| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
As happened previously, my initial thread
regarding phrasing was diverted to other issues. Here I will reset the
discussion back to the form of the reach objectives. While few have
chimed in publically on form, I have received helpful correspondence privately
and now have a better understanding of interpretations of the words “up
to at least” and “consistent with”. The phrase “up to at least” can
be interpreted as conveying a range (from near 0 m) to X m. Such
interpretations would preclude solutions that work at X m but do not work at
arbitrarily shorter distances, for example, such solutions lacking the required
equalization variation or overload tolerance. However, as is clear from
my objections, I do not find this phrase to unambiguously convey that
interpretation. If what is intended within the objective is to ensure a
range of reaches, then that range can and should be explicitly stated in the
objective. The phrase “consistent with” in
the circuit board objective is apparently meant to allow definition based on
channel loss characteristics rather than absolute length. This is a way
of separating performance from a specific implementation, so I now support its
use. However, the present word ordering seems to still introduce prescriptive
implementation. Moving the word “consistent” forward
alleviates that problem. To address these issues I would propose
modifying the reach objectives as follows. Define a 4-lane 100 Gb/s
PHY for operation over links consistent with copper traces on “improved
FR-4” covering a range of lengths from “W” m to at least
“X” m. Define a 4-lane 100 Gb/s PHY for operation
over copper twin-axial cables covering a range of lengths from “Y”
m to at least “Z” m. Given that the circuit board solution is
meant to support backplane connectivity, the interface points that define the
ends of the link need to be clearly stated. Otherwise that ambiguity can
easily lead to misunderstandings about how the channel impairments are
distributed between the backplane and the cards plugged into the backplane. The
outcome of such misunderstandings is lack of interoperability, which defeats
the primary purpose for setting a standard. This problem does not exist
for the twin-axial cable solution because the interface is known to be at the
point of interconnection to the cables. If the circuit board reach is
also similarly understood, for example, to be at the point of connection to the
backplane, then it should be straight forward to make that clear. I’d
like to get some discussion on this topic. Regards, Paul Kolesar From: John, Thanks for posting your planned
co-authored contribution on objectives. As you said, this contribution proposes
the form of the objectives and leaves the key parameters open. I have
previously commented to this forum about a particular form that, while
apparently approved in the past, is illogical, conflicting and ambiguous
because it mixes words that define minima with words that define maxima.
Specifically, slide 4 states the reach requirements using the phrase “up
to at least X m”. One can take this to define a maximum of X
m (i.e. ”up to”) or a minimum of X m (i.e. “at least”).
Given this conflict the only logical interpretation that can
simultaneously meet both is a reach of exactly X m. I do not find
this acceptable and have previously proposed to state these objectives clearly
as minima using the non-conflicting phrase “of at least X m”.
This will avoid some problems that have occurred in the past with
interpretation when the value of X was challenged. I also do not understand the utility of
the phrase “consistent with”. It appears to be a way to
interject some flexibility. But I do not have an appreciation for what
that implies. Perhaps someone could enlighten me. Without that insight I
would further simplify the form to the following. Define
a 4-lane 100 Gb/s PHY for operation over copper traces on “improved FR-4” for
link lengths of at least “X” m. Define
a 4-lane 100 Gb/s PHY for operation over copper twin-axial cables for link lengths of at least “Y” m. Regards, Paul Kolesar From: John D'Ambrosia
[mailto:jdambrosia@xxxxxxxxxxxxxxx] All, This email is to inform you of two updates to the Study
Group Web Pages. First, Mark Bugg has provided a calibration file, which will
allow de-embedding of the board traces. This file may be found at http://www.ieee802.org/3/100GCU/public/ChannelData/Molex_11_0210/2XCAL.zip.
My thanks to Mark for providing this file. Next, as discussed at our prior meeting, the SG needs to
focus on setting its objectives. To that end I have worked with Howard
Frazier and Regards, John |