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

Re: [8023-10GEPON] PIN-PD Power Budget



Hello, Marek
  I would like to answer your concerns as Rx sensitivity issues.

1. The sensitivity improvement of PIN receiver mostly depends on TIA technology progress. However, the any TIA technology improvement could affect to APD Rx sensitivity improvement too. According to current TIA Ics technology situation, we assume 1-2dB sensitivity improvement in both PIN and APD in 2-3 years. The 2dB sensitivity improvement of APD sensitivity could make us to use normal EML replacing high power EML in OLT or eliminating FEC chip in ONU, which will make cost down significantly. But the 2dB improvement doesn't make any difference to EDFA-PIN.
   However, could we get the improvement of 2dB sensitivity from TIA? From current technology, it's difficult.

2. Current PIN/APD sensitivity:
   The current good (Best)PIN Rx sensitivity (BER=1E-12) is typical -20dBm at beginning of life and high extinction ratio in mass production. But in over conditions and EOL, the spec may be -18dBm or worse in production level. 
   In module level, the spec of max. -16dBm at EOL and over conditions is a good number in production. The sensitivity of module at over conditions and EOL is the most important reference spec for us to define the power budget.
   The good APD sensitivity: typ about -28dBm, -26dBm in over conditions and EOL.
   Module: -24dBm@EOL and -26dBm in some conditions.

3. New technologies:
   I don't know what kind of new technology you are talking about. As my knowledge, there are few technologies to improve sensitivity.

 (1). TIA: as I said above, the sensitivity improvement is limited and difficult to solve the high launch power issues.
 (2). "Super PIN": This could improve the "PIN" sensitivity a lot. However, this is a kind of "APD", not a real "PIN". You may know "APD" is "a special 'PIN' with internal gain". So any additional function to improving "PIN self sensitivity", the low cost advantage of "PIN" may disappear.
 (3). SOA+PIN: The approach could make PIN to have the sensitivity as APD or above, but it's not a low cost solution. Some people aspect this technology applied to 100G. 

4. The cost roadmap of 10G APD: Please see the material 3av_0707_Feng_2.pdf in SF meeting.
   Actually, I hear from GPON maker, from GPON case, the cost ratio of GPON Triplexer (APD+DFB) and BPON Triplexer (PIN+FP) is about 1.5 in 2008. From GPON history, we could image 10GE PON's cost trend.

I am glad of discussing any issues with you at any time. Please let me know if you have any questions.

Best regards
Hao Feng
Eudyna Devices USA

-----Original Message-----
From: Hajduczenia, Marek [mailto:marek.hajduczenia@SIEMENS.COM] 
Sent: Monday, September 03, 2007 1:50 AM
To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] PIN-PD Power Budget

Dear Hamano-san,
Thank You for the clarification of this topic. I believe we might have been riding on the bubble of the potential PIN evolution which was already mentioned several times and which seemed to indicate that PINs can get better in the next 2-3 years (I think this was mainly mentioned during discussion and not included formally in any presentation). Should the scenario You mention be what we have to cope with, it all comes down to selection of PIN or APD. That also means that either group must persuade the large share of the TF on the selection of the device for the ONU, thus defining the power budget for the system and Rx/Tx parameters. I would like to hear more opinions on the feasibility of any improvements in termas of PIN sensitivity in the following years. Should there be any, we will have to either live with high launch power at the OLT or slightly higher initial ONU cost (providing that the APD cost curves do come down nicely as quoted in the presentations). 
Are there any members of the PIN-Team who are willing to enligthen us on the potential sensitivity improvements in the following years ? I do not think we need a presentation on this - it would be rather good to know where the experts in the field expect to reach in the years to come. 
Thank You for the discussion

Marek Hajduczenia (141238)
NOKIA SIEMENS Networks S.A., Portugal - R Rua Irmãos Siemens, 1 Ed. 1, Piso 1 Alfragide
2720-093 Amadora
Portugal
* Marek.Hajduczenia@siemens.com
http://www.marekhajduczenia.info/index.php
(+351.21.416.7472  4+351.21.424.2082
"C makes it easy to shoot yourself in the foot; C++ makes it harder, but when you do, it blows away your whole leg." - Bjarne Stroustrup 

-----Original Message-----
From: Hiroshi HAMANO [mailto:hamano.hiroshi@JP.FUJITSU.COM]
Sent: segunda-feira, 3 de Setembro de 2007 7:04
To: STDS-802-3-10GEPON@LISTSERV.IEEE.ORG
Subject: Re: [8023-10GEPON] PIN-PD Power Budget

Dear Dr. Parruck and Dr. Hajduczenia,

Thank you for your discussions for solving the PIN/APD issue.
Perhaps I am not an appropriate person to discuss, because some PIN group member should be responsible.  But I try some, because my explanation, in my former E-mail, was perhaps not good and misunderstood.

There is 8dB sensitivity difference between 10G PIN and APD, including about 1-1.5dB E-FEC gain difference.  It is the 10G optics background and the main reason for the long PIN/APD battle.  
Your discussions are ignoring this background and only driving the PIN sensitivity close to APD, but I am afraid this approach is not fruitful.
As for the power budget spec. lists, all the vendors carefully investigated the tight B++ requirement and decided the spec. numbers with their optics experts, who have huge production experience of XFPs and other 10G transceivers.  I believe they started their assumption with the typical PIN-ROSA alone sensitivity of around -21dBm@10-12, same as that Dr. Parruck has indicated in his E-mail, and they decided, in the end, the ONU sensitivity spec. -20dBm with E-FEC.  
They have no room to improve as long as they assume current PIN-ROSA performance.

The following shows the PIN receiver sensitivity experimental results on the page 7 of 3av_0611_Chang_1.pdf.
3	10Gb/s	-19dBm		PIN-HBT		1998
4	10Gb/s	-20.4dBm	PIN-HBT		1995
5	10Gb/s	-22.4dBm	PIN-HEMT	1996
6	10Gb/s	-23.5dBm	PIN-HEMT	1995
I do not know other reports so far breaking this record -23.5dBm, and that means there were no big improvements for more than 10 years.  
And also future 10G PIN-ROSA sensitivity improvement seems limited unless there comes a drastic breakthrough.

Your discussions, requesting the PIN-ONU sensitivity lower, seem to me that you are indirectly persuading the PIN group to give up PIN and to choose APD, but I do not think it is acceptable for them.

This shows only my personal understanding, please ask some PIN group member for clear answer.

Best regards,
Hiroshi Hamano

%% "Hajduczenia, Marek" <marek.hajduczenia@SIEMENS.COM> %% Re: [8023-10GEPON] PIN-PD Power Budget %% Fri, 31 Aug 2007 10:36:22 +0100

>  
> Dear Bidyut,
> That would be splendid since it would allow us to end the PIN/APD wars with a single neat aggressive power budget which would satisfy everyone. 
> The problem is that we seem to get not much closer to consensus than we were during the last two meetings. I am not a PIN / APD expert but isn't it possible to create a single proposal which could satisfy both camps and allow the rest of the group to back up the proposal ? 
> Forgive my ignorance if such a solution is impossible due to some 
> obvious reasons I am not aware of
> 
> Marek Hajduczenia (141238)
> NOKIA SIEMENS Networks S.A., Portugal - R Rua Irm竢s Siemens, 1 Ed. 1, 
> Piso 1 Alfragide
> 2720-093 Amadora
> Portugal
> * Marek.Hajduczenia@siemens.com
> http://www.marekhajduczenia.info/index.php
> (+351.21.416.7472  4+351.21.424.2082
> "C makes it easy to shoot yourself in the foot; C++ makes it harder, 
> but when you do, it blows away your whole leg." - Bjarne Stroustrup
> 
> -----Original Message-----
> From: Bidyut Parruck [mailto:Bidyut@CORTINA-SYSTEMS.COM]
> Sent: quinta-feira, 30 de Agosto de 2007 19:17
> To: STDS-802-3-10GEPON@listserv.ieee.org
> Subject: Re: [8023-10GEPON] PIN-PD Power Budget
> 
> Hi Hamano-san,
> 
> My apologies - I didn't look for the PIN-PD receivers in July. However 
> in light of where we are with respect to power budget, the discussion 
> is relevant and worth revisiting.
> 
> Some of the same PIN-PD vendors, who claim RX sensitivity of -20 dBm 
> for BER of 1x10E-3, publish datasheets showing -25 dBm for BER of 1x10E-3.
> If PIN-PD vendors could agree to a better sensitivity, we could all 
> agree to one power budget. Could we hear from some of them if they are 
> willing to move from -20 dBm?
> 
> Bidyut
> 
> -----Original Message-----
> From: Hiroshi HAMANO [mailto:hamano.hiroshi@JP.FUJITSU.COM]
> Sent: Wednesday, August 29, 2007 8:20 PM
> To: STDS-802-3-10GEPON@listserv.ieee.org
> Subject: Re: [8023-10GEPON] PIN-PD Power Budget
> 
> Dear Sirs,
> 
> Thank you for discussing the power budget table in 
> 3av_0705_takizawa_1.pdf.
> 
> For your interest, I copied again the E-mail discussion with 
> Dr.Effenberger below.
> You can also find it in the E-mail reflector archive.
> 
> http://www.ieee802.org/3/10GEPON_study/email/msg00644.html
> 
> Best regards,
> Hiroshi Hamano
> 
> ------- Forwarded Message
> %% Hiroshi HAMANO <hamano.hiroshi@JP.FUJITSU.COM> %% Re: 
> [8023-10GEPON] [POWER BUDGET] resolving differences over PIN vs.
> APD
> %% Tue, 10 Jul 2007 20:48:35 +0900
> 
> Dear Dr. Effenberger,
> 
> I cannot understand why you say that there exists unfairness between 
> PIN
> 
> and APD sensitivities.  I believe both the RX sensitivities are quite 
> fair enough with regard to production feasibility.
> No such arguments arise among vendors here, because they always think 
> about both receiver types seriously and check both budget tables.
> If the RX-type will be once decided, all the vendors have to put the 
> one
> 
> in production anyway, even though their RX preference is the other.
> 
> If you think that PIN-RX sensitivity in our vendor summary looks 
> conservative, same as Dr. Frank Chang thought that APD-RX looks so --- 
> he pointed that out at the last Geneva meeting ---, I understand 
> somewhat both of your feelings, considering such a tight Class B++ 
> power budget.
> But please understand the 10G transceiver production facts, for 
> instance, XFPs.
> It is not the laboratory experiments or single ROSA performance.  
> Vendors also have to spend a lot care for product deviations, 
> especially
> 
> for ONU, to avoid yield which makes its cost jump up.
> Besides, as it is the specification of ONU sensitivity, additional 
> unknown WDM filter loss onto the bare transceiver cannot be ignored, 
> which Takizawa always comments on his power budget presentation.
> I once suggested that a slight change (0.5dB) for RX sensitivity, and 
> got a strong booing from all the vendors.
> So, the vendors cannot change their RX sensitivities of both PINs and 
> APDs at moment.
> 
> Best regards,
> Hiroshi Hamano
> 
> ---------------------
> 
> %% Bidyut Parruck <Bidyut@CORTINA-SYSTEMS.COM> %% Re: [8023-10GEPON] 
> PIN-PD Power Budget %% Wed, 29 Aug 2007 16:59:41 -0700
> 
> > 
> > 
> > 
> > Hi Frank,
> > 
> >  
> > 
> > Thanks for clearly stating the issue. The datasheets I'm "casually 
> > looking on the web" are from "major" vendors.
> > 
> >  
> > 
> > Both examples I showed had a sensitivity of -21 dBm for BER of
> 1x10E-12.
> > I can understand 2-3 dB of margin - our current equations are using 
> > 5 dB. I do understand, initially, this could increase the cost of
> PIN-PD.
> > But in the long run the same companies will innovate and bring the
> cost
> > down. I think we should revisit this subject. 
> > 
> >  
> > 
> > Regards,
> > 
> >  
> > 
> > Bidyut
> > 
> >  
> > 
> > From: Frank Effenberger [mailto:feffenberger@HUAWEI.COM]
> > Sent: Wednesday, August 29, 2007 4:16 PM
> > To: STDS-802-3-10GEPON@listserv.ieee.org
> > Subject: Re: [8023-10GEPON] PIN-PD Power Budget
> > 
> >  
> > 
> > All,
> > 
> >  
> > 
> > The 1e-12 BER PIN sensitivity that is being used in the slide below 
> > is
> > -16 dBm.  
> > 
> > The slide says that the 1e-4 BER sensitivity is -19 dBm.  
> > 
> > The slide says that the 1e-3 BER sensitivity is -20 dBm.  
> > 
> >  
> > 
> > What Bidyut is noticing is that many receivers he is casually 
> > looking
> at
> > on the web seem to have a much better sensitivity curve than what 
> > this slide says.  The typical sensitivity numbers are -20dBm, and 
> > I'd
> believe
> > that the 'hard spec' would be about -18 dBm.  I've been saying the
> same
> > thing for many months.  But "team PIN" seems convinced on -16 dBm, 
> > for reasons of economy and yield, I've been told.  But that is a 
> > matter of how aggressive you want to be with the specification....
> > 
> >  
> > 
> > Sincerely,
> > 
> > Frank E.  
> > 
> >  
> > 
> >  
> > 
> >  
> > 
> > ________________________________
> > 
> > From: Bidyut Parruck [mailto:Bidyut@CORTINA-SYSTEMS.COM]
> > Sent: Wednesday, August 29, 2007 6:05 PM
> > To: STDS-802-3-10GEPON@listserv.ieee.org
> > Subject: Re: [8023-10GEPON] PIN-PD Power Budget
> > 
> >  
> > 
> > Hi Ryan,
> > 
> >  
> > 
> > I'm not sure where the FEC numbers are baked in. Here's the slide 8
> from
> > 3av_0705_takizawa_1.pdf.
> > 
> >  
> > 
> > Bidyut
> > 
> >  
> > 
> >  
> > 
> >  
> > 
> > From: Ryan Hirth [mailto:ryan.hirth@TEKNOVUS.COM]
> > Sent: Wednesday, August 29, 2007 2:10 PM
> > To: STDS-802-3-10GEPON@listserv.ieee.org
> > Subject: Re: [8023-10GEPON] PIN-PD Power Budget
> > 
> >  
> > 
> > Bidyut,
> > 
> > The additional gain from operating at BER 10E-12 to 10E-3 in your 
> > example is the FEC optical gain.  You cannot increase the receiver 
> > sensitivity and add gain for FEC since this would be adding in the
> same
> > gain factor twice.
> > 
> > Regards,
> > 
> >  
> > 
> > Ryan Hirth
> > 
> > Director of ASIC Engineering
> > 
> > Teknovus Inc.
> > 
> >  
> > 
> > ________________________________
> > 
> > From: Bidyut Parruck [mailto:Bidyut@CORTINA-SYSTEMS.COM]
> > Sent: Wednesday, August 29, 2007 1:31 PM
> > To: STDS-802-3-10GEPON@listserv.ieee.org
> > Subject: [8023-10GEPON] PIN-PD Power Budget
> > 
> >  
> > 
> > I have been looking at publicly available 10Gbps PON photodiode 
> > datasheets from major suppliers. For the receiver sensitivity, they
> all
> > claim -21 dBm for BER of 1x10E-12. The typical characteristics curve 
> > suggests a BER of 1x10E-3 for input power level of -25 dBm.
> > 
> >  
> > 
> > With input BER of 1x10E-3, we can find FEC algorithms which would 
> > provide output BER of 1x10E-12 or better. Excuse my ignorance of the 
> > subject, but why can't we use -24 dBm as the receiver sensitivity 
> > for PIN-PD receivers? That will give us an additional 4 dB.
> > 
> >  
> > 
> > Here are some examples:
> > 
> > 
> 
> 
> 
> 
> ---
> ---------------------------------------------
> HIROSHI HAMANO         Network Systems Labs.
> FUJITSU Labs. Ltd., Kawasaki, 211-8588 JAPAN
> TEL: +81-44-754-2641  FAX: +81-44-754-2640
> E-mail: hhlsi@flab.fujitsu.co.jp
> ---------------------------------------------
> 
> 




---
---------------------------------------------
HIROSHI HAMANO         Network Systems Labs.
FUJITSU Labs. Ltd., Kawasaki, 211-8588 JAPAN
TEL: +81-44-754-2641  FAX: +81-44-754-2640
E-mail: hhlsi@flab.fujitsu.co.jp
---------------------------------------------