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

Millenimum MAS mailing


This might be my last e-mail of 1999.  More notes
to your notes; see below.

>For a multi-level system, The AGC would have to be linear and capable of
>maintaining equal effective BER's for all levels. In general, this requires a
>far more sophisticated AGC than is commonly employed in binary signaled
>A carefully designed multi-level system would establish and maintain an
>dynamic range at the receiver by accurately controlling the current
applied to
>the laser for each level. 

Whatever the transmit power limits are, there most 
assuredly will be an AGC at the receiver.  Perhaps
there will be some relief in the AGC design if it
operates over a narrower dynamic range - I can't think
of any right now.

>Page 15 of the same tutorial discusses alternate modulation techniques
such as
>Optical QAM. My Kauai technology update further discusses alternatives past
>PAM5. See page 24 of 

I am looking it up now.  One theme which comes to mind is 
linear FM.  It is a fact the most linear characteristic of 
common lasers is their time domain, or phase performance.  I 
did not want to have any missiles launched at me before Y2K, but
QPSK or MPSK are more attractive alternatives to me than MAS 
in terms of their technical risks.
>You are grossly misinterpreting my intentions. My PAM5x4 proposal is a
>proposal. Most aspects of the proposal would have to change to support PAM10.
>I've submitted close to a hundred proposals related to gigabit link
>architectures to the IEEE 802.3 or NCITS T11 committees in the past decade
>am ALWAYS open to constructive criticism. P802.3ae is free to accept all,
>or none of the proposal and amend and evolve it at will.
>I simply said that I would not support 10 level PAM knowing what I know about
>PAM, laser noise, SNR, alternate modulation technologies, etc. If you
would like
>to propose PAM10, please feel free to use my proposal as a template. In the
>absence of a counter proposal from you or a third party on PAM10, the
issue will
>go nowhere. Based on your previous notes, you are 100% against MAS. 

I am not 100% against MAS.  I can't afford to be.
What if it works and the committee adopts it? I would
lose too much face if I was 100% against the outcome.

My position on MAS is rather one of skeptical optimism.
I am aware of the many benefits of MAS should it pan out
into fine gold dust.  I have some doubts based on my previous
experiences with analog CATV and CCTV linear optical links.
I worked for one company which went broke in this business
in 1993, and have several colleagues who have echoed my 

Despite this sunburn, I am ready to go out and get a tan
as soon as the IEEE membership speaks. The analysis you 
have presented looks very good on paper, but my experience
teaches me manufacturing numbers of quality analog laser links 
is a lot like herding cats.  Consider this anecdotal evidence
in contrast to your well prepared study.

>Given this
>position, you won't run with the idea. That's why I asked you to drop it. I
>apologize if I ruffled your feathers with this remark. I'm just trying to
>forward progress with MAS.  

You are probably right.  I would most likely not develop a PAM9
or PAMN proposal.  Since there is no other PAM proposal on the 
table, it might be a better use of time to debate your PAM5.

I accept.  My feathers are now in flying condition.  I will try
to lay off the MAS puns for a while.

>One more time with gusto :-)
>2.5 GHz is the fastest effective symbol rate and corresponds to 5 GBaud.
>Look at is this way: The 1000BASE-X signaling rate is 1.25 Gbaud. The
coding is
>8B/10B. The highest effective frequency is generated by the 101010... pattern
>and is equivalent to 1/2 the baud. 

True enough if you filter out all but the fundamental.
But then you would wind up with a sine wave which is 
not a very useful as information.  Don't forget the 1st,
3rd, and 5th harmonics of the square wave at 1/2 the 
baud (I won't say rate).  It is necessary to include
some portion of the 3rd in order to properly condition
the signal for transmission over a dispersive media.

It also helps to include higher frequencies than the 
1/2 baud peak in order to maximize SNR assuming the 
PSD of an NRZ signal.  I have no problem with your 
statement other than it emphasizes a mathematical truth
which obscures practical considerations of bandwidth.

>Therefore, the maximum effective 1000BASE-X
>signaling rate is 625 MHz. The same general rule applies to PAM5x4 making the
>highest effective frequency 2.5 GHz. This has NOTHING to do with bandwidth of
>the laser, driver, receiver, TIA, rise time, etc.
>5.5 GHz is the proposed laser bandwidth using the formula ~1.1x Baud. Laser,
>driver and rise/fall times are all considered in this number. Compare this to
>the corresponding required laser bandwidth for a 12.5 Gbaud serial system.
>answer is ~13.75 GHz to transport 10 Gbps of Ethernet data.
>Where again is the confusion and inconsistency?

It is a bit scary to think about 13GHz.  But it is within our
capability.  The nice thing about those SONET folks is they
always seem to be a step ahead of us in bandwidth.  

I have stated my piece above about the differences between
baud and GHz.  I do not suppose it will do to repeat it again.
>9.22 is the correct number. I did say "approximately 2.3". 1.22 bits per
>symbol is the overhead allocated to FEC, special symbols, transition density
>guarantees and "thrown away" as invalid codes. See page 10 of
>> I have seen numerous postings which indicate 8B/10B encoding will
>> be supported.  I understand from reading your proposal and by
>> assuming you have calculated the bit rate correctly, the 8B/10B
>> might be used at the MAC, XMII, etc. but would be stripped away
>> in the transceiver and replaced with Viterbi or Reed-Solomon code.
>> If this is not the case, there would be a penalty imposed due to
>> supporting the 8B/10B encoding in addition to Viterbi or Reed-
>> Solomon which would reduce the information bandwidth another 20%.
>My PAM5x4 proposal supports Hari and strips away the 8B/10B code. 
>> ----------------------------------------------------------
>> >> I do not agree that MAS and laser safety are independent for
>> >> the case of the 850nm lasers I cited in my previous posting
>> >> which I have left in below.
>> >
>> >I believe that you'll need to make a better case for a dependency
>> >because I blew your previous case out of the water above.
>> Referring once again to your paper, on page 37 you acknowledge
>> the advantage 1310nm lasers have in a higher Class 1 safety
>> limit.  This limit is ~6mW for 1310nm while the limit for 850nm
>> is ~ 400uW.  If this advantage you claim for the 1310nm lasers
>> higher Class 1 limit is not relevant to the MAS proposal for
>> other wavelengths such as 850nm, it probably should not be listed
>> as an advantage.  It appears your opinion has changed.
>> I agree with your original conclusion stated in your white paper;
>> the higher Class 1 safety limit creates an additional degree of
>> freedom for 1310nm transmitter design.  One of the design choices
>> may be to overcome the SNR penalty due to MAS with additional
>> transmit power rather than to rely solely upon the benefits of
>> FEC.  Of course, any increases in transmit power must be rational
>> with respect to the overall system design.
>I agree with all you say. But what does it have to do with your original
>of a dependency between MAS and laser safety? Once again, MAS is a signaling
>method and is independent of laser wavelength and transmit power. 

There is no such freedom to increase power with 850nm 
lasers at this time.

MAS itself is not the laser safety issue anticipated at 
850nm, it is the requirements of MAS on the 850nm power
levels which may cause the difficulties with Class 1 limits.

I say "may" because I have not found sufficient details on
the signal processing gain due to the FEC to allay my
suspicious mind's suspicions.  Perhaps there is a reference
I have overlooked; if so please elucidate. 

>> 1) Inclusion of HARI increases the jitter budget by your
>>    logic above because it increases the number of elements
>>    over the conventional serial approach.
>You are incorrect. Hari has its own jitter budget which is separate from the
>PMD-medium-PMD jitter budget. A link architecture employing Hari is
>significantly different than a conventional link architecture such as GbE
or 1/2
>Gbps Fibre channel.

I suspect we are discussing a semantic, rather than substantive
difference here.  Jitter should not be substantially altered
in either case, unless a number of combinational stages are
required to decode the HARI and encode the MAS driver should
it in fact exist as an interface to the PMD/transceiver.

What really counts is the total jitter at the transmitter port.

>> 2) Part of reducing the losses in high speed serial transmission
>>    involves reduction of radiative losses by providing continuous
>>    ground planes around transmission lines.  This also produces
>>    a beneficial effect in EMI reductions.  Improving control of
>>    transmission line impedance reduces reflections and has a
>>    similar  EMI benefit due to the reduction in peak voltages.
>>    Keep in mind HARI reduces the transmission bandwidth by four,
>>    but it also multiplies the number of transmission lines by a
>>    factor of four.  So while the EMI peaks have been reduced in
>>    frequency, they have the potential to be greatly increased in
>>    amplitude.
>Yes. But a 10-12.5 Gbps trace won't work on FR-4, CMOS, and for trace

FR-4 is not the only choice.  I do not know if it is a 
clear directive we should not entertain miniature coax
for selected high speed lines.  Like it or not, we have 
a microwave transmission problem on our hands even with
HARI parallelism.  Coax certainly does wonders for EMI

>required in 10 GbE products. We've been through this in previous notes.
Also the
>EMI on a 16 line + clock 622 MHz parallel interface would be greatly
increase in
>amplitude compared to Hari by your logic.

Design trade-offs depend on one's experiences and 
expectations; I do not consider EMI concerns to be
of major significance in terms of our choice of which 
architecture to employ as a parallel interface.

>> 3) A 10Gbit serializer/deserializer could be implemented any number
>>    of ways.  The choices of technology are well established in this
>>    thread, but they include CMOS/SiGe, GaAs, SiGe, etc.
>Yes, but not economically. There's that darn Economic Feasibility PAR Critter
>I'm trying to address. My personal belief is this this is the most
important but
>elusive critter.

We have received commercial SiGe post amplifiers for
2.5Gbit/s receivers which are in the range of a Whopper
with cheese.  We can't use dollar signs here, can we?

>> 4) The traditional serial approach has the advantage here over
>>    a HARI approach because of the removal of the MAS encode/decode
>>    logic at the serial output.  This conclusion is consistent with
>>    your logic in (1) above.
See response to (1).  I will amplify if you like
in the next millenium.

 Best Wishes to all for the next Thousand Years,
 Patrick Gilliland