[10GMMF] Notes for TP2 call, 8/26/04

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

All - here are my notes for the 8/26 TP2 call. Please send comments or corrections if needed.

Attendees (no order)

Jesper Hanberg, Intel
Jan Peeters Weem, Intel
John Jaeger, Big Bear
Lew Aronson, Finisar
Yi Sun, OFS
Yu Sun, Optium
John Dallesasse, Emcore
Bala Mayampurath, Vitesse
Piers Dawe, Agilent
Tom Lindsay, ClariPhy
Others?

Summary objective (repeated from before)

Present a proposal for TP2 signaling parameters and associated conformance testing at the September Meeting. The work must consider and provide tradeoff information among component cost, test cost, and power penalties.

Reduction in power may be another outcome to consider.

Housekeeping

Proposed agenda was approved (see below).
Previous notes (from 8/19 call) were approved.

Progress, technical discussions

There were no new presentations this week.

The only significant discussions involved follow-on to the Intel presentation from the week before, 8/19. Jesper sent an email 8/26 that addressed some of the questions posed from the 8/19 call.

Laser modeling

A slide was included in the 8/26 email that depicted a laser response (perfect rectangular current waveform) with more peaking than used in the modeling in the 8/19 presentation. Higher peaking was created by reducing the gain compression coefficient and by reducing the operating (now 50 mA) and modulation currents (now 20-80 mA). The latter changes should also reduce ROF. The slide shows some important laser values.
Some comments were that ROF is high (shown at 14.6 GHz, try cases closer to 10 GHz), that some lasers have less damping (more ringing), and that the filtered mask limits could be used to set limits.

The drive waveform in the 8/19 presentation used filtering but no reflections (or multiple reflections).
If anyone has suggestions on laser settings or drive waveforms, please get in touch with Jesper.
A MMSE method was used to determine the EDC coefficients.
hanberg_1_0304 briefly mentions the method for determining BER. It appears to use the signal to rms noise ratios for 8 binary triplets at a slicer input.

I suspect there are more questions on the details of these 2 methods (speaking for myself, anyway...).

A single fiber was used in the 8/19 presentation. It is the one with the 22nd worse PIE-L response in the Cambridge model set (out of 65 fibers?), taken from the end with the greatest PIE-L value.
Jesper acknowledged that his work did not include the effects of RIN, but expected the additional impairment would be on the order of 1/2 dB.
Intel did some investigation of more taps for FFE (up to 9), but did not see much improvement compared to the 5 used in the 8/19 presentation. Jesper believes the limitation may be due to laser nonlinearities. DFE taps showed greatly improved performance over FFE.
There was a question on why the results of some of the DFE simulations were better than predicted by (ideal) PIE-D. Jesper thinks it may be a combination of rounding errors and that the simulations were done for a BER of 1E-9.

Next call (9/2, 9 AM Pacific time)

Cost vs. performance studies (Opnext, Bookham)?
Other presentations encouraged
- Cost trends
- Penalties and test metrics vs. parameters (and try to relate to cost trends)
- Other?
TP2 test metrics (test and methods)