Re: [8023-10GEPON] Define TDP values

[Frank Effenberger <feffenberger@xxxxxxxxxx>]

Dear Pete,

 

Thank you for your comments.  Let me just add some extra comments:

 

1. On the numbers from the PX-20 budget – the group took a look at those a few meetings ago, and found the same difficulties as you mention.  In fact, we had Piers Dawe take a look at those numbers, and he couldn’t quite explain how they got to be the way they are.  There were small discrepancies, as you mention.  Even more disturbing was the fact that the discrepancies were positive for PX-20 and negative for PX-10.   At that time, the discussion suggested that maybe the values that ended up in the budget were “rounded” or otherwise modulated.  So, we stopped trying to do math at that point. 
2. On the exact definition of TDP from the past and how we are using it now, I cannot claim that they are identical.  However, what you say does fit with another comment I heard a few meetings ago that “stressed sensitivity” does not include jitter-related impairments (that is, stressed sensitivity includes “vertical eye closure” only).  Indeed, if that is the case, then we should make a move to include “horizontal eye closure” in our definition.  If we need to create a new term, like “Really stressed sensitivity” (!), then I’m fine with that.  

 

Sincerely,

Frank E.  

 

 

 

 

---

From: Pete Anslow [mailto:pja@NORTEL.COM]
Sent: Tuesday, December 04, 2007 6:31 AM
To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] Define TDP values

 

Dear Frank,

 

I agree with most of what you say.

 

One place where I think that your statement is not generally true, however, is when you say:

 

IEEE stress sens. = worst-case Tx + worst-case path + worst-case Rx 

 

If one takes the values for 1000BASE-PX20-D from clause 60.4 we have:

 

Transmitter and dispersion penalty (max) 2.3 dB

Receiver sensitivity OMA (max) -26.2 dBm

Stressed receive sensitivity OMA (max) -23.6 dBm

 

So the difference between Receiver sensitivity and Stressed receive sensitivity is 2.6 dB

 

Also in clause 58.7.11.1 it says:

“The optical power penalty for the stressed eye is intended to be similar to its vertical eye closure penalty. This is not necessarily the same as the highest TDP anticipated in service, but represents a standardized test condition for the receiver.”

 

I think that you can choose to make the stressed Rx sens so that the value of TDP is the difference between the stressed sensitivity and the “ideal sensitivity”, but this is not generally true in IEEE budgets.

 

Regards,

Pete Anslow

 

Nortel Networks UK Limited, London Rd, Harlow, Essex CM17 9NA, UK

External +44 1279 402540 ESN 742 2540

Fax +44 1279 402543

 

---

From: Frank Effenberger [mailto:feffenberger@HUAWEI.COM]
Sent: 03 December 2007 18:30
To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] Define TDP values

 

Dear Discussing parties:

 

I would submit that all of this is a huge confusion – and before we change the spreadsheet, I want to make sure that we all are in complete agreement on the definition of terms, and what use they are.  

 

For the spreadsheet as it is: we have presented how this spreadsheet and the budget calculations are supposed to work.  While there are lots of things in that spreadsheet, the only thing that matters is this:  Tx_OMA_min = Channel_loss + Stressed_Rx_sensitivity_OMA.  

For example, if we launch at +5, and the channel loss is 15, then the stressed sensitivity is -10.   This is the simplest budget calculation possible.

 

I believe that the definitions of Tx_OMA_min and channel loss are self-evident.  

As for the meaning of “stressed RX sensitivity”, this is the sensitivity that I would measure with a worst case Rx, worst case Tx, and worst case optical path.  

 

So, to the definition of TDP.  The value of TDP given here is the difference between the stressed sensitivity (above) and the “ideal sensitivity”. 

The ideal sensitivity is already in the spreadsheet, named simply “sensitivity”.  

This “ideal sensitivity” is what I would measure with a worst case Rx, but a best case Tx (perfect pulses) and a best case optical path (that is, an attenuator!).

Note that a perfect Tx is expressed in terms of OMA, so the ER penalty *is* captured.  

 

Currently, TDP is manually entered, and the ideal sensitivity is calculated to fit.   If people want to reverse that, and enter the ideal sensitivity and calculate the TDP, well, go ahead – it will make no difference at the end of the day.  

 

As for what TDP does for us, it is primarily a means of controlling the transmitter imperfections, some of which are ‘stand-alone’ (like waveform imperfections), and some of which are interactions with the optical path.  The IEEE idea is that we throw all of it into a common bucket, and let the Tx builder optimize his heart out.  

 

In contrast, the ITU way keeps them separate.  The ITU specifies an optical path penalty, and the “transmitter penalty” is folded into the ITU concept of Rx_sensitivity.  The ITU Rx Sensitivity is measured with a worst cast Tx but a best case optical path (i.e., an attenuator).  So, it’s all there, just re-arranged. 

 

In summary, the following table of definitions can be stated:

IEEE       sensitivity = best-case Tx + best-case path + worst-case Rx

ITU       sensitivity = worst-case Tx + best-case path + worst-case Rx

IEEE stress sens. = worst-case Tx + worst-case path + worst-case Rx 

 

Now: if people want to break up the TDP into “transmitter penalty” and “Dispersion penalty” (which is very close to the optical path penalty), then fine.  This is, in fact, practically the ITU way of doing things.  But, I thought that the IEEE method was somewhat better for low-cost reasons.  

 

Sincerely,

Frank Effenberger

 

 

---

From: Ken Maricondo [mailto:mariconk@GMAIL.COM]
Sent: Monday, December 03, 2007 10:39 AM
To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] Define TDP values

 

Pete,

 

I think that there is some misunderstanding.  I do not disagree with your earlier response to the definition of Transmitter_+_Dispersion_Penalty (TDP), nor am I opposed to having clear definitions of each value in the standard and spreadsheet; but what I was suggesting was a compromise in the absence of hard data for the TDP.   As I understand propose of the current spreadsheet, the spreadsheet is designed to show a worse case scenario, not every minute value and calculation.  

In response to your current thread, what you have pointing out is a more refined definition of the TDP value which is not part of the scope of the 802.3av task force, as I understand it.  What I have personally done is modify the spreadsheet to include the values that I feel are pertinent.   

 

Best regards,

 

Ken

 

On Dec 3, 2007 4:11 AM, Pete Anslow <pja@nortel.com> wrote:

Ken,

 

The equation you propose for row 53 still does not make sense.  TDP and ITU_Optical_Path_Penalty are not equivalent measures since TDP includes the non-ideality of the transmitter waveform and ITU_Optical_Path_Penalty does not.

 

If the group wants to call out the two components of TDP separately then we need:

 

A user input cell for Maximum Transmitter Penalty TP (penalty due to non-ideal eye shape at the transmitter)

A user input cell for Maximum Dispersion Penalty DP (further penalty caused by the link dispersion)

A calculated cell for Maximum TDP (TDP = TP + DP)

A cell which calculates the actual dispersion penalty (which is already there)

 

I think that it would probably be a good idea to remove the reference to ITU_Optical_Path_Penalty from the DP row as it seems to cause confusion rather than help.

 

The test in row 53 would then become Is Calculated_dispersion_penalty <= DP?

 

We have no way of calculating an estimate of TP, so there is no way to make this test involve TDP rather than DP alone.

 

As discussed earlier, if an additional cell for achievable receiver sensitivity was to be introduced then a second test for the overall power budget could be added.

 

Is Min_Tx_Pow – TDP – Max_channel_loss > Rx_Sens?

 

The Min_Tx_Pow and Rx_Sens would be in OMA and Rx_Sens would have to be referred to the sensitivity you would get with an ideal transmitter for this to work.

 

Regards,

Pete Anslow

 

Nortel Networks UK Limited, London Rd, Harlow, Essex CM17 9NA, UK

External +44 1279 402540 ESN 742 2540

Fax +44 1279 402543

 

---

From: Ken Maricondo [mailto: kmaricondo@IEEE.ORG]
Sent: 30 November 2007 04:17

To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] Define TDP values

 

Dear All,

 

Thank you all for the feedback and clarification.  In reviewing all of the recent threads, I think that we all can agree that a penalty is a penalty.   Therefore, I suggest that the spreadsheet be change to reflect the following:

1.        Change the TDP cell (A39) to only be TP

2.        Change the Transmitter Dispersion Penalty cell (D39) to Transmitter Penalty

3.        Change Dispersion_Penalty <= ITU_Optical_Path_Penalty cell (A53) to Transmitter_+_Dispersion_Penalty <= ITU_Optical_Path_Penalty or TDP <= ITU_Optical_Path_Penalty

4.        Change cell (D53) formula to =IF((B38+B39)<=B30,"PASSED","FAILED")

 

In the absence of hard TDP data, I think that this will allow the user to put in a transmitter penalty value, while other users who might think that the TP is overkill can put in a value of zero.  At a later date and when TDP data is available, I think that we can readdress this issue.  What do you think?

 

Best regards,

 

Ken Maricondo

 

 

On Nov 29, 2007 3:10 PM, Frank Chang <ychang@vitesse.com> wrote:

Hi Pete;

 

I am very happy you chime in to clarify the confusion which exists for a while in the email thread and also associated mtg discussions so far. I also feel the term "TDP" or even "stress RX sens" was misinterpreted in link budget formalism, which is quite inconsistent with what is defined in IEEE 802.3 for the TP2 and TP3 methodology.  

 

FYI- In line with what you said, actually I provided a tutorial to elaborate this during July mtg as follows:

http://www.ieee802.org/3/av/public/2007_07/3av_0707_chang_1.pdf

 

Best Regards

Frank C.

-----Original Message-----
From: Pete Anslow [mailto: pja@nortel.com]
Sent: Thursday, November 29, 2007 3:13 AM
To: STDS-802-3-10GEPON@listserv.ieee.org
Subject: Re: [8023-10GEPON] Define TDP values

Hi,

The way the term " TDP" is being discussed in this thread seems to me to be inconsistent with the way it is defined in IEEE 802.3.

TDP stands for Transmitter and Dispersion Penalty.   It is the penalty due to the combination of the eye closure of the transmitter and the further eye closure caused by the link dispersion.

The TDP measurement procedure for 1000Base PX10 and PX20 is defined in subclause 58.7.9 .  The sensitivity of the reference receiver is measured with as near an ideal test transmitter as possible and then this is corrected for any residual transmitter eye closure to give the sensitivity with an ideal transmitter S.  Then the receiver sensitivity is measured again using the transmitter under test through the worst case dispersion .  T he TDP value is then the difference between the second measurement and S.

If we label the two penalty components as EP for the transmitter Eye Penalty (penalty due to non-ideal eye shape at the transmitter) and DP for the transmi tter Dispersion Penalty (further eye closure caused by the link dispersion ) then we can say:

TDP = EP + DP

Now, for most ITU-T power budgets Path Penalty is approximately equal to DP (and the specified receiver sensitivity has to be met using a transmitter with a worst case EP ).

Consequently, Dispersion_Penalty <= ITU_Optical_Path_Penalt y makes reasonable sense.

The inequality (Dispersion_Penalty+TDP) <= ITU_Optical_Path_Penalt y makes no sense at all as it is roughly equivalent to saying:

DP + (EP + DP) < = DP

I agree that Dispersion_Penalty <= ITU_Optical_Path_Penalt y is not sufficient to establish that the power budget is feasible with available optics.  As I understand the current spreadsheet, it calculates the receiver sensitivity that would be required given the various input parameters.  In order to test for the feasibility of this sensitivity value an additional input cell containing the achievable receiver sensitivity with a n ideal transmitter would be required.

Regards,

Pete Anslow

Nortel Networks UK Limited, London Rd, Harlow, Essex CM17 9NA, UK

External +44 1279 402540 ESN 742 2540

Fax +44 1279 402543

_____________________________________________
From: Hiroshi Hamano [ mailto:hamano.hiroshi@JP.FUJITSU.COM]
Sent: 29 November 2007 02:55
To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] Define TDP values

Dear Dr. Maricondo,

Thank you very much for your explanations, and I apologize for my late reply.

Now, I understand that, in your E-mail, (Dispersion_Penalty+TDP) means the dispersion penalty value using a realistic transmitter with worst case TDP, in contrast to that using an ideal or perfect transmitter with small or no TDP.

I am not sure, whether (Dispersion_Penalty), in the Spreadsheet, is figured out based on such an ideal transmitter or not, but I agree that the result should indicate the worst case value in order to d ecide the fail/pass condition.

But I am not sure either, how TDP should be counted into such transmitter parameters for calculating (Dispersion_Penalty+TDP), in the Spreadsheet, and how big its impact will be.  If you have any suggestions, that will be qu ite helpful.

If TDP value should be derived from the measurement results, not from Spreadsheet calculations, some penalty value may remain only assumed and uncalculated, unless the relationship between the measured TDP and Dispersion_Penalty is justified.

Any comments or discussions will be highly appreciated.

Best regards,

Hiroshi Hamano

%% "Ken Maricondo" <kmaricondo@ieee.org> %% Re: [8023-10GEPON] Define TDP values %% Fri, 23 Nov 2007 17:07:53 -0500

>

> Dear Hamano-san and Hajduczenia,

>

> I agree (from all that I have read) that the ITU_optical_path_penalty

> basically includes no transmitter penalty and that receiver

> sensitivity value in ITU formalism should share the TDP within the

> margin. The rationale as to why I suggested adding the TDP to

> *Dispersion_Penalty <=

> ITU_Optical_Path_Penalty* is as follows:

>

> 1.    Any penalty such as chirp, extinction ratio, MPN, etc., will require

> that the photodiode receiver to receive an increased proportional

> optical receive level in order to maintain the same BER verse a system

> without the same penalties. The *Dispersion_Penalty <=

> ITU_Optical_Path_Penalty* from what I can tell assumes an ideal

> transmitter and receiver over a given optical path and that only if

> the dispersion penalty exceeds the optical path penalty then the link

> fails.  If I use only dispersion penalty in a network calculation

> without the TDP, then I am not truly taking into account worse

> case/end of life network performance.  In my point of view, a network

> designed with a TDP of 3dB for example will have a shorter operational

> lifetime or be performance limited then an identical network with an

> ideal transmitter with no TDP or lower value of TDP.  So to compensate for the TDP a more robust FEC scheme or higher quality receiver might be in order.

>

> 2.    I do not discount or object to the TDP value from being subtracted

> from the *IEEE_Rx_Stressed_Sensitivity_OMA* to get *IEEE_Rx_Sen_OMA*,

> but I think that TDP should show up in the system margin as worse

> case/end of life calculation within the spreadsheet.  The fact that

> the receiver sensitivity has to go lower to compensate for the TDP

> only points out that I have to have a higher quality ideal receiver at first glance.

>

> I am acutely aware of the fact that 10GEPON standard will allow for

> degree of flexibility in the network design to compensate for network

> design short comings.  My suggestion for *(Dispersion_Penalty+TDP) <=

> ITU_Optical_Path_Penalty* to be changed is to make sure that there is

> no ambiguity in the standard (spreadsheet) and to point out to the

> adopter of the 10GEPON standard that *all* penalties have been

> accounted for, analyzed and documented.  At the end of the day it is

> up to task force as whole to adopt what they feel is appropriate for the standard.

>

> Best regards,

>

> Ken Maricondo

>

>

>

>

> On Nov 22, 2007 5:27 AM, Hiroshi Hamano

> <hamano.hiroshi@jp.fujitsu.com>

> wrote:

>

> > Dear Dr. Maricondo and Dr. Hajduczenia,

> >

> > Thank you for your quick response and discussion.

> > I am not sure how TDP values have been defined in the previous

> > specifications such as IEEE 802.3ae and 802.3ah.  If they have also

> > reflected the vendor data based on the transmitter measurement

> > results, vendor feedbacks may be quite important similarly for

> > 10GE-PON.

> >

> > > but should read *(Dispersion_Penalty+TDP) <=

> > > ITU_Optical_Path_Penalty* to insure all possible noise and

> > > penalties

> > are

> > > accounted for.

> >

> > Perhaps, I misunderstand the last sentence of Dr. Maricondo's E-mail.

> > But my understanding is that ITU_optical_path_penalty basically

> > includes no transmitter penalty.  Receiver sensitivity value in ITU

> > formalism should share the margin, instead, for possible transmitter

> > penalty compared to the ideal one.

> > Would you please explain again your intension about the sentence??

> >

> > Best regards,

> > Hiroshi Hamano

> > Fujitsu Labs. Ltd.

> >

> > %% "Ken Maricondo" <kmaricondo@ieee.org> %% Re: [8023-10GEPON]

> > Define TDP values %% Wed, 21 Nov 2007 19:15:31 -0500

> >

> > >

> > > Dear Hamano-san,

> > >

> > > Although the TDP is a controversial value to be added to the

> > > 10GEPON standard, I have to agree with the previous committee (

> > > 802.3ah) for assigning a TDP value; albeit not an apparent

> > > benefit, the TDP does set

> > a

> > > reference value/point for determining transmitter quality which

> > > does

> > impacts

> > > system performance.  I agree with your assessment that the lack of

> > > high power reference transmitter to make a TDP measurement at this

> > > time is a problem.  In the absence of such a reference

> > > transmitter/s, I think that

> > the

> > > TDP values you have chosen are a good reference point to start

> > > with and

> > I

> > > support you on this issue.

> > >

> > > I also think that the spreadsheet should reflect the impact of TDP

> > > on

> > the

> > > systems' overall performance.   From what I have been able to determine,

> > the

> > > TDP value is subtracted from the

> > > *IEEE_Rx_Stressed_Sensitivity_OMA* to

> > get *

> > > IEEE_Rx_Sen_OMA*, which does not readily translate into a system

> > performance

> > > impact/limitation.  The system pass/fail calculation is based on

> > > *Dispersion_Penalty <= ITU_Optical_Path_Penalty* only, but should

> > > read

> > *(Dispersion_Penalty+TDP)

> > > <= ITU_Optical_Path_Penalty* to insure all possible noise and

> > > penalties

> > are

> > > accounted for.

> > >

> > > Best regards,

> > >

> > > Ken Maricondo

---

-----------------------------------------

Hiroshi Hamano

Network Systems Labs., Fujitsu Labs. Ltd.

Phone:+81-44-754-2641 Fax.+81-44-754-2640 E-mail:hamano.hiroshi@jp.fujitsu.com

-----------------------------------------