RE: Interferometric noise, version 2
Petar,
I read your response with interest and I think there are a few things
that need to be said.
In our document we devote a section to describe the large effects an
accidental air gap has on the interferometric noise. You think that we
should exclude the possibility of an air-gap in the link because it is
outside of the standard. I share you opinion here, but some people do
not. I think we should get this nailed down, and then take care of the
implications is has. If we can get an agreement on this we should
revisit the 1550 spec where a transmitter is required to withstand 12 dB
return loss coming from an accidental air gap in the link. If we can
not, we should take this into account when we consider interferometric
noise.
For 1300 nm, you think the new version of our document represent a shift
in right direction, getting closer to the "real numbers". However, you
still don't recognize that our approach does not represent the worst
case. If you have a short link with two path cords with 26 dB connectors
(one path cord at each end of the link), the total reflection at each
end will vary between 10.5 and 14.3 dB. The effective return loss of
both the Tx and Rx will be below 11 dB simultaneously with 8%
probability. That gives you 5 dB penalty for a very short link. Now this
is when the polarization line up which they will not always do, but some
days you will get penalties of this order in a short link. Then add
overshoots that you have in directly modulated lasers. Another thing we
need to make sure is that the receivers that are available to us operate
perfectly OK with such kind of penalties. Personally I am not confident
that they do, but I have not scanned the market.
In essence, the point that I as well as other people have made is that
by taking this approach we enter a domain people generally avoid. I am
surprised at how easy dismiss this as a non-issue if we add 0.4-0.5 dB
power penalty in the link budget. I know that some of the effects we
have not yet taken into account work in opposite directions, but we
really don't know how they add up. It might be the case that we're OK,
but we might also end up on the wrong side of the line. Achieving a
return loss of 20 dB (or somewhat less) is not
very difficult and would fix the problem.
/Peter
> -----Original Message-----
> From: Petar Pepeljugoski [mailto:petarp@us.ibm.com]
> Sent: den 29 januari 2001 21:45
> To: Peter Öhlén
> Cc: stds-802-3-hssg-serialpmd@ieee.org
> Subject: Re: Interferometric noise, version 2
>
>
>
> Peter,
>
> I read your write-up with interest and I must say that the
> approach that
> you take represents a shift in the right direction, getting
> closer to the
> real numbers. I find your results to confirm my assertions
> (that I voiced
> during the meeting in Irvine and later during our serial PMD
> teleconference) that the IN penalty for the longest links is
> between 0.4
> and 0.5 dB. I believe that your results are also implying
> that the penalty
> for the shortest links is such that it does not cause
> negative margin (fig.
> 4, 5 and 6). They also confirm the numbers that are in the
> spreadheet that
> I posted on the reflector (interferometric noise 3a.xls), which can be
> found on the serial pmd site.
>
> Having said that, I also have to express my surprise with
> your conclusions.
> You continue to make the same conclusions, and propose the
> same remedy as
> before, although your simulations now show reduced IN
> penalty, consistent
> with my numbers. Your simulations confirm that the steps
> taken so far (3dB
> ER limit for the transmitter, with 12 dB RL for both the TX
> and the RX) are
> sufficient to guarantee the performance of the link with
> positive margin.
> However, you do not comment on those things.
>
> The only thing that we need to do is to add the 0.4-0.5 dB IN
> penalty to
> the link budget (we already have 0.61 dB of margin, so people
> know that it
> is not unallocated margin and can't be used for other things.
> As for the
> argument that we need to accomodate air gaps, I think we should not
> consider things that are outside of the standard, since that road is
> slippery and can lead us to considering other conditions that
> are not part
> of the standard.
>
>
> Regards,
>
> Peter
>
> Petar Pepeljugoski
> IBM Research
> P.O.Box 218
> Yorktown Heights, NY 10598
>
> e-mail: petarp@us.ibm.com
> phone: (914)-945-3761
> fax: (914)-945-4134
>
>
> Peter Öhlén <Peter.Oehlen@optillion.com>@ieee.org on
> 01/29/2001 02:24:47 PM
>
> Sent by: owner-stds-802-3-hssg-serialpmd@ieee.org
>
>
> To: "_Serial PMD Ad Hoc Reflector (E-mail)"
> <stds-802-3-hssg-serialpmd@ieee.org>
> cc:
> Subject: Interferometric noise, version 2
>
>
>
>
> There have been some concerns about the validity of the worst-case
> approach that was
> taken in our previous document about interferometric noise. In this
> document (see link below) we will extend the previous document in two
> different ways. In both approaches worst-case polarization is assumed.
> In real life the polarization state will vary slowly with
> time. However,
> the polarization state will persist for a fairly long time and could
> give random and unpredictable outages when the polarizations are
> aligned.
>
> In the first approach, instead of using the worst-case phase
> between the
> main signal and the reflected signal, the BER is averaged
> over the phase
> difference between the interfering signals. Chirp and other dynamic
> effects are ignored in this approach. In the second approach a
> rate-equation model for a single-mode laser is used, and the resulting
> output signal is used to calculate the effect of the interferometric
> noise. Here dynamic effects such as chirp and overshoots are
> taken into
> account. Although the semi-analytical approach makes a number of
> simplifications it compares fairly well to the more detailed simulated
> approach.
>
> The document was filtered by the exploder, please use this URL:
> http://hem.passagen.se/peter.ohlen/interferometric2.pdf
>
>
> ================================================================
> OPTILLION --- High speed optical transceivers
> Peter Öhlen Phone: +46 8 477 41 56
> Kronborgsgränd 9 MPh: +46 70 181 52 05
> S-164 87 Kista Fax: +46 8 477 41 51
> http://www.optillion.com mailto:peter.ohlen@optillion.com
>
>
>
>