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

[802.3_ISAAC] Latency discussion in 802.3dg

All,

 

I would like to point you all to a very interesting presentation on latency by Günter Steindl in 802.3dg:

https://www.ieee802.org/3/dg/public/May_2024/new-steindl-PHY-delay-requirements-0924-v01.pdf

 

The presentation was made in the context of industrial automation, but many of the same concerns apply to automotive.

 

On slide 4 of the presentation the Rx and Tx latency is limited to <1us in general, but <5us for 10BaseT1L.

 

Slide 5 makes the following important points:

  • Introducing a new PHY for copper should not require changes in the machine and automation cell
  • Roundtrip delays for motion or other high-speed applications shall still be achievable without new machine designs

While these requirements are listed in the context of factory automation, I believe that they apply equally to what we are doing in 802.3dm. In the automotive camera applications, there are existing systems today that are based on legacy proprietary solutions, and 802.3dm should not introduce significant additional latency. Günter clarified that if additional PHY latency is introduced, then it may require significant redesign of the whole system.

 

In follow up questions, Günter clarified that the tight latency requirements can not be relaxed if an accurate timestamping is used. In particular, he clarified that even with accurate timestamping, it would NOT be feasible to relax the latency requirements to 100us.

 

In follow up discussion, Günter also pointed out that link up time is very important.

 

Ragnar

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