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

Re: [8023-10GEPON] Jitter ad-hoc status



Dear Dr. Hirth, Jitter Adhoc members

I greatly appreciate Dr. Hirth for your effort taking this ad hoc leadership.

I would like to comments on feasibility of TP7->TP8 jitter,
which I could not point out at Adhoc meeting on March 18th evening.



10G jitter budget is based on the concept of CEI-11G serdes jitter budget.
At CEI serdes situation, the effect on thermal noise across voltage axis
(not phase noise), which is generated at analog front-end amplifier, may not
be considered, because the amplitude of received signal is as large as 
several 10's mV range and is much larger than thermal noise.

However, in a fiber optic receiver, thermal noise in TIA and/or shot noise
in APD across voltage axis is not negligible.  Attached is a simplified
model of an optical receiver output waveform with appropriate roll off.
As shown in the slide, rms value of noise is as large as 1/6.18 of OMA at 1E-3 
BER. This noise is converted to RJ with rms value (1 * sigma) of (1/6.18)*UI.
It was not concluded the BER condition to determine jitter budget at the Adhoc, 
  1E-3 (P-P jitter due to RJ is 3.09 * sigma) or
  1E-12 (P-P jitter due to RJ is 7.03 * sigma),
the above basic consideration suggests that TP7->TP8 jitter value (delta TJ)
of 0.1UI is not realistic, even if we consider only RJ.

In addition to RJ, DJ is also crucial issue in a burst mode optical receiver.
DCD at higher input power will be the dominant factor of DJ, rather than ISI 
in AC coupling, through my experience of burst mode receiver up to 1.25Gbps. 
The phase offset between 1010.... preamble and 64B/66B coded DATA area also 
acts as DJ with appoximately 0.05UI to 0.1UI.

Considering above situation, I propose to modify the upstream jitter budget
to allow TP7->TP8 jiter value (delta TJ) of 0.25. 
It is too challenging to realize burst mode CDR with jitter tolerance of 
more than 0.7, the only solution will be to reduce the jitter at TP5.
Referred to SONET spec, 0.1UI TJ at TP5 will be feasible.

With regard to phase offset between 1010.... preamble and 64B/66B coded
DATA area, I have an idea to reduce this effect to be presented later.


Best Regards,
Takeshi Nagahori
NEC


------------ Original Message ------------

>Hi all,
>First of all I'd like to thank Vijay for leading the ad-hoc meeting in Orlando.  I was select to lead the ad-hoc for our meeting in Tokyo since Vijay will not be able to attend.
>
>The jitter ad-hoc met in Orlando and proposed jitter budgets for the upstream and downstream 10G-BasePR PHYs.  We left the meeting needing to verify the nominal jitter values selected.  The 10G jitter budget is very tight.  Please review the jitter table and study the feasibility of each test point.
>
>Orlando summary:
>http://www.ieee802.org/3/av/public/2008_03/3av_0803_pathak_3.pdf
>
>Jitter table:
>10G-BasePR
>TP1 24ps 0.25UI
>TP2 34ps 0.35UI
>TP3 53ps 0.55UI
>TP4 68ps 0.70UI
>TP5 24ps 0.25UI
>TP6 39ps 0.40UI
>TP7 48ps 0.50UI
>TP8 68ps 0.70UI
>
>
>1. For TP1 - Is serial output jitter feasible?
>    OIF-CEI-02.0 11G-SR and XFPr4.5 Specifications are 0.3UI Tj.  Should this value be increased to comply with these specifications?
>
>2. For TP1->TP2 - Is jitter added by an EML feasible?
>    Can optical experts confirm this? (10ps,0.1UI)
>
>3. For TP2->TP3 - Is jitter due to chromatic dispersion correct?
>
>4. For TP3->TP4 - Is jitter due to PIN/APD, Limiting AMP, AC coupling, and cross talk feasible?
>
>5. For TP4 - Is receive jitter tolerance feasible?  Is their margin for sinusoidal jitter test?
>  OIF-CEI-02.0 11G-SR agrees 0.7UI Tj.  XFPr4.5 requires 0.65UI.
>
>6. For TP5 - Is serial output jitter feasible with loop-timing?
>    OIF-CEI-02.0 11G-SR and XFPr4.5 Specifications are 0.3UI Tj.  Should this value be increased to comply with these specifications?
>
>7. For TP5->TP6 - Is jitter from DML Laser feasible?
>    Can optical experts confirm this? (15ps,0.15UI)
>
>8. For TP6->TP7 - Is jitter due to chromatic dispersion correct?
>
>9. For TP7-TP8 - Is jitter due to burst-mode PIN/SPD, LA, AC coupling, and cross talk feasible?
>
>10.For TP8 - Is jitter tolerance capable of burst synchronization?
>  OIF-CEI-02.0 11G-SR agrees with 0.7UI Tj.  XFPr4.5 requires 0.65UI.  These values are not intended for burst mode operation.
>
>
>Other issues:
>
>
> 1.  Haim is to send out the jitter spreadsheet used by the jitter ad-hoc in the Orlando meeting. (Haim)
> 2.  Should the sinusoidal jitter corner frequency be at 5Mhz? (Haim)
> 3.  Is the jitter transfer curve feasible? (Ryan)
>
>
>Please provide feedback promptly so that we can be prepared for the April meeting.  Feedback is open for anyone and is not limited to the members of the Orlando group.  Please let me know if there any other issues that should be added to the jitter ad-hoc list of issues.  I will prepare a summary of the feedback I receive by April 11 to report at the Tokyo meeting.
>
>Thank you,
>
>Ryan Hirth
>Teknovus Inc.
>
>
-----------------------------------------------------
長堀 剛
t-nagahori@ah.jp.nec.com
日本電気(株) 光デバイス事業部 第二商品開発グループ
〒270-1198 千葉県我孫子市日の出1131
社内メール: 26-52801
TEL: 04-7181-8738(内線8-26-76324)
FAX: 04-7185-7926(内線8-26-57926)
-----------------------------------------------------
Takeshi Nagahori
Manager, Engineering
Optical Access Groop
Fiber Optic Devices Division
NEC Corporation
t-nagahori@ah.jp.nec.com

nagahori_jitter080402.pdf