|Thread Links||Date Links|
|Thread Prev||Thread Next||Thread Index||Date Prev||Date Next||Date Index|
Larry – IEEE 802.3 standards generally provide link segment transmission parameters and noise parameters (such as coupling between link segments). They also provide PHY receiver specifications which specify the levels of noise, interference, and attenuation that a PHY needs to tolerate. It is unusual for 802.3 standards to supply physical reference specifications. However, particularly for backplane PHYs, there are informative reference configurations contained in IEEE 802.3 standards – see, for example, informative Annex 69B, or, a more evolved version – clause 92.10 (normative transmission characteristics) together with annex 92A (informative details). these do not substitute for link segment or PHY implementations.
Where possible, though, being able to reference other standards for the construction of the channels is the preferred approach. However, as above, there are cases where this can’t be done, and an informative annex may be helpful.
While I agree that the same, interoperable PHY description can and should be in the specification and I totally support the additional use case that this change supports, but I do have a question.
As a committee should we provide guidance to futures users of this specification a reference channel model for the various use cases that we support?
In other specifications I have been involved with this has been accomplished through normative annexes. I believe that some guidance should be required if for no other reason to provide future implementers at least a baseline reference as to how the specification limits were developed and to allow duplication and verification of this work.
I would think that this work does not need to be all inclusive (i.e. every potential configuration) but should at least provide a baseline functional system as a reference.
Delphi Automotive LLP
I had stated the electron’s don’t care whether the wires are twisted.
What I added to that, and maybe didn’t emphasize enough, was that they care about the transmission and interference parameters of the channel – not its physical construction. These parameters can be achieved by PCB traces, shielded pairs, coax, or other means. The two applications fit within the same set of transmission (and interference) parameters – then the physical construction is outside our scope (a problem left for the implementer).
Having done this for a number of decades, I firmly believe it is possible and even likely that the same PHY developed for the short-reach channel will serve the need for the backplane application. There is a long history of using ethernet PHYs both for cabled and backplane channels. BASE-X PHYs have been used on coax & board traces, 1000BASE-T has been used on backplanes, and even run on open-wire, DSL transceivers (Ethernet in the First Mile) have been similarly used both on traces, open wire, twisted pair, and even coax.
Additionally, we are not saying that the same chip is used both in backplane and, for example, automotive applications – only that it is the same, interoperable PHY description. There are other requirements on the chips, for example, the thermal environment, which may make them different chips. However, if you plug two PHYs of the same specification into a common link segment, they should work interoperably (at least at the PHY level).
George Zimmerman, Ph.D.
President & Principal
CME Consulting, Inc.
Experts in Advanced PHYsical Communications