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

Re: [802.3_100GNGOPTX] PAM-N discussion in today's NG 100G OE SG



Hello Vipul, Chris,

 

Thank you for your comments. Sadly, you are correct that I get a lot of my information from refereed published results. But it is exciting to learn that the actual state of the art has moved far ahead of reported results, and that even better we will get to learn how far it has advanced in March. I will be suggesting to some of the Silicon Photonics researchers I know at IBM, ALU, Intel, and other companies that they should definitely attend the IEEE 802.3 March Plenary meeting. Not only is Hawaii a magnificent location for the meeting, but they will learn how much they have fallen behind.

 

Having my bubble pricked at the March meeting will be painful, but well worth the educational experience. Although it does raise the question as to why you are waiting until March to do it. Why not prick my bubble at the 802.3 Interim meeting this week? You have been working on this technology for years, so would it not have been better to assemble the pin prick sooner rather than later?

 

I did a quick count of the pages in Sudeep’s presentation and there are roughly 20 with technical content of which best case 2 could be argued have material on Silicon Photonics. The rest are superb work by Sudeep and his colleagues on System Modeling, Signal Processing, and IC architectures, demonstrating something everyone already knew that Broadcom really understands how to design Signal Processing ASICs.

 

However as you point out none of us know the actual state of the art of Silicon Photonics, including not realizing that the real state of the art is 3x faster than the best reported results. Since this is an Optical track and we are discussing adopting optical objectives, it is an odd strategy to have no material on optical performance, and to only present ASIC feasibility material to an optics audience.


You have now set high the expectation for you presentation in March. Hopefully, you will present more than rise/fall results, but show comprehensive measurements, discuss ALL the important optical impairments that effect PAM-N modulation, and discuss in detail how these will be solved in production silicon processes. If this does not happen, you will find it much harder to use as an excuse the audience’s lack of knowledge about your area of expertise.

 

I look forward to the seeing your state of the art results in March.


Chris

 

From: Vipul Bhatt [mailto:Vipul_Bhatt@xxxxxxxx]
Sent: Tuesday, January 24, 2012 9:37 PM
To: STDS-802-3-100GNGOPTX@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GNGOPTX] PAM-N discussion in today's NG 100G OE SG

 

Chris,

 

I am sorry to prick your bubble, but your simulations are based on references that are less relevant now. The state of the art of silicon photonics modulators has moved ahead, though not fully reported in public domain.

 

The co-authors and supporters of PAM proposal are preparing to bring some very encouraging data in March, assuming this study group will give them an opportunity.

 

Regards,

Vipul

 

Vipul Bhatt

Lightwire, Inc.

(408) 461-8521 mobile

 

 

 

From: Chris Bergey [mailto:cbergey@xxxxxxxxxxx]
Sent: Tuesday, January 24, 2012 6:01 PM
To: STDS-802-3-100GNGOPTX@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GNGOPTX] PAM-N discussion in today's NG 100G OE SG

 

Chris

 

Your comments are noted and clearly the rise and fall times required for a PAM8 or PAM16 system would need to be greater than 20ns.  I would note that many of the papers you highlighted were focused on 10G NRZ systems for which a 20nS rise time would be quite sufficient and desired.  As a data point, I would point you to the Palkert 01_1111 presentation from Atlanta showing 4x28G transceiver eyes.  These eyes exhibited a rise and fall time <15ps and an ER>3dB, although we took out the measurement details out of the posted IEEE presentation we can provide them again.

 

Regards

Chris

 

 

 

On Tue, Jan 24, 2012 at 4:53 PM, Chris Cole <chris.cole@xxxxxxxxxxx> wrote:

 

> During this afternoon’s NG 100G OE SG meeting, a proposal for PAM-8 and

> PAM-16 modulation 100GE alternatives were presented.

> 

> 

> It appears that required silicon modulator speed as measured by rise and

> fall times is about 3 times faster than the fastest reported devices in the

> literature.

> 

> The PAM-8 and PAM-16 eyes when simulated with the speed reported by other

> companies (IBM, ALU, Fujikura, Intel, Luxtera) result in largely closed

> eyes.

> 

> We have submitted a post-deadline presentation showing ideal simulations

> where the ONLY optical device limitation is rise/fall time which is the

> simplest impairment to look at.

> 

> Because of the full schedule, we may not get the time to present it. The

> presentation is posted at the following ftp site and available for download

> for those interested:

> 

> site:                  ftp://ftp.finisar.com/

> 

> username:         ieee

> 

> password:         ieee8023

> 

> file name:            cole_03_0112_NG100GOPTX.pdf

> 

> 

> Chris