|Thread Links||Date Links|
|Thread Prev||Thread Next||Thread Index||Date Prev||Date Next||Date Index|
I agree that efficiency and PD cost is important factors. The question is if the cost of not having plug&play is balanced or not.
System wise; what cost more?
I think there are good arguments on both sides, I’d like to give some information that may be useful to know about.
This is about efficiency, which depends on selecting a typical low cost schottky diode bridge versus selecting an expensive active (FET) bridge.
In the case of 51W PD, a low cost bridge (schottky rectifiers) will introduce close to 0.9 W x 2, which is a real concern for PD equipment makers, who struggle with thermal and efficiency.
If now we look at the whole PSE-PD-cable ecosystem, this has high impact on system efficiency, which partly defeats the purpose of driving 4P.
It is true that using an active bridge could eliminate a high portion of that loss, typically probably 75% of it, but it then becomes a quite expensive solution.
On the other hand, a single rectifier solution can be based on any of these two options:
1. single schottky diode: can already remove 50% of that loss with negligible cost,
2. one-FET solution: eliminate about 90% of the loss with just one FET (much simpler and lower cost) instead a full FET bridge (4 FET).
And we all know how cost-sensitive the PD equipment market is.
I fully understand the concerns mentioned below, and I’m sure they are valid concerns, but I think that efficiency and cost should also be part of the discussion.
Lest I be misunderstood, I am not supporting any position with regard to including bridge diodes.
My view is that wiring errors DO exist and that crossover, while rare any more, also does exist.
My position is that either diode bridges should exist in the PD, or a means of detecting wiring defects, crossovers, etc.. must exist and perhaps auto-correction of power application to address it. An "Auto-MDIX-like" application of power so to speak.
When these alternatives are weighed, diode-bridging may be the simplest, lowest cost, most robust means to address the problems cited above.
On a side note, plug-n-play being cited, we should note that prior to Auto-MDIX, it was not unusual for someone to burst into the lab frantically looking for a crossover cable. Ethernet was not as plug-n-play as we like to think.
Dove Networking Solutions
530-906-3683 - Mobile
On 2/12/14 10:44 AM, Rimboim, Pavlick wrote:
Just to add my 0.02$,
When POE LED lighting becomes a commodity anyone could install it on his desk lighting by using /running RJ45 cable ,
you cannot trust everyone to be a professional installer , so any robustness issue need to be looked at with this understanding that we are looking at a commodity solution at the end,
and the end user might be the one to install it and not only a professional installer.
Hugh, and John – Yes, and Yes. Crossovers are a part of life, and are valid. Wiring errors are also part of life, even if they are invalid according to standards.
Immunity to pair swap is generally listed in the objectives of modern BASE-T PHYs. As a result, data transmission in Ethernet systems is immune to, if not blind, to these defects, whether they are legal or not.
If we are going to consider removing the robustness against pair swaps from powering, what we will need is some kind of data-gathering exercise that spans multiple installations and installation classes. It’s a big job, and has to include not just the professional installation and cabling community but also the end-user IT manager community – because these things change over time, and we will be running on cabling installed years ago, with additions and changes made over time.
Principal, CME Consulting
Experts in Advanced PHYsical Communications Technology
(PLEASE NOTE NEW EMAIL ADDRESS. THE OTHER WILL STILL WORK, BUT PLEASE USE THIS FOR CME BUSINESS)
We need to be clear – crossover cables are not, in any way, faulty. Crossover and straight-through are part of standard connectivity.
Has anyone spent time with cable installers? Bad installations or crossed cables will always be blamed on the vendor
Sent via the Samsung Galaxy Note® II, an AT&T 4G LTE smartphone
-------- Original message --------
From: Dan Dove <dan.dove@xxxxxxxxxxxxxxxxxx>
Date: 02/12/2014 9:46 AM (GMT-08:00)
Subject: Re: [802.3_4PPOE] Are diode bridges really needed (2).pptx
Do you have any idea what the percentage of installations with inherent wiring defects may be? If its 20%, your point makes sense. If its .01%, it would seem that those customers may be willing to fix their wiring rather than pay additional cost and efficiency (OPEX) and have the entire industry taking a hit on OPEX.
Such wiring defects should be identifiable with a cable tester. Most professional installers certify wiring upon installation, and so this seems like a really small exposure. But if you have data otherwise, it would be useful.
Dove Networking Solutions
530-906-3683 - Mobile
On 2/12/14 9:15 AM, Darshan, Yair wrote:Dave and all,The "ALT C" looks a nice idea however I don't s how it would solve the concern that Hugh raised regarding PD working in some locations and in some it will not.Please consider the following situation:You have a multiport Switch that through that patch panel, goes to several locations in the building. The cabling installation is already there. So some PDs that have the diode bridge will work and some PDs that doesn't will not. This is problem #1.Now your suggestion may be good for new cabling installations. What will be the "interoperability index" now with mixed installation. This is problem #2.In cases that you have large switches with spare ports, the IT manager probably will use the same switch, will not install new cables (so no guarantee now of straight cables with fixed polarity)so we back in square 1.I believe that PD vendors that doesn't care about polarity insensitive (just care to prevent damage) may decide to do so regardless of the standard and we may not need to preclude it just leave it "out of scope" etc. If you add solution options by the spec, it increases the concerns for interoperability.Yair-----Original Message-----From: Dave Dwelley [mailto:ddwelley1@xxxxxxxxx]Sent: Wednesday, February 12, 2014 6:31 PMSubject: Re: [802.3_4PPOE] Are diode bridges really needed (2).pptxIt would seem to me that we need to define a new "Alt-C", which has all four pairs powered with a defined polarity (like Alt-B). This is similar to Christian's and Ken's suggestions, with a new name attached. Then we can either:- make full autopolarity optional in a .bt PD, but mandate some sort of notification (like an LED indicator) when it's plugged in backwards. One can argue that this implements some amount of functionality with either polarity.- make autopolarity mandatory in all PDs (as it is now), but the defined Alt-C polarity makes it pretty obvious what the PD should do if it doesn't care about full compliance.Dave