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

Re: [802.3_B10K] Dispersion Penalty Presentations



Dear Colleagues,

 

I have received a number of questions about the use of statistical data analysis to quantify the yield to a specification. This is just basic SPC (Statistical Process Control) developed by Shewhart and Deming in the ‘30s, and then popularized by Deming, most famously in Japan in the ‘50s. A search on SPC will reveal a wealth of references. I suggest some brushing up on this prior to the 802.3cn Interim Teleconference  which will be held Tuesday, Sept. 24, 10AM ET to discuss publicly submitted comments against Draft 3.0; please see enclosed invite from the Chair.

 

The measurement data used to for the analysis is listed and summarized in Helen’s presentation:   http://www.ieee802.org/3/cn/public/18_11/xu_3cn_02b_1118.pdf#page=4

It should be viewed as having an optimistic bias in predicting yield. Defective or low performance device data does not make it into presentations. In a proper data set, all devices would be counted, including defective ones.  

 

Since SPC may not be intuitive to everyone, I added a Margin metric to each of the three tables below. This is the difference in dB between the TX OMA (min) and RX Sens OMA (max), and their corresponding measurement data averages. This is what we typically use to qualitatively judge the robustness of a proposed spec. At this stage we typically like to see 3dB or margin. 2dB is OK, and 4dB is great. Below 1.5dB we start to be concerned, and below 1dB we have a problem. This is an incomplete metric because it doesn’t factor in the distribution. Encouragingly the qualitative metric leads us to the some conclusions as the statistical yield metric.

 

Table 1 shows that ER8 transmitter and receiver specs. have little margin. This is consistent with what Yamamoto-san and Yoshimatsu-san already showed at the Jan. 2018 meeting.

 

http://www.ieee802.org/3/B10K/public/18_01/yamamoto_b10k_01a_0118.pdf#page=8

 

They also showed that if we really want 40km, then we need some additional technique to make it feasible. They proposed the use of strong FEC. This can be considered in the future by Operators as a proprietary spec. based on an IEEE 30km spec.

 

Table 3 shows that the ER8 specifications can be adjusted to have good margin. Notice that Table 3 also shows the limitation of this Margin metric. TX has lower margin even though it has higher yield than RX. This is because the margin calculation doesn’t factor in that TX sigma is lower than RX sigma.


Thank you

 

Chris

 

 

Table 1.  P802.3cn draft D3.0 Select Optical Specifications

Transmitter Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

40

km

TX-OMA (avg)

EML .3cu data

or lower if optimum

7.4

8.5

8.5

dBm

TX OMA (sigma)

EML .3cu data

0.4

0.4

0.4

dBm

TDECQ

EML .3cu data

2

2

2.5

dB

Mux loss (max)

0

2

3

dB

TX OMA (max)

7.4

7.4

6.4

dBm

TX OMA (min)

3.4

3.4

2.4

dBm

TX OMA (min) margin

3.1

1.1

0.6

dB

TX OMA Yield

100.0

97.6

48.1

%

 

 

 

 

 

Receiver

Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

40

km

RX Sens OMA (max)

SECQ = 1.4

-15.1

-15.1

-16.1

dBm

DeMux loss (max)

0

2

3

dB

RX Sens OMA (avg)

SECQ = 1.4

APD .3cu data

-19.0

-19.7

-19.7

dBm

RX Sens OMA (sigma)

SECQ = 1.4

APD .3cu data

0.8

0.8

0.8

dBm

RX Sens OMA (max) margin

4.6

2.6

0.6

dB

RX Sens OMA Yield

100.0

99.8

14.1

%

 

 

 

 

 

TX OMA * RX Sens Yield

100.0

97.3

6.8

%

 

 

Table 2.  P802.3cn draft D3.0 Select Optical Specifications with changes proposed in public comments

Transmitter Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

TX-OMA (avg)

EML .3cu data

or lower if optimum

7.4

7.2

7.7

dBm

TX OMA (sigma)

EML .3cu data

0.4

0.4

0.4

dBm

TDECQ

EML .3cu data

2

2

2.5

dB

Mux loss (max)

0

2

3

dB

TX OMA (max)

7.4

5.2

4.2

dBm

TX OMA (min)

3.4

1.2

0.2

dBm

TX OMA (min) margin

3.1

3.3

2.8

dB

TX OMA Yield

100.0

100.0

100.0

%

 

 

 

 

 

Receiver

Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

RX Sens OMA (max)

SECQ = 1.4

-15.1

-17.3

-15.3

dBm

DeMux loss (max)

0

2

3

dB

RX Sens OMA (avg)

SECQ = 1.4

APD .3cu data

-19.0

-19.7

-19.7

dBm

RX Sens OMA (sigma)

SECQ = 1.4

APD .3cu data

0.8

0.8

0.8

dBm

RX Sens OMA (max) margin

4.6

0.4

1.4

dB

RX Sens OMA Yield

100.0

24.6

72.7

%

 

 

 

 

 

TX OMA * RX Sens Yield

100.0

24.6

72.7

%

 

 

Table 3.  P802.3cn draft D3.0 Select Optical Specifications with additional proposed changes

Transmitter Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

TX-OMA (avg)

EML .3cu data

or lower if optimum

6.0

8.5

8.5

dBm

TX OMA (sigma)

EML .3cu data

0.4

0.4

0.4

dBm

TDECQ

EML .3cu data

2

2

2.5

dB

Mux loss (max)

0

2

3

dB

TX OMA (max)

6

7

5.4

dBm

TX OMA (min)

2

3

1.4

dBm

TX OMA (min) margin

4.5

1.5

1.6

dB

TX OMA Yield

100.0

99.9

99.9

%

 

 

 

 

 

Receiver

Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

RX Sens OMA (max)

SECQ = 1.4

-16.5

-15.5

-14.1

dBm

DeMux loss (max)

0

2

3

dB

RX Sens OMA (avg)

SECQ = 1.4

APD .3cu data

-19.0

-19.7

-19.7

dBm

RX Sens OMA (sigma)

SECQ = 1.4

APD .3cu data

0.8

0.8

0.8

dBm

RX Sens OMA (max) margin

3.2

2.2

2.6

dB

RX Sens OMA Yield

99.9

98.8

99.5

%

 

 

 

 

 

TX OMA * RX Sens Yield

99.9

98.7

99.4

%

 

 

 

From: Chris Cole
Sent: Thursday, September 19, 2019 11:34 PM
To: 'STDS-802-3-B10K@xxxxxxxxxxxxxxxxx' <STDS-802-3-B10K@xxxxxxxxxxxxxxxxx>
Cc: 'STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx' <STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx>
Subject: RE: [802.3_B10K] Dispersion Penalty Presentations

 

Dear Colleagues,

 

I received several emails asking about the reasons for the proposed changes to Draft 3.0. The good news is that at least some people are reading my entire emails. The flip side is that their questions may be motivated by a suspicion that the new values were pulled out of the air.

 

Table 1 below lists select draft D3.0 Transmitter and Receiver Optical Specifications. It also includes supporting data presented in 802.3cu, in yamamoto_b10k_01a_0118, yu_b10k_01c_0319, huang_b10k__01a_0918, and jackson_b10k_01_0918. The transmitter section entries are the average and sigma of all reported EML TX OMA data. TX penalties are rough averages of the reported data. The receiver section entries are the average and sigma of all reported APD RX Sens data. These distributions are used to calculate the yields of the transmitter and receiver due to TX OMA (min) and RX Sens OMA (max) limits, respectively.

 

Table 1 highlights in red the entries that are of concern. ER8 transmitter and receiver have unacceptable yields, resulting in a specification that is not manufacturable. During Task Force meetings, multiple participants raised this as a serious concern.  For example, yamamoto_b10k_01a_0118 shows that there is insufficient margin, and offers a remedy of a stronger FEC to increase margin by 1dB. These data driven concerns should not have been ignored.

 

Table 2 below lists draft D3.0 Select Optical Specifications with changes proposed in the submitted public comments. ER8 transmitter and receiver have good margins, although the receiver yield needs improving. Unfortunately, the ER4 receiver yield was severely degraded and an additional change is required to fix this.

 

Table 3 below lists draft D3.0 Select Optical Specifications with additional proposed changes to address the concerns identified in Table 2. Additionally, there is a proposed reduction to ER1 TX OMA (min), which was not one of the publicly submitted comments, although it was flagged as a concern in an earlier email. Use of high power EML for ER1 unnecessarily increases cost. Lower output power EML still results in high yield specifications. All the specification changes are highlighted in blue.

 

Thank you

 

Chris

 

Table 1.  P802.3cn draft D3.0 Select Optical Specifications

Transmitter Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

40

km

TX-OMA (avg)

EML .3cu data

or lower if optimum

7.4

8.5

8.5

dBm

TX OMA (sigma)

EML .3cu data

0.4

0.4

0.4

dBm

TDECQ

EML .3cu data

2

2

2.5

dB

Mux loss (max)

0

2

3

dB

TX OMA (max)

7.4

7.4

6.4

dBm

TX OMA (min)

3.4

3.4

2.4

dBm

TX OMA Yield

100.0

97.7

49.2

%

 

 

 

 

 

Receiver

Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

40

km

RX Sens OMA (max)

SECQ = 1.4

-15.1

-15.1

-16.1

dBm

DeMux loss (max)

0

2

3

dB

RX Sens OMA (avg)

SECQ = 1.4

APD .3cu data

-19.0

-19.7

-19.7

dBm

RX Sens OMA (sigma)

SECQ = 1.4

APD .3cu data

0.8

0.8

0.8

dBm

RX Sens Yield

100.0

99.8

14.1

%

 

 

 

 

 

TX OMA * RX Sens Yield

100.0

97.4

6.9

%

 

 

Table 2.  P802.3cn draft D3.0 Select Optical Specifications with changes proposed in public comments

Transmitter Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

TX-OMA (avg)

EML .3cu data

or lower if optimum

7.4

7.2

7.7

dBm

TX OMA (sigma)

EML .3cu data

0.4

0.4

0.4

dBm

TDECQ

EML .3cu data

2

2

2.5

dB

Mux loss (max)

0

2

3

dB

TX OMA (max)

7.4

5.2

4.2

dBm

TX OMA (min)

3.4

1.2

0.2

dBm

TX OMA Yield

100.0

100.0

100.0

%

 

 

 

 

 

Receiver

Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

RX Sens OMA (max)

SECQ = 1.4

-15.1

-17.3

-15.3

dBm

DeMux loss (max)

0

2

3

dB

RX Sens OMA (avg)

SECQ = 1.4

APD .3cu data

-19.0

-19.7

-19.7

dBm

RX Sens OMA (sigma)

SECQ = 1.4

APD .3cu data

0.8

0.8

0.8

dBm

RX Sens Yield

100.0

24.6

72.7

%

 

 

 

 

 

TX OMA * RX Sens Yield

100.0

24.6

72.7

%

 

 

Table 3.  P802.3cn draft D3.0 Select Optical Specifications with additional proposed changes

Transmitter Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

TX-OMA (avg)

EML .3cu data

or lower if optimum

6.0

8.5

8.5

dBm

TX OMA (sigma)

EML .3cu data

0.4

0.4

0.4

dBm

TDECQ

EML .3cu data

2

2

2.5

dB

Mux loss (max)

0

2

3

dB

TX OMA (max)

6

7

5.4

dBm

TX OMA (min)

2

3

1.4

dBm

TX OMA Yield

100.0

99.9

99.9

%

 

 

 

 

 

Receiver

Description

50GBASE-ER1

200GBASE-ER4

400GBASE-ER8

unit

Reach

40

40

30

km

RX Sens OMA (max)

SECQ = 1.4

-16.5

-15.5

-14.1

dBm

DeMux loss (max)

0

2

3

dB

RX Sens OMA (avg)

SECQ = 1.4

APD .3cu data

-19.0

-19.7

-19.7

dBm

RX Sens OMA (sigma)

SECQ = 1.4

APD .3cu data

0.8

0.8

0.8

dBm

RX Sens Yield

99.9

98.8

99.5

%

 

 

 

 

 

TX OMA * RX Sens Yield

99.9

98.7

99.4

%

 

 

From: Chris Cole
Sent: Tuesday, September 17, 2019 5:59 PM
To: 'STDS-802-3-B10K@xxxxxxxxxxxxxxxxx' <STDS-802-3-B10K@xxxxxxxxxxxxxxxxx>
Cc: 'STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx' <STDS-802-3-100G-OPTX@xxxxxxxxxxxxxxxxx>
Subject: FW: [802.3_B10K] Dispersion Penalty Presentations

 

Dear Colleagues,

 

I have received several private emails correcting errors and making suggestions for clarifications, to my previous email.

Below is the corrected text, with the addition of IEEE Public Review Comments that have just been submitted on the IEEE Standards site. The deadline for comments is today, so if you have any, now is the time to submit them.

 

The following is an explanation of Chromatic Dispersion Penalty (CDP) scaling with reach and Baud rate, and interplay with transmitter chirp characteristics.

 

CDP is proportional to CD coefficient, link length, and Baud rate squared.

 

CDP CD * L * B2

 

Calculating the exact penalty is complex, however the CD * L * B2 term can be used for direct, relative comparisons, by defining it as a CDP Figure of Merit (CDP FM).

For convenience, 10km and 25Gbaud are used for normalization.

 

CDP FM = CD * (L/10) * (B/25)2

 

Table 1 below lists key transmitters specs. and CDP FM for various codes, with red highlighting concerns. The take away points are:

 

  • 400GBASE-LR4 10km reach proposal had excessive CDP FM.
  • 400GBASE-LR4 6km reach baseline has reasonable CDP FM.
  • 400GBASE-ER8 40km reach baseline has excessive CDP FM for CD (min).

(Qualitative statement that 50G PAM4 “is a lot easier” than 100G PAM4 is not true at 40km. In fact the opposite it true;  400G ER8 40km (50G PAM4) is harder than 400G LR4 6km and 100G LR1 10km (both 100G PAM4). )

  • 400GBASE-ER8 30km reach proposal will have reasonable CDP FM.
  • 200GBASE-ER4, 400GBASE-ER8 40km baselines have excessive TX OMA (min). 50GBASE-ER 40km baseline may have an excessive TX OMA (min).

(This is higher than previous TX OMA values (0.3 and 0.1 dBm for 40G ER4 and 100G ER4, respectively), and is higher than other codes in Table 1. It will significantly drive transmitter cost up by likely requiring a TX amplifier. This undermines the Broad Market Potential, and is counter to the expressed need for low cost by Network Operators.)

  • 200GBASE-ER4, 400GBASE-ER8 30km proposals will have reasonable TX OMA (min).
  • 50GBASE-LR, 50GBASE-ER have not changed their suffixes to LR1 and ER1 to align with 802.3 convention established this year.

 

TABLE 1

~Baud Rate

Reach

TX OMA

TX OMA – TDECQ

TDECQ

~CD

CDP FM

~CD

CDP FM

Codes

each lane

(max)

each lane (min)

each lane (min)

(max)

(min)

for CD (min)

(max)

for CD (max)

Gbaud

km

dBm

dBm

dB

ps/nm-km

normalized

ps/nm-km

normalized

4WDM-40 (100G ER4f)  (TDP not TDECQ)

25

40

0.5

-0.5

3

-3

12

1

4

50GBASE-LR

25

10

-1.5

-2.9

3.2

-2

2

1.5

2

50GBASE-ER

25

40

3.4

2

3.2

-2

8

1.5

6

100GBASE-LR1

50

10

0.7

-0.6

3.4

-2

8

1.5

6

200GBASE-LR4

25

10

-0.4

-1.7

3.4

-3

3

1

1

200GBASE-ER4

25

40

3.4

2

3.2

-3

12

1

4

400GBASE-LR8

25

10

0.2

-1.1

3.1

-5

5

1

1

400GBASE-ER8

25

30

0.2

-1.2

3.4

-5

15

1

3

400GBASE-ER8

25

40

2.4

1

3.4

-5

20

1

4

400GBASE-LR4

50

6

0.2

-1.1

3.5

-6

14

3

7

400GBASE-LR4

50

10

0.2

-1.1

3.5

-6

24

3

12

 

Table 2 and 3 below explains the dynamics of the review process that happened in 802.3cu and 802.3cn.

 

In 802.3cu the proponents showed favorable data for 400GBASE-LR4 10km using TX favorable to the proposal. The opponents showed unfavorable data using TX unfavorable to the proposal. Therefore, the Task Force had a complete data set for its review of Technical Feasibility.

 

In 802.3cn the proponents showed favorable data for 400GBASE-ER8 40km using TX favorable to the proposal. No unfavorable data was shown because no other type of TX was used. Therefore, the Task Force had an incomplete data set for its review of Technical Feasibility. And the missing data is exactly in the area of concern as identified by Chromatic Dispersion Penalty Figure of Merit analysis. Further, the data presented in 802.3cu for 400GBASE-LR4 10km, which has similar CDP FM for CD (min) to 400GBASE-ER8 40km was determined by the 802.3cu to have insufficient margin to be considered Technically Feasible. Therefore, 802.3cn has not established Technical Feasibility.

 

TABLE 2

EML TX

EML TX

SiPIC TX

802.3cu area of concern

high positive chirp

moderate positive chirp

no chirp

CDP effect

for CD (max)

big detriment

moderately detriment

no detriment

Proponent data brought in

No

Yes

Yes

Opponent data brought in

Yes

No

No

 

 

TABLE 3

EML TX

EML TX

SiPIC TX

802.3cn area of concern

high positive chirp

moderate positive chirp

no chirp

CDP effect for

CD (min)

big help

moderate help

no help

Proponent data brought in

Yes

No

No

Opponent data brought in

No

No

No

 

802.3cn needs to go back and take more measurements to establish a complete data set to be in a position to determine Technical Feasibility.

 

Thank you

 

Chris

 

 

Category

Page

Sub-clause

Line #

Comment

Suggested Change

General

22

45

1

50GBASE-FR, 50GBASE-LR, 50GBASE-ER need 1 suffix to be consistent with 802.3 convention, pages 22 to 27,  lines 1 to 54 pn each page

Change to 50GBASE-FR1, 50GBASE-LR1, 50GBASE-ER1 everywhere in the document

Technical

45

122

16

400G ER8 can only support 30km, lines 16 to 18

Keep 200GBASE-ER4 2 m to 30 km || 2 m to 40 kma, Change 400GBASE-ER8 to 2 m to 30 km only, make same changes everywhere else in the entire document

Technical

46

122

21

Outer Optical Modulation Amplitude (OMAouter), each lane (min)b for 200GBASE-ER4 of 3.4 dBm it too high, lines 21 to 22

Change Outer Optical Modulation Amplitude (OMAouter), each lane (min)b for 200GBASE-ER4 to 1.2 dBm

Technical

46

122

27

Launch power in OMAouter minus TDECQ, each lane (min): for extinction ratio > 4.5 dB for 200GBASE-ER4 of 2 dBm is too  high, lines 27 to 28

Change Launch power in OMAouter minus TDECQ, each lane (min): for extinction ratio > 4.5 dB for 200GBASE-ER4 to -0.2 dBm, Change all other limits in this table consistent with the OAM (min) reduction by 2.2dB.

Technical

48

122

18

TX OAM (min) reduced by 2.2dB in Table 122-9, lines 18 to 54

Change all limits in this table consistent with the TX OAM (min) reduction by 2.2dB in Table 122-9

Technical

47

122

26

Outer Optical Modulation Amplitude (OMAouter), each lane (min)b for 400GBASE-ER8 of 2.4 dBm is too high, lines 26 to 27

Change Outer Optical Modulation Amplitude (OMAouter), each lane (min)b for 400GBASE-ER5 to 0.2 dBm

Technical

47

122

32

Launch power in OMAouter minus TDECQ, each lane (min): for extinction ratio > 4.5 dB for 200GBASE-ER4 of 1 dBm is to high, lines 32 to  33

Change Launch power in OMAouter minus TDECQ, each lane (min): for extinction ratio > 4.5 dB for 200GBASE-ER4 to -1.2 dBm, Change all other limits in this table consistent with the OAM (min) reduction by 2.2dB.

Technical

49

122

10

TX OAM (min) reduced by 2.2dB in Table 122-10, lines 10 to 54

Change all limits in this table consistent with the TX OAM (min) reduction by 2.2dB in Table 122-10

General

73

139

1

50GBASE-FR, 50GBASE-LR, 50GBASE-ER need 1 suffix to be consistent with 802.3 convention, pages 73 to 87, lines 1 to 54 on each page

Change to 50GBASE-FR1, 50GBASE-LR1, 50GBASE-ER1 everywhere in the document

 


To unsubscribe from the STDS-802-3-B10K list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10K&A=1

--- Begin Message ---

All,

Comments have been received via the 60-day public review of IEEE P802.3cn draft D3.0. 

 

Therefore the Interim Teleconference call will happen 24-Sept-2019, 10:00 to 12:00pm ET. 

As noted in the meeting announcement email (sent 5-Aug-2019) and during the IEEE P802.3cn presentation given at last week’s IEEE 802.3 WG Meeting , registration for the meeting is required.  To register, please send an email to the chair at jdambrosia@xxxxxxxx.  Please note that registration closes on Thursday, 19 September 2019.

 

An additional note - I will be traveling next week, and while I don’t anticipate problems with connectivity, as chair I need to anticipate any potential issues.  Therefore, if I should experience a connectivity issue during the call, Pete Anslow will assume my responsibilities for chairing the call.

 

Regards,

 

John D’Ambrosia

Chair, IEEE P802.3cn Task Force

 


To unsubscribe from the STDS-802-3-B10K list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10K&A=1


--- End Message ---