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

Re: [802.3_B400G] waterfall curves

Hi all,

 

I am agreeing with Chris that it is of importance to see waterfall curves.

That was also the point of my presentation given to the dj ad hoc on 2 November:

https://www.ieee802.org/3/dj/public/adhoc/optics/1023_OPTX/stassar_3dj_optx_01a_231102.pdf

If the BER performance is OK, then we will be able to derive appropriate Receiver sensitivities and establish suitable power budgets and reaches.

 

Kind regards,

 

Peter

 

From: Chris Cole <chris@xxxxxxxxxxxxxxx>
Sent: Wednesday, November 15, 2023 10:34 PM
To: STDS-802-3-B400G@xxxxxxxxxxxxxxxxx
Subject: [802.3_B400G] waterfall curves

 

During welch_3dJ_01a_2311 Q&A, an assertion was made that the presentation proved that there is no error floor because it showed a single point BtB measurement with an impressive BER.

 

That is an incorrect conclusion. A single point BtB measurement is encouraging, and suggestive of overall good performance, but it is not proof. For that we need waterfall curves. 

 

It may help to understand that waterfall curves are not just an aesthetic format choice, but the foundation of communication. 

 

It starts with Shannon-Hartley theorem:  C = BW ln(1 +S/N)

 

We have generally accepted that when BW (bandwidth) changes we need a new set of measurements. No one would make a proposal for a new rate, for example 200G/L, and claim that 50G/L or 100G/L measurements demonstrate technical feasibility. 


The same holds for the S/N term. It is insufficient to show a single point measurement, unless one can guarantee that in operation that S/N is the only operating point that will be seen by the receiver. If we specify a range of S (power) that a receiver will see then we must measure over that full range, and ideally beyond to show margin. 

 

If we look at a single point BtB measurement with power level unspecified, it just means that at the unknown power level we get a good BER. It doesn't tell us where within the specified power range the point falls, which means that there could be a much higher error floor at a lower power level, or more importantly at a higher power level. This is what often happens. Overload problems arise as power is increased even when the error floor at mid power range is low.

 

The point has been made that all that matters in 802.3 is 75%. However, let's hope that technical feasibility is based on more than subjective opinion, and instead on objective measurements.

 

Thank you

 

Chris

 

 

To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1

To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1