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

Re: [10GMMF] Minutes from TP2 call 2/17/05

All - due to a conflict, there will be no TP2 call this week (2/24). If there are any particular issues that need to be discussed, please use the reflector, or give me a call. Otherwise, please be working on comments for the ballot due 2/25.
 
Tom Lindsay
ClariPhy Communications
tom.lindsay@clariphy.com
phone: (425) 608-0209 or (949) 480-9210
cell: (206) 790-3240
fax: (425) 608-0232
 
----- Original Message -----
Sent: Thursday, February 17, 2005 11:00 AM
Subject: [10GMMF] Minutes from TP2 call 2/17/05

Attendance
  • Andre Van Schyndel
  • Vivek Telang
  • Piers Dawe
  • Tom Lindsay
  • Norm Swenson
  • Ali Ghiasi
  • Joe Gwinn
  • Greg Lecheminant
  • Lew Aronson
  • Jan PeetersWeem
Agenda
  • Attendance (see above)
  • Agenda (approved)
  • Review old minutes (2/10 approved)
  • TP2 topics (see below)
  • Next meeting/call - 2/24
TP2 technical topics
  • TWDP test pattern (from AnAi or BnBi)
    • No progress. Explained background of pattern presently in D1.1. This pattern is a candidate but needs validation.
  • RJ 
    • Can RJ magnitude, currently 0.033 UI rms be reduced?
      • Piers - Tx noise may be swamped by Rx noise.
      • Tom - may be less true for LRM since min OMA is 30 dBe above rms noise floor in budget.
      • Ali - if all Gaussian, would be 0.46 UI pk-pk.
      • Ali - for 10G serdes and clocks, <1 psec rms is common. Laser noise might raise this to 2 psec rms.
      • Tom - want to keep a simple test, reducing to ~2 psec rms or so is one approach.
      • Ali - otherwise, could spec/control unbounded RJ and bounded DJ. This would require a detailed proposal.
    • Scope folks to report on instrumentation RJ capabilities. Most comments are from Greg.
      • 2 grades of scopes: ~800 fsec rms and ~200 fsec rms for high grade.
      • CDRs more typically 300 fsec to 1 psec rms.
      • All these values are okay if LRM spec is 2 psec or more.
  • Clock recovery/tracking
    • Uncorrelated jitter test - CR should track per 4 MHz high pass spec. Most comments are from Greg.
      • Some new scope CR units track transition density to maintain 4 MHz. Old ones may not, and some may lose lock if transition densities are too low. That is, some CR units may have troubles with low frequency square waves, and patterns with more normal transition densities are preferred.
      • Jitter measurement times are ~1 sec for a 20 bit pattern, ~4 sec for a pattern around the length of AnAi, and ~5 sec for prbs15. Greg wants to measure these times a bit more, but other methods including histograms methods should scale with similar ratios. Test times for short patterns dominated by scope sampling rate, long pattern times dominated by pattern rep rate.
      • Does it matter which edge of a long pattern jitter is measured on? General belief is that differences should be minor.
      • Testing with Tx and Rx running with mixed data patterns will excite self-noise in a more natural way.
      • Conclusions are that uncorrelated jitter testing should be done on a mixed repetitive pattern and that any edge can be used.
    • TWDP test
      • No discussion.
  • OMA square wave pattern
    • Piers - cannot officially change OMA pattern range.
    • Piers - has heard that measuring OMA above and below the crossing regions is very workable.
    • Tom - OMA must represent low frequency portion of signal strength such that it can be used for loss budgeting - that is, it should track the DC loss.
    • The standard could suggest that LRM use a range towards the long end. However, no specific actions taken or expected.
  • MATLAB code review
    • Piers - trying to create waveforms and run through IEEE code to compare to Abbott's PIE-D values for Gen67. Piers should talk to John about results, since John may be trying to do the same thing.
    • Tom to send Ewen PIE-D results that he used to compare IEEE code against to Piers.
    • Vivek has gotten excellent agreement with ClariPhy code, both with linear waveforms vs. PIE-D, and with a nonlinear waveform and his own EDC simulator.
    • How many channels/fibers are required for TWDP test? - not discussed.
  • TWDP and TP3 limits currently ~5 dB.
    • More work encouraged to establish if the line is drawn correctly.
  • Rx overload (driven by Rx folks?)
    • Andre offered to explore this topic further. Otherwise, no actions or responses.
  • Other topics? Key comment items for D1.1?
    • General point that much of LRM still needs validation through real testing. Part of feasibility demonstration.
Next call
  • Date: Thurs, Feb 24, 05 (regular day/time)
  • Time: 9:00 AM Pacific
  • Duration: 1:30 max
  • Number: 401-694-1515
  • Access code: 421721#
    • Thanks,
    Tom Lindsay
    ClariPhy Communications
    tom.lindsay@clariphy.com
    phone: (425) 608-0209 or (949) 480-9210
    cell: (206) 790-3240
    fax: (425) 608-0232
     
    ----- Original Message -----
    Sent: Wednesday, February 16, 2005 5:14 PM
    Subject: [10GMMF] Next TP2 call 2/17/05

    Next call
  • Date: Thurs, Feb 17, 05 (regular day/time)
  • Time: 9:00 AM Pacific
  • Duration: 1:30 max
  • Number: 401-694-1515
  • Access code: 421721#