Archived from groups: comp.dcom.lans.ethernet (More info?)

I'm reading about PoE and I wonder why there are 2 alternatives (A and
B) to implement a PSE. Which are the benefits and drawbacks of each of
them?

I am designing an end-span Ethernet switch with PoE (PSE) and I don't
know which alternative is better. The IEEE 802.3af allows a PSE to use
either of the two alternatives, but not working simultaneously. So,
products the like in the market, which alternative do they implement?

It is cheaper to implement arternative B because you don't need a
special transformer. So, what's the point in using alternative A?

Regards,
Chesi

Archived from groups: comp.dcom.lans.ethernet (More info?)

"chesi" <cesteban75@gmail.com> wrote:

> I'm reading about PoE and I wonder why there are 2 alternatives (A and
> B) to implement a PSE. Which are the benefits and drawbacks of each of
> them?
>
> I am designing an end-span Ethernet switch with PoE (PSE) and I don't
> know which alternative is better. The IEEE 802.3af allows a PSE to use
> either of the two alternatives, but not working simultaneously. So,
> products the like in the market, which alternative do they implement?
>
> It is cheaper to implement arternative B because you don't need a
> special transformer. So, what's the point in using alternative A?

One option simply uses the four spare conductors in a Cat 5e cable for
power. Simple, right? Problem is, with 1000BASE-T, you don't have four
spare conductors. All eight wires are necessary for data transfer in
1000BASE-T. So this simple approach works for 10/100 Mb/s Ethernet
twisted pair systems only.

The solution for 1000BASE-T is to send power as DC along the same
conductors as data, as described in the spec, and this can also work
with the slower Ethernets.

Bert

Archived from groups: comp.dcom.lans.ethernet (More info?)

Albert Manfredi wrote:

(snip regarding power over ethernet)

> One option simply uses the four spare conductors in a Cat 5e cable for
> power. Simple, right? Problem is, with 1000BASE-T, you don't have four
> spare conductors. All eight wires are necessary for data transfer in
> 1000BASE-T. So this simple approach works for 10/100 Mb/s Ethernet
> twisted pair systems only.

> The solution for 1000BASE-T is to send power as DC along the same
> conductors as data, as described in the spec, and this can also work
> with the slower Ethernets.

This is commonly done for TV antenna amplifiers, including large and
small satellite dish amplifiers. Though in many cases it is done with
60Hz AC instead of DC to avoid electrochemical problems in a damp
environment. I am not sure where people use PoE, but it might also
be true there.

At TV frequencies it isn't hard to do with LC or RC filters.
It is harder to do with ethernet and keep that signals balanced.
I think my choice would be center tapped transformers, though I
believe it has been suggested that it isn't so easy.

This reminds be of a discussion before about whether 10baseT will
work with a 120VAC common mode voltage on the pair. I never got
around to actually (and carefully) testing it, though.

-- glen

Archived from groups: comp.dcom.lans.ethernet (More info?)

"chesi" <cesteban75@gmail.com> wrote:

>I'm reading about PoE and I wonder why there are 2 alternatives (A and
>B) to implement a PSE.

Can you point us to descriptions of the two options?

Archived from groups: comp.dcom.lans.ethernet (More info?)

"chesi" <cesteban75@gmail.com> wrote:
>I'm reading about PoE and I wonder why there are 2 alternatives (A and
>B) to implement a PSE. Which are the benefits and drawbacks of each of
>them?

Ah, found it at
http://poweroverethernet.com/assoc [...] 621476.pdf

/*
The spec does not allow both sets of wires to be used – a choice must
be made. The Power Sourcing Equipment (PSE) applies power to either
set of wires. The Powered Device (PD) must be able to accept power
from both options.
*/

If you are designing a gigabit switch I don't think you have any
choice, as you don't have any spare pairs. I suppose you could read
this as allowing transformer-coupled power over the 4/5 and 7/8 pairs,
but if you are going with transformer-coupled PoE, why not do it over
the 1/2 and 3/6 pairs and avoid the confusion?

If it's only a 10/100 switch you could probably get away with the
'spare pairs' thing, but I'd have to believe that phantom-power
Ethernet magnetics are going to be as easy to come by as the regular
type in the near future, and the phantom-power over 1/2 and 3/6
'feels' more compatable to me.

Dunno why they allowed the spare-pairs thing, must be some
backwards-compatability thing to some pre-standard kludge.

Doesn't someone use phantom power on all pairs to get 26W at the
terminal device? I suppose that's out of spec as well...

Archived from groups: comp.dcom.lans.ethernet (More info?)

"chesi" <cesteban75@gmail.com> wrote:
>I'm reading about PoE and I wonder why there are 2 alternatives (A and
>B) to implement a PSE. Which are the benefits and drawbacks of each of
>them?

Also found http://www.powerdsine.com/Products/PoE_Tester.asp which
tries to make it sound like switches use phantom power and midspan
injectors use spare pairs (and "PowerDsine High Power Midspan" does
both simultaneously).

Archived from groups: comp.dcom.lans.ethernet (More info?)

<William P. N. Smith> wrote in message
news:ncjhb19a7gr4sh7q2ta8m5rtv6slco03h8@4ax.com...
> "chesi" <cesteban75@gmail.com> wrote:
> >I'm reading about PoE and I wonder why there are 2 alternatives (A and
> >B) to implement a PSE. Which are the benefits and drawbacks of each of
> >them?
>
> Ah, found it at
> http://poweroverethernet.com/assoc [...] 621476.pdf
>
> /*
> The spec does not allow both sets of wires to be used - a choice must
> be made. The Power Sourcing Equipment (PSE) applies power to either
> set of wires. The Powered Device (PD) must be able to accept power
> from both options.
> */
>
> If you are designing a gigabit switch I don't think you have any
> choice, as you don't have any spare pairs. I suppose you could read
> this as allowing transformer-coupled power over the 4/5 and 7/8 pairs,
> but if you are going with transformer-coupled PoE, why not do it over
> the 1/2 and 3/6 pairs and avoid the confusion?

same is true for 10/100 that use Telco connectors (12 interfaces on a 50 pin
connector) - no spare pairs.

there are implementations that support more than one type of PoE - a lot of
the cisco stuff supports their prestandard system as well as 802.3af (they
more or less have to since a fair number of cisco end points like IP phones
havent been updated to 802.3af).

FWIW i cant see why you could put both flavours of 802.3af into a switch and
let end point discovery switch on the right version on a per port basis?

after all plenty of networks are built as multi vendor and the last thing a
switch manufacturer is going to want to do is to reduce the size of their
potential market.

i seem to remember that the spare pairs system was meant to be easier to
build for in a simple device (at the client end i think) - but it cant be
used everywhere so we end up with 2 flavours.
>
> If it's only a 10/100 switch you could probably get away with the
> 'spare pairs' thing, but I'd have to believe that phantom-power
> Ethernet magnetics are going to be as easy to come by as the regular
> type in the near future, and the phantom-power over 1/2 and 3/6
> 'feels' more compatable to me.
>
> Dunno why they allowed the spare-pairs thing, must be some
> backwards-compatability thing to some pre-standard kludge.
>
> Doesn't someone use phantom power on all pairs to get 26W at the
> terminal device? I suppose that's out of spec as well...
>
--
Regards

Stephen Hope - return address needs fewer xxs

Archived from groups: comp.dcom.lans.ethernet (More info?)

stephen wrote:

>> If you are designing a gigabit switch I don't think you have any
>> choice, as you don't have any spare pairs. I suppose you could read
>> this as allowing transformer-coupled power over the 4/5 and 7/8 pairs,
>> but if you are going with transformer-coupled PoE, why not do it over
>> the 1/2 and 3/6 pairs and avoid the confusion?
>
> same is true for 10/100 that use Telco connectors (12 interfaces on a 50
> pin connector) - no spare pairs.

I have also seen an installation where a single 4 pair cable is used to feed
two 10/100 connectors. Again, no spare pairs.

Archived from groups: comp.dcom.lans.ethernet (More info?)

chesi wrote:

> I am designing an end-span Ethernet switch with PoE (PSE) and I don't
> know which alternative is better. The IEEE 802.3af allows a PSE to use
> either of the two alternatives, but not working simultaneously. So,
> products the like in the market, which alternative do they implement?
>
> It is cheaper to implement arternative B because you don't need a
> special transformer. So, what's the point in using alternative A?

There may be situations where the spare pairs aren't available.

Archived from groups: comp.dcom.lans.ethernet (More info?)

stephen wrote:

> FWIW i cant see why you could put both flavours of 802.3af into a switch
> and let end point discovery switch on the right version on a per port
> basis?
>
> after all plenty of networks are built as multi vendor and the last thing
> a switch manufacturer is going to want to do is to reduce the size of
> their potential market.

According to the spec, linked to in another note, the powered device must be
able to work with either method.

Archived from groups: comp.dcom.lans.ethernet (More info?)

"stephen" <stephen_hope.xx@ntlxworld.com> wrote:
>FWIW i cant see why you could put both flavours of 802.3af into a switch and
>let end point discovery switch on the right version on a per port basis?

I'm not sure I'm parsing the above properly, but there's no need to do
both, as the spec (apparently) says the end-point has to support both,
so the switch only has to provide one.

[Plus or minus support for the pre-standard stuff, and the
post-standard high-power stuff... That's the wonderful thing about
standards, there are so many to choose from!]

Archived from groups: comp.dcom.lans.ethernet (More info?)

On 21 Jun 2005 01:13:58 -0700, "chesi" <cesteban75@gmail.com>
wrote:

>I'm reading about PoE and I wonder why there are 2 alternatives (A and
>B) to implement a PSE. Which are the benefits and drawbacks of each of
>them?
>
>I am designing an end-span Ethernet switch with PoE (PSE) and I don't
>know which alternative is better. The IEEE 802.3af allows a PSE to use
>either of the two alternatives, but not working simultaneously. So,
>products the like in the market, which alternative do they implement?
>
>It is cheaper to implement arternative B because you don't need a
>special transformer. So, what's the point in using alternative A?

The magazine EDN had a very clear article on the subject not so
long ago.
www.edn.com and look for PoE :-)

--
Kind regards,
Gerard Bok