RE: 1000BASE-T PCS question

["Chang, Edward S" <Edward.Chang@xxxxxxxxxx>]

Dave:

I agree that the intention of SONET scrambler is to achieve sufficient DC
balance; nevertheless, from the scrambling probability, a run length of 70
bits is inevitable which causes the problem. 

ATM_ Forum could not tolerate the run length, and further added additional
scrambling circuit to reduce the run length, which is still affecting DC
wander.  It is not practical to keep adding another layer of scrambling
circuit to eliminate run-length affect.

The optical power is only as good as to the O/E conversion, to correctly
convert optical signals to electrical signals.  The DC-wander problem is the
electrical, bit-boundary detection issue, then in-turn it relates to the
recovered clocking, accuracy issue.  All these issues are electrical, timing
issues, which are independent of optical power.  In another word, even we
have a zero error at O/E conversion, these timing issues exists after that. 


In addition, there always a link runs much better BER than specified, for
example, 10^-15, but it is not typical.  If 10^-15 is typical for 10^-12
specification, then it is not optimum cost-effective design.   


Ed Chang. 
Unisys Corporation             

   



-----Original Message-----
From: David Martin [mailto:dwmartin@xxxxxxxxxxxxxxxxxx]
Sent: Thursday, May 27, 1999 5:50 PM
To: Chang, Edward S
Cc: HSSG_reflector
Subject: RE: 1000BASE-T PCS question




I guess my point wasn't clear. It isn't the RX circuitry after the O/E
conversion that is the limiting
factor. The SONET scrambler provides sufficient DC balance that the data
slicing threshold (related
to base-line wander) and sampling point (related to clock drift) are not
difficult to optimize. The
limiting factor is enough RX optical power to provide a sufficiently open
eye. 

It looks like there will be lots of good discussion next week.

...Dave

	-----Original Message-----
	From:	Chang, Edward S [SMTP:Edward.Chang@xxxxxxxxxx]
	Sent:	Thursday, May 27, 1999 4:07 PM
	To:	Martin, David [SKY:1I63-M:EXCH]; Chang, Edward S
	Cc:	HSSG_reflector
	Subject:	RE: 1000BASE-T PCS question


	Dave:

	Thanks for inputs.

	I am not surprised that some link will keep running for equivalent
of seven
	years to get one error.  

	I agree if you allocate more photons/bit, higher S/N ratio, BER will
be
	improved.  You mean OC-192 specification sets the receiver minimum
input
	power at the BER of 10^-12.  Sure it can be done.  We can even
assume that
	no errors up to O/E conversion just for the sake of discussion.
However,
	after O/E conversion, the base-line wander and PLL clock drift
caused by
	excessive run-length will generate bit errors.  Therefore, just more
	photons/bit will not improve over all BER, you may have something
more. 

	Ed Chang
	Unisys Corporation
	Edward.Chang@xxxxxxxxxxx         
	 

	-----Original Message-----
	From: David Martin [mailto:dwmartin@xxxxxxxxxxxxxxxxxx]
	Sent: Thursday, May 27, 1999 1:09 PM
	To: Chang, Edward S
	Cc: HSSG_reflector
	Subject: RE: 1000BASE-T PCS question



	I'd like to clarify the residual BER spec for telecom systems. For
OC-48
	rate SONET systems, the
	end-of-life BER customer requirement is 1E-12 (no FEC). For OC-192
rate the
	value is the same, 
	1E-12 without FEC (c.f. BCR GR-1377-CORE Section 4.2.1, page 4-3)
and 1E-15
	with FEC (customer
	requirement).

	Note that some of our OC-48 customers, who have had systems running
for up
	to 10 years, are
	measuring RBERs in the 1E-16 range (about one error every 7 years on
a DS-3
	trib). The OC-192
	systems out there will need to run awhile longer to acquire similar
	performance data.

	The point here is that the SONET scrambler is not the limiting issue
in
	achieving low error rates. The
	issue is having enough photons/bit, or optical SNR (eye-Q) to
accurately
	recover the data. 

	...Dave

	David W. Martin
	Nortel Networks
	+1 613 765-2901   
	+1 613 763-2388 (fax)
	dwmartin@xxxxxxxxxxxxxxxxxx
	========================

		-----Original Message-----
		From:	Chang, Edward S [SMTP:Edward.Chang@xxxxxxxxxx]
		Sent:	Thursday, May 27, 1999 9:16 AM
		To:	Jaime Kardontchik; rtaborek@xxxxxxxxxxxxxxxx
		Cc:	HSSG_reflector
		Subject:	RE: 1000BASE-T PCS question


		Jami:

		We have been discussing scramble code versus block code,
8B/10B in
		particular, for a while on the reflector. Many people have
the same
	feeling
		that scrambled code has run length much longer than
desirable to
	cause
		base-line wander, and PLL clock drift; as a result, it can
not meet
	the BER
		of 10^-12

		The SONET using scramble code has BER of 10^-10, which is
not
	recommended
		for the datacom file transfer.  I believe the BER of
1000BASE-T is
	10^-10,
		again, which is not recommended for file transfer.

		You can prove the BER is 10^-12 for the 4D symbol code to
enable it
	to be
		used for all purposes, or stay at 10^-10 BER to be used, as
802.3ab,
	for
		less critical data handling. 

		Please clarify.

		Ed Chang
		Unisys Corporation
		Edward.Chang@xxxxxxxxxx