Hi Claude and Frank,
Thank you for preparing
ACT and TDD Comparison.
After reviewing the slides, I and my co-authors believe several statements about ACT – and several places where ASA material is used as a proxy for the IEEE TDD proposal – are inaccurate or insufficiently supported. To keep our technical record clean, I’d
appreciate a review and correction of the items below (slide refs in brackets point to the evidence in my slide deck).
1) ASA-ML/MLE ≠ IEEE TDD (scope/attribution).
Your slides use ASA-ML/MLE results and language in support of the IEEE TDD proposal. ASA-ML and ASA-MLE are both TDD-style duplexing method but differ materially from the IEEE TDD under consideration. They are not interchangeable and are not guaranteed interoperable.
Please label ASA content explicitly and avoid using it as evidence for IEEE TDD unless the underlying specification and test conditions are made available for review. [See overview and interoperability notes in my deck, Slides 3–4, 16, 23.]
2) Crystal-less operation, PoC & EMC.
ACT has demonstrated crystal-less operation with PoC and has passed EMC in like-for-like setups. Replacing a crystal with an external signal generator with 100ppm offset and then modulating with an unspecified additional ppm offset. This is not a correct demonstration
of crystal-less. The EMC slides do not present comparable TDD data (only ASA data). Please provide TDD evidence with test setups and limits, or remove the comparison. Also, comparing CISPR 25 “noise floors” across different receivers/setups is not valid. [Slides
3, 11–12.]
3) Latency/link-delay accounting (uplink, buffers, IBGs).
Some TDD latency calculations omit uplink timing and fixed buffering (e.g., ~933 ns) and do not include inter-burst gaps. Please revise to include DS+US, buffers, and IBGs for each speed grade, and distinguish payload vs total link latency. [Slides 9–10, 20.]
4) Bandwidth/ISI and low-frequency behavior.
TDD’s higher downstream bandwidth drives more ISI and is sensitive to longer propagation delays, raising collision and baseline-wander risks on long links. Please address these effects directly (including insertion-loss line choices and low-frequency impedance
behavior), rather than assuming parity with ACT. [Slides 4, 13, 19.]
5) Power, complexity, and die-size claims.
Claims that a multi-Gbps TDD PHY can be smaller/simpler or lower-power than a 100 Mbps architecture ignore equalization, linearity, and additional low-power state machinery. EEE is not required to hit low power with ACT, and adding state machines for aggressive
power modes increases area. Please update or remove these assertions unless supported by silicon-equivalent block-level accounting. [Slides 5–8, 12.]
6) “Reversibility” and other non-PAR features.
Reversibility and several other items cited are not part of the 802.3dm PAR/objectives. If you intend to propose PAR changes, please call them out explicitly and provide rationale. Otherwise, these should be removed from the comparison claims. [Slides 3, 15,
21–22.]
7) Cable-length market data.
Please expand beyond a single passenger-vehicle example (BMW X5/X6/X7) and address commercial/off-highway needs. Earlier industry inputs request up to ~40 m for certain use cases. [Slide 19.] The use case listed is for a specific OEM.
8) 25 Gbps note (why it appears).
Please check your math, as there is a miscalculation. I also mention 25 Gbps not to extend the scope but to check forward scalability—an interest you (Frank) raised previously—and to highlight where timing/echo/IBG assumptions may need to evolve at higher rates.
All such discussion is clearly marked as forward-looking. [Slide 20.]
For convenience, here is the deck that enumerates these corrections with citations and calculations:
“ACT vs TDD Comparison: Evidence-Based Results”.
Houck_etal_3dm_01_0925.pdf and my previous presentation to pull references from that I put together with my co-author(s)
Houck_Cordaro_3dm_01b_07292025.pdf
If you can share an updated deck or an errata list on the reflector, I’m happy to do a joint slide-by-slide review so we land on an accurate, apples-to-apples comparison before the next TF meeting.
Best regards,
TJ Houck
Infineon Technologies Americas Corp. – Detroit
System Architect
tj.houck@xxxxxxxxxxxx
To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1