RE: Signal vs. Idle debate (A picture is worth a thousand words)
Dan's picture sums it all up. Let's be clear about what the objectives are for
this standard. Mid-span isertion is not an option... It is a requirement. As
per the March Plenary minutes :
Move to add a requirement that:
The solution for DTE powering shall support mid-span insertion of the power
Moved: A. Anderson Second: D. Dove
Y: 33 N: 0 A: 1 Time: 2:45PM
The benefits of this requirement is obvious:
1. It allows a low-cost retrofit to existing hardware in the wiring closet.
2. It does not require the replacement of existing switch hardware.
The concept of adding mid-span isertion via the un-used pair is straight-forward
and does not include any hardware that is not readily available, no non-standard
transformer is required and, as Dan points out, it does not affect return loss.
Phantom power via the signal pair requires the addition of an additional
transformer in the signal path; the effect on the return loss is unknown and is
may be detrimental. Geoff Thompson stated recently:
"There is currently very little margin in a Class D link for new elements.
Every new element that is introduced into a link has losses and mismatches.
In the cabling community the competition for the exisiting margins is
already oversubscribed by legitimate factions with very real needs."
In the last couple of days there have been alot of voices justifying using the
signal pairs because it allows inclusion of 1000BASE-T. It seems to me that
using the unused pairs, rather than the signal pair, will allow the return loss
specs of the 100BASE-TX standard to be met with no additional high quality
transformer elements. If we as a task force can show that a transformer exists
that will allow the return loss specs to be met, as the Phantom group assures us
it can, then this solution solves the Gigabit need as well. Two birds, one
stone, with the risk assigned where it belongs. We need to remeber that Gigabit
inclusion is not required of this task force.