Geoff –d thanks for the clarification, and
this is a fine point.
First, I think we agree that the MDI
referred to in standards is an interface plane, a reference
point, not the actual physical connector, although common
usage may use the term “MDI” to refer to what would be more
precisely the “MDI connector”.
Technically, the MDI (reference point) cuts
right through the middle of a physical connection to the
medium, at its mating surface, and is therefore hard to
measure. Always a sticking point, because, as you point out,
the cabling standards don’t include even the cabling side of
the connector (a plug in the case of RJ-45).
However, I disagree that the link segment
includes the MDI connector for the following reasons:
First, the electrical characteristics of
the MDI connector for most PHYs are defined separately.
(they’re usually in a xxx.8 whereas the link segment is in
xxx.7).
Second, the definition of the ‘link
segment’ in clause 1.4 is “the medium connection between” two
MDIs. To my understanding, the words “medium” and “between”
exclude the connectors which define what the “medium” is
“between”. when the connector is a jack/plug arrangement,
with the jack on the equipment, the jack side of the interface
plane is part of the equipment (or “port”) which we specify,
and the plug side is part of the link segment (whose
composition we do not).
Third, it is not possible to measure the
link segment characteristics, independent of the equipment,
and therefore impossible to specify them, if you include the
port side of any connector.
- again, this is a fine point. As far as
the discussion we were having in the ad hoc today (which
regarded the inliners vs. the cable segments themselves), I
think we agree. they are all parts of the link segment.
-george
George-
Actually, to be precisely correct,
The "Link Segment" is the MDI to MDI
piece and includes the connector.
Just as the MDI on the DTE includes the
connector.
MDI means exactly what it stands for, Media
Dependent INTERFACE as distinct
from the "MDI connector."
Interface means "at the
connecting surface." The distinction is very important for
copper connectors which generally attach pins to the cable
conductors. Not as important, but not of zero importance in
fiber connectors where the connector in usually used to
position the cable fiber directly against the mating
surface.
Our system separation is: Equipment -
Terminated Medium - Equipment
Cabling standards are different, they put
the edge of their standards and performance specs at the
back side of the connector which is application specific to
the DTE.
This difference is not huge but it is
real. Again, it tends to be more significant with copper
than it is with fiber.
Thus your statement about the definition
of a "link segment" below is
incorrect.
Chris – I think the below is a
great start, and yes, we need to be rigorous about
saying “link segment”.
I haven’t heard Geoff (Thompson)
on our calls in a while, but I’m hearing his voice
in the back of my head, and in this case, his voice
is one of wisdom. Words have power and meaning – we
need to use them correctly. Good standards are
unambiguous to the reader, and do not rely on
assumptions. I know you know this, and the below,
it is for other’s benefit I expound.
To that extent, we talk about the
components used, such as cable and connectors, in
order to understand the technical feasibility of
what we are requiring. However, I suggest that what
we SPECIFY in the document is only at the link
segment level. I write the below with great
trepidation, because someone will have a nit. The
important thing is that the link segment is a
defined thing and we don’t go within it – just as
Chris said.
link segment: this would
include inline connectors, and cable (or other
conductor) segments, but not the MDI connectors.
When the MDI is at the edge of the PCB, it does not
include PCB losses as well. It is a defined term in
IEEE Std 802.3-2018 (see 1.4.309), so we can’t
change it, it is point-to-point and includes
everything between the two MDIs. 802.3 does not
generally sub-divide the link segment into
components in its requirements (on occasion it may
offer informative guidance on how components might
relate to construction of a conforming link
segment).
If we want to go further in
specifying requirements and include more than just
the link segment, “channel” (1.4.179) is often used
for these purposes (but see below for a more
rigorous approach - test points). “channel” has been
a bit of a hot-button in 802.3 specs, and, except
for systems which use a notion of a channel as a
frequency (like a radio or TV channel), is usually
locally defined within clauses (“a defined path
along which an electrical or optical signal
passes”). As a result, the precise meaning, depends
on the clause and the context, and therefore, in my
opinion, the word “channel”, while common, is better
avoided in 802.3 standards, and needs extra
precision. (because the reader THINKS they know it,
but it has some specific, precise meaning).
However, for us, and often elsewhere, channel is
useful to mean everything between the PMA/PMD
(transmitter) and PMA/PMD (receiver), including PCB
and MDI.
The rest of the words we use are
not strictly defined in IEEE Std 802.3, they
include “cabling” which is generally “cable and
connectors” and roughly equivalent to link segment,
except that it isn’t restricted to MDI to MDI
connections (“cabling” may be a subset of a link
segment as a result), “cable” which is just the
wiring of an individual segment (which may be a
subset of a link segment), and “connectors”,
“inliners”, or “connecting hardware” which are
mostly self-explanatory for connecting hardware
within the link segment (not the MDI connector).
Components of the link segment
are generally in the domain of other organizations.
In this case it might be SAE, might be OPEN, might
be IEC, … not really sure. This is where liaisons
help, and if some of our
automotive-industry-affiliated participants can help
clarify which groups write normative specifications
that we might reference, that would be great.
Following the ad hoc meeting discussion; I
suggest when we mean "link segment", we say
"link segment".
If
I understand correctly, methods for
restricting micro-reflections in
jonsson_3cy_01_12_08_20.pdf would yield
restricting reflections in "link segments"
characterized as residual return loss (TBD).
Towards baseline, residual return loss would
be specified as a link transmission parameter
(using 802.3ch terminology).
XXX.X
Link segment characteristics
XXX.X.X
Link transmission parameters
XXX.X.X.X
Residual Return loss
Brief
description on residual return loss...In
order to limit the noise at the receiver
due to residual return loss.....
To
unsubscribe from the STDS-802-3-B10GAUTO list, click
the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10GAUTO&A=1
To
unsubscribe from the STDS-802-3-B10GAUTO list, click
the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B10GAUTO&A=1
