Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [8023-POEP] System Architecture



Title: System Architecture

Hi Steve,

 

Thanks for the review. See my answers below.

Yair


From: owner-stds-802-3-poep@xxxxxxxx [mailto:owner-stds-802-3-poep@xxxxxxxx] On Behalf Of Steve Robbins
Sent: Wednesday, June 14, 2006 3:26 AM
To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: Re: [8023-POEP] System Architecture

 

Yair,

 

I have a few comments:

 

  1. I think we can all agree that the issue of current sharing is dead. 

Yair: We agree that current sharing can be implemented in many ways including functional isolation.

        From standard point of view it is dead issue.

        From practical point of view we can use current sharing without isolation hence Idiff (max current difference between pairs when the current is in the range of Icut) is required to be in the specification in order to allow any kind of implementation.

VERY IMPORTANT COMMENT: isolation between the 4P PD which is based on 2x2P is not mandatory. It depends if it environment A or B. Only B requires isolation. A can use isolation or active current sharing or nothing pending on the power and implementation.

----------------

We’ve already agreed (or at least I think we have) that a 4P system will consist of 2 independent 2P systems. 

Yair: Yes we agree on this concept.

-------------

The two power sources can have voltages that differ by several volts, and may be on completely different grounds, again with several volts difference. 

 

Yair: This is true for ENV B configuration. In ENV A the voltage difference is around 1-2V or less.

        Due to that facts we have two options: a) The PD will support worst case i.e. largest voltage differences.

b)       The PD will support max voltage differences pending the advertised ENV A or B.

         I would support in option b) since it allows using active linear means with low power dissipation in many ENV A applications which is the larger part of the market.

         Using option a) will kill a lot of current and future application.

-----------------

 So no form of linear current sharing (before the input to the converter inside the PD) will be practical because of heat dissipation. 

Yair: I agree however see my solution above. It is ENV dependent.

---------------

Therefore, any PD that draws power from more that one PSE must have isolation between the inputs. 

Yair: I disagree. See my distinction between ENV A and B.

--------------

 When you have isolation, current sharing is inherent, although not perfect; the converter with the lowest input voltage will draw more current, but the current ratio shouldn’t be too severe.

Yair: I agree. Isolation is one of the implementation to get inherent current sharing but not the only one in ENV A.

-----------------------------------------------------

 

  1. I can’t picture a PD that uses more than 2 PSEs.  I don’t think we need to worry about this “Nx2P” architecture.

Yair: This conclusion of general N pairs system is a direct outcome of the 2*2P concept.

We don’t have to address it in the standard. It don’t cost us anything. It is implementation issue and I might say a good outcome. It allows future expansion of the PoE technology.

The fact that we can not picture N pairs application now is normal in new technology. The issue is that our standard will not block new applications in the future.

--------------------------  

 

    1. I can easily picture a PD that has two converters inside, to draw power from two PSE sources.  Load sharing with two converters isn’t bad.  But load sharing with N converters becomes messy.  I don’t think anyone would want to bother with it.

Yair: Load sharing between N pairs can be done easily however it is out of scope for us now to show how to do it. The issue is how not to block it. When the need will rise simple solutions will be found too.

--------------------------

 

    1. If someone’s PD really needs that much power, I don’t see how PoE could be an economical solution.  Why not just use AC power?  Why run N cables, using up N ports on your Ethernet switch or midspan?  If the PD really needs so much network bandwidth that it needs N ports just to achieve the data throughput, then okay.  But using N ports just to get more power?  I don’t believe it.  Unless someone can site a practical example I think this topic is a waste of time, and we should just focus on 2x2P.

Yair: This is valid argument. Why people buy 1port PSE and not using wall adapter?. Probably they have found some advantages. You can never know.

------------

 

 


From: owner-stds-802-3-poep@xxxxxxxx [mailto:owner-stds-802-3-poep@xxxxxxxx] On Behalf Of Yair Darshan
Sent: Tuesday, June 13, 2006 4:15 PM
To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: [8023-POEP] System Architecture

 

Hi all,

Please find my action item for tomorrow's meeting.

It is revised presentation from May meeting that focus on the concept 4P=2x2P and the more general case of N Pairs = Nx2P..

Please feel free to comment.

Thanks

<<Flexible PD implementation driven Architecture 002.pdf>>

Yair

Darshan Yair
Chief  Engineer
PowerDsine Ltd.  - 
The Power over Ethernet Pioneers

1 Hanagar St., P.O. Box 7220
Neve Ne'eman Industrial Zone
Hod Hasharon 45421, Israel
Tel:  +972-9-775-5100, Cell: +972-54-4893019
Fax: +972-9-775-5111

E-mail: <mailto:yaird@xxxxxxxxxxxxxx>.  
<http://www.powerdsine.com>