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

[10GMMF] TP3: Suggested stressor levels collected responses

Hello TP3 group,

Here are the responses to request for suggested TP3 stressor levels.  The responses were slower arriving than I had hoped, so here is what I have received so far.  Per the request, I have listed the responses with any identifying information removed so that people would be more at ease to speak freely.  One of the suggestions was posted to the reflector (so everyone knows the author's name and can read the full text), but I listed it like the others.

Also, since I have received no objections to the request to push the next TP3 teleconference to April 5, that is what we are going to do.  This should allow people time to prepare and be available to discuss the TP3 stressors in detail.

Thank you for all of your participation.

- Jim McVey

+ 1 650 740 7732

Suggestion A:

Max PIE-D 4.25 dB, prefer 4.0 dB

Consider the project's objectives for cost effectiveness and small form factor / high port density friendliness.  Note that there are many causes of implementation penalties, so we have to be conservative.  None of the stressors should be harder (for the set of perfect finite equalizers) than is represented by a PIE-D of 4.25 (for the perfect infinite equalizer).  4 dB would be preferable.

Suggestion B:

PIE-D 4.5 dB

My suggestion is for the stress level to be based on a PIE-D metric of 4.5dB.

This is based on detailed finite-length equalizer simulations of the Cambridge 108, and Monte Carlo OM1 and OM3 models.  We have done a first pass feasibility analysis of an EDC chip both from a power and cost point of view, based on existing and projected IP.  All this leads us to believe that we will be able to deliver a robust, viable and manufacturable EDC chip that will meet the standard and the cost and power constraints that the market demands.

Suggestion C:

PIE-D  4.5 dB

Our recommendation is for a PIE-D of 4.5 dB.  Our rationale is as follows:

1) The three cases (pre-cursor, post-cursor & symmetric) are all equally likely and equally stressful from a PIE-D perspective.  Hence the number of 4.5dB is proposed for all of them.

2) John Ewen's methodology can create pre-cursor cases by translating the PIE-D number into nn% for a range of finite EDC's through the CDF.  The procedure for creating post-cursor & split-symmetrical cases needs be understood.

3) In the Atlanta meeting an EDC integrated into a module was shown to pass the D1.2 stressors (bhoja_1_0705).  These IPR's have PIE-D values up to 5.1dB.  Sufficient margin in the EDC exists to develop low power, high yielding optical modules.

4) Lab results on sample fibers such as TIA 12/96 as well as the Gen67YY fiber simulation model indicates that a number smaller than 4.5dB PIE-D will result in many failing fibers even with dual launch.  The standard will then be below market expectations.

5) Industry experience with 100BASE-TX, 1000BASE-T etc. have shown that the power consumption & performance of PHY solutions improve dramatically over relatively short time.  However, if customers are not satisfied with the quality of early modules because the specification was not set high enough to yield good performance such gains due to Moore's law will not be given an opportunity.

Suggestion D:

PIE-D 4.50 dB

We recommend a PIE-D target for LRM links of no less 4.5 dB.  There are several reasons for this:

We are investing heavily into LRM and want it to be successful.  Initial impressions are critical to that success, and if there is any perception that LRM is not robust, we are worried that users will hesitate, and the market will further commit itself to LX-4.  We are not supportive of any sacrifice of expectations at this critical and competitive time.

We accept the dual launch approach, but are concerned in how it might actually play out.  There is sufficient margin in LX and LX-4 that an offset patch cord is seldom required, but we are less comfortable with the margins of LRM to accept a target of less than 4.5 dB.  If patch cords are too frequently required, we are again uneasy about any
There are currently numerous uncertainties in modeling and budgeting involving center launch, connectors, performance of real laser transmitters, lack of OM2 models, measurement methods, etc.  We fear that some of these uncertainties will turn out to be real and serve to reduce coverage, so it is imperative that we are conservative and do not start with a lowered target.

There are several companies that believe 4.5 dB is achievable in a timeframe that will support the successful launch of LRM.  This will only improve with time.  We should not base decisions with long-term impact on technologies that have not yet had a chance to mature.
To TP3's specific question, we are very confident that the value of 4.5 dB can be achieved.  Even with changes to 300 meters and tightening of models, we still believe in this, and have every confidence in our ability to equalize to the 4.5 dB level, and more.

Suggestion E:

1) My first choice is to leave the TP3 pulses as they were adopted in Vancouver by >75% vote.

2) My second choice, for 99%tile coverage, is to run John Ewen's model aiming at DFE coverage between 96 and 99%tile for three stressor cases, expecting the precursor penalties to be nearer 99%tile while post cursor and quasi-symmetric penalties will be lower in the given range.

3) If we must reduce the coverage goal to 95%tile, and if David tells us to enforce that on two-way link coverage, then my third choice would be to aim at DFE coverage at the 95 to 97.5%tile range.

4) If we must reduce the coverage goal to 95%tile, and if David tells us to enforce that on one-way link coverage, then my fourth choice would be to aim at DFE coverage of 94 to 95%tile.  It should get easier to hit a more narrow range of penalties with the three IPR categories at those reduced %tile levels.