

 

http://www.ieee802.org/3/bs/public/14_07/cole_3bs_01a_0714.pdf#page=20

 

Chris

 

From: Brian Welch [mailto:bwelch@xxxxxxxxxxx]
Sent: Wednesday, September 17, 2014 9:16 AM
To: Chris Cole; STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: RE: Sept Interim Meeting Material Posted

 

My proposal is not to assume very low loss technologies for technical feasibility, while simultaneously assuming that future integration will bring the cost down. I’ve heard many times on questioning that products will be cost reduced in the future with ‘Integration and SiP’, although no presentations as to whether these proposals would allow for such measures.

 

As per what I think are likely losses for WDM MUX/DEMUX I’ll refer you to my prior bm and bs presentations on the matter, at least for 2L and 4L. I don’t have anything on 8L as I don’t believe such solutions will ever allow for integrated solutions, although I may be able to provide more quantitative answers in our next meeting.

 

Brian

 

From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx]
Sent: Wednesday, September 17, 2014 9:05 AM
To: Brian Welch; STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: RE: Sept Interim Meeting Material Posted

 

Brian

 

What is your proposal?

 

Chris

 

From: Brian Welch [mailto:bwelch@xxxxxxxxxxx]
Sent: Wednesday, September 17, 2014 9:01 AM
To: Chris Cole; STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: RE: Sept Interim Meeting Material Posted

 

What technologies/integration are you assuming in these MUX/DEMUX figures? Conservative for one technology (ie, discrete thin film) may not be for another (ie, integrated SiP).

 

Brian

 

From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx]
Sent: Tuesday, September 16, 2014 4:15 PM
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: [PossibleSpam] Re: [STDS-802-3-400G] Sept Interim Meeting Material Posted

 

Dear Task Force Participants,

 

Last week, the TF gave permission to update cole_3bs_02 to include TX and RX analog BW used in TDP calculations, which was requested during Q&A. The updated deck was posted last week.

http://www.ieee802.org/3/bs/public/14_09/cole_3bs_02b_0914.pdf#page=5 (pages 5 and 6)

 

On a separate topic, in this presentation, we use 2.5dB for 4:1 WDM Mux and 1:4 WDM DeMux loss, which is conservative. Other presentations use loss values of 1.5dB and 2.0dB. Since these values do not enter into normative specifications, normally there is no need for agreement. However, when making comparisons between different approaches, if the loss values are not the same, the results are not apples-to-apples.

 

To enable fair comparison of different results I would like to propose that we all use 2.0dB as the value for 4:1 Mux loss, and for 1:4 DeMux loss. This value was used by Alan Tipper in his analysis. This is neither conservative nor aggressive. Using this number in no way constrains implementations. When we adopt normative specifications, everyone is free to set their design and production targets as they please.

 

If we can agree on a consensus 4:1 Mux and DeMux loss value for use in analysis, I will update our optical specifications analysis to reflect it.

 

Thank you


Chris