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

Re: [8023-10GEPON] [CHANNEL_LINK] New Excel spreadhseetavailable ...



Dear Tatsuta-san, 
I comment inline to facilitate tracking
BR

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
 

-----Original Message-----
From: TATSUTA [mailto:tatsuta@ANSL.NTT.CO.JP] 
Sent: terça-feira, 22 de Maio de 2007 6:22
To: STDS-802-3-10GEPON@listserv.ieee.org
Subject: Re: [8023-10GEPON] [CHANNEL_LINK] New Excel spreadhseetavailable ...

Dear Marek,

I recognized one thing and I have to correct my former e-mail.

It looks that TDP in Table 60-3 and Table 60-6 of IEEE802.3ah came from a 
row "W" of the spread sheet, though PX20-1490 is different between the spec 
and the spread sheet (Maybe it is a mistake?).

<MH>
I doubt whether that is a mistake - I was not participating in the EFM Task Force and I think that it will be though right now to precisely state why the given value was selected as differing from the actual calculated value. I would bet on the problems with achieving specific parameters and the lowered production yield. That is only a wild guess and please do not treat it as a binding answer. Perhaps someone participating in the EFM could answer this question.
</MH>

And the allocation for penalties in Table 60-9 of IEEE802.3ah looks like 
calculated by TDP + 0.2dB (I do not know where +0.2dB came from).

<MH>
The politically correct version of the answer is as follows: "No clue". 
</MH>

Anyway, do we need to use the row "W" of 
"10GEPON-D[1].0.4_with_examples.xls" to estimate a power budget?
Now it is an enormously big value, Downstream of PX-30 is 4.2dB and 
Upstream is 3.5dB.

<MH>
I would say that if these values hold for 10G channel (the model was not changed and the same one is used in 10GE links), we are supposed to use them. Nevertheless, the magic trick with lowering the TDP values which You also noticed for PX20-1490 in 802.3ah indicates that this value is a pessimistic (perhaps overly pessimistic) estimation of the real TDP. Where the real value is, I guess we would have to find out by actually asking the component vendors.
</MH>

I apologize for confusing you.

<MH>
I have not been confused. There is nothing to apologize for :-)
</MH>

Sincerely yours,
Tsutomu TATSUTA


At 11:54 07/05/22, TATSUTA wrote:
 >Dear Marek,
 >
 >Sorry for my late reply.
 >
 >A following comment relates to your file.
 >>>5. Cell "B42" in file "10GEPON-D[1].0.7" is specified as 
other_penalties which represents other system penalties. However a value is 
derived from cell "L7" in file "10GEPON-D[1].0.4_with_examples", which 
represents Connections etc. I think this value should be included in ChIL 
and a penalty should come from cell "V34".
 >>>[Marek Hajduczenia] In 
<http://www.ieee802.org/3/av/public/tools/10GEPON-D.0.7h.xls>http://www.ieee802.org/3/av/public/tools/10GEPON-D.0.7h.xls, 
the cell other_penalties (B42) is a total of the non-linear penalties (B21) 
and fibre plant penatlies (B22), whic includes splices, connectors etc. 
Cell B41 (fibre_loss) includes pure fibre loss only, assuming there are no 
connectors, PSC etc. and the channel is ideal.
 >>>[Marek Hajduczenia] As far as I understand the question, what You're 
saying Tatsuta-san is that we should take the splices, connectors etc. and 
include them in the CHIL and take the system penalties as the maximum value 
calculated in cell V34. Please note however that the said system penalty 
value You refer that is column V is not used in any calculations apart from 
charting ... It is hard for me to say whether we should therefore use this 
value or not.
 >>Umm..... a little complicated.
 >>Is not A value of "V34" used for anything except for making the chart?
 >>[Marek Hajduczenia] As far as I can say .... no. In terms of 
calculations, we take the difference between the Tx power and Rx 
sensitivity, substract the fibre loss, PSC loss and additional channel loss 
(splitters etc.) and what's left is what goes for penatlies and power 
margins. I am not aware how precisely one can model the RIN penalities etc. 
That would have to be explored in depth but I would say those values are 
pessimistic at best. That is at least the impression I get from the white 
paper and discussions with Piers and Pete ... In other words, in the 
spreadsheet we estimate the maximum value of the penalities we can tolerate 
in the system and do not try to estimate their values per say ... I hope 
that is clear right now.
 >>I thought this sheet was used for calculating a power penalty as well as 
calculating a ChIL
 >>[Marek Hajduczenia] Indeed it is. It just gives You the upper bound on 
how much penalty You can tolerate in the system ... As I said before, the 
formulas governing calculation of individual penalities are approximated at 
best so do not count on really precise modelling. Matlab would have to be 
used to calculate them precisely.
 >>If you are right, how can we calculate the penalties?
 >>[Marek Hajduczenia] By calculate You mean take a formula and plugin in 
some numbers and get a result. You can estimate their values in the 
extended tables (lines 15 - 34 in D.0.4) but I am not sure about their 
precision level ... Piers would have to answer that but from what I gather 
from the white paper, the modelling is simplified to be used in Excel. You 
might need to use Matlab models to produce more precise results.
 >>
 >>Sincerely yours,
 >>Tsutomu TATSUTA
 >
 >I rechecked a relationship between IEEE802.3ah spec and the spread sheet, 
and I understood your answers. If my understanding is right, there is not 
so strict relationship between a value of the penalty in IEEE802.3ah spec 
and the spread sheet.
 >So, there is still remaining one question in my head.
 >
 >How can we calculate a penalty value for estimating a power budget?
 >I think there are two ways to estimate the power budget.
 >
 >1. We decide the power budget and CHIL at first, and then we calculate 
the allocation for penalty by subtracting the CHIL from the power budget.
 >2. We decide CHIL and the penalty at first, and then we calculate the 
power budget by adding the CHIL and the penalty.
 >
 >Do you have any opinions?
 >
 >Sincerely yours,
 >Tsutomu TATSUTA