Summary

  • A September 8 contribution to IEEE P802.3ds expresses several transmit, receive and illustrative link-budget entries in terms of the same still-unsettled maximum TDECQ.
  • The purchasing implication is to keep limits, test methods and specification versions together. A proposed budget is neither an adopted standard nor spare margin in a delivered link.

The most useful character in a new optical-networking presentation is a letter, not a headline data rate. In Ramana Murty’s supporting material for IEEE P802.3ds, T stands for a transmitter-quality limit still under discussion. It appears repeatedly: in transmit characteristics, receiver-test conditions and the illustrative allocation of optical power. Several unfinished entries are not separate mysteries. They depend on a common choice.

The contribution is dated September 8 and was posted ahead of the task force’s scheduled electronic meeting. The editors’ report for that meeting groups the central limit comments with related items and still lists maximum TDECQ as unresolved. This article examines those public materials at 11:15 UTC on September 8, before the announced meeting time. It does not report a vote or claim a value has been selected.

The budget is an allocation, not a bonus

P802.3ds concerns 200 Gb/s per wavelength multimode-fibre physical interfaces. The approved objectives span Ethernet rates from 200 Gb/s to 1.6 Tb/s. That programme matters to data-centre buyers contemplating short optical connections, but its objectives are not evidence that particular modules have shipped or that existing cabling will support them.

TDECQ means transmitter and dispersion eye closure for PAM4. It is a signal-quality quantity within a defined measurement procedure, rather than a second way of quoting throughput. Murty’s new deck assumes the maximum for the related transmitter eye-closure measure, TECQ, equals the maximum TDECQ. Under that assumption, selecting T carries consequences beyond one row labelled “transmitter”.

The 1060 nm example makes the accounting accessible. Its illustrative power budget is T + 1.9 dB: 1.7 dB of channel insertion loss and T + 0.2 dB allocated to penalties. Substituting 4.4 dB for T would produce 6.3 dB overall. It would not give the buyer 6.3 dB of unused headroom. The table has already assigned that budget to the channel and impairments, with no additional insertion-loss allowance in this example.

That distinction is commercially important even before a purchase is made. A larger number under “budget” can look like a more forgiving link when copied into a comparison sheet. Here, raising the assumed quality limit also raises the associated penalty allocation. A reader cannot infer an improvement in usable margin merely by observing that the total went up. Nor can this conditional calculation establish the performance of an actual transmitter and receiver pair.

A disagreement with a technical address

The D1.2 proposed-response record contains different suggested limits. Their scope matters. Ali Ghiasi proposes 4.0 dB in comments concerning both the 850 nm and 1060 nm clauses; David Lewis proposes 4.4 dB for the 1060 nm clause; Eric Bernier proposes 4.6 dB TDECQ for the 850 nm transmit table. These are proposals, not three freely interchangeable settings for every interface.

Lewis’s earlier 1060 nm contribution already highlighted how a provisional 4.4 dB choice would resolve several dependent numbers. The new supporting deck makes that relationship symbolic instead of requiring the reader to infer it from coloured cells. Neither presentation is, by itself, a final disposition of the comments.

The measurement definition deserves equal attention. The unresolved-items contribution discusses a reference equalizer with 15 feed-forward taps and one feedback tap, alongside the calculation of its DC gain. A threshold divorced from the procedure that produces the measured result is an incomplete acceptance condition. A purchasing team does not have to adjudicate the equalizer debate to insist that a supplier identify the procedure used.

There is no scandal in a developing standard containing unresolved choices. Exposing the dependence is a useful part of the work. The avoidable error comes later, if a slide date becomes shorthand for approval, or if a sales specification combines a favourable limit from one proposal with a receiver assumption from another.

What the buyer can decide now

A laboratory evaluation can be valuable before the standard is complete. It can establish how the selected equipment behaves under explicitly provisional conditions. The result should travel with those conditions: wavelength, physical-interface variant, fibre assumptions, reference test method and specification revision. Removing them to simplify a management presentation discards the information needed to compare a later result fairly.

Lu Heng’s writing on representations and executable power provides a useful distinction here. A table represents a technical proposition; a purchase contract turns selected requirements into consequences for acceptance and payment. His discussion of a strict common minimum and local adoption also helps separate two decisions. A shared interface needs precise common rules. An operator still has to decide when the evidence is sufficient for its own deployment. This is the author’s application of those ideas, not Lu Heng’s assessment of the project.

Local discretion does not mean changing the test and claiming unchanged conformance. It means deciding what to evaluate, what uncertainty to retain and when to make a commitment that is harder to reverse. The small letter T is a reminder that a procurement checklist should preserve technical dependencies rather than make them disappear.

Sources