Is IEEE 802.3af PoE truly downwards compatible?

Hi Ramon,

I suspect you are confusing the various elements of 2 different environments, DSL and Cable, and the bearer encapsulation being used on either (PPPoA or PPPoE). Also, your use of the term "PoE" is quite confusing, because to me, PoE means Power over Ethernet, something quite different.

You may find your situation to be isolated to your specific environment, because in 7 years of using DSL I have never "lost" the Phone line that carries DSL (except once when a digger chopped the cable just down the road....;-)).

The interesting thing with DSL is that its a digital methodology deployed over what is essentialy an analog environment. Surprisingly, from a transmission perspective, Cable and DSL have a lot of elements in common.

Both CABLE and DSL employ a Transmisison environment for the WAN component, and they only really differ in how that environment itself is built (the modulation schemes differ). DSL is based on ATM, that is also deployed in other transmission environments, however Cable uses what was originally a proprietary, now published for all to use, completely different scheme The one common factor is that BOTH only use a SINGLE copper pair (in the general domestic market) for the WAN part of that delivery. The main difference between them here is the Modulation scheme they each use to deliver the higher layers (the higher layers being the PPP part of the composite).

At the copper level, it is well known that when you combine 2 modulation schemes on a single copper pair, there WILL be some interaction between the 2 signals, that may cause an issue with one of the environments. That is why full DSL requires a Splitter to be used to help isolate the environments and allow older POTS hardware to operate in the presence of the VHF signals that are not "normally" present on the copper, and why "spitterless" DSL is just a way of lowering the effect of that interaction on older H/W, but there is still an interaction taking place.

The major effect that humans MAY notice is an effective REDUCTION in signal levels for voice communications (POTS) on the Wire. This means (to me) the voice level received at my home from the caller is slightly reduced compared to when the phone line did not also have ADSL present, however this reduction has never been great enough to completely kill the POTS service.

Assuming you mean PPPoE, well then PPPoE has nothing at all to do with that actual transmission mode of the WAN to reach your Home, however it DOES have something to do with the ENCAPSULATION of the data stream ONTO that WAN component. So I then suspect you really mean "Cable" as the WAN component, as you can do either PPPoA or PPPoEoA, for DSL as the WAN transport, and for Cable you usually find PPPoE or one of the RFC Bridge styles being used, however the reality is that NEITHER cable nor DSL (for consumer grade use) are yet capable of 1GB WAN transport! So if you are looking at Cable, then you are still looking at a SINGLE Copper cable pair delivering the WAN component to your home, exactly the same way that DSL uses ATM technology for DSL.

So your provider does NOT deliver Ethernet using UTP cabling to your home.

The part of your question that reads -

is actually irrelevant, as this has nothing to do with PPPoE (or PPPoA) at all. You have to dig LOWER in the ISO standards to find where DSL (or Cable) comes in.

Nope, PPPoE or PPPoA just ride on the back of DSL or Cable or Ethernet itself. The reality is that these are just ENCODING schemes to package up data, and have NOTHING to do with the actual transmission media being used.

Pehaps a good place to start is to try and understand the bottom 3 layers of the ISO 7 layer protocols and see exactly where DSL/CABLE/ PPP all sit in that structure.

One other thing to consider, is that because DSL is based on ATM technology, and DSL v1 was approx 8Mb, DSl 2 go up to around 24Mb adn DSL 2+ even further, then consider that ATM itself goes way above that to 500+Mb then speed enhancments to DSL technologies look to be far more likely than when using Cable technologies, which require a re-write of the Cable protocols to go anywhere near these speeds. Then again, things can change pretty quick in this world....;-)

I hope this helps..............pk.

I've got many customers that have probably have had years of zero problems over all time of them using shared DSL/POTS service... It usually does work flawlessly, as much as the LEC will do it.

There are several modes of operation for 802.3AF. The standard is free to get as part of the 802 package from the IEEE (well, at least at one point in time it was). One mode, operating with Gigabit Ethernet, does use all 4 pair for signal, as well as PoE. It does really work with the proper gear.

If a device isn't 802.3af aware, it doesn't complete the initial handshake, and the head-end unit won't send down more than miliamps of current needed just to start talking. The end device has to request how much juice is sent down its way through the protocol negotiation. Normal devices should be able to deal with the initial startup protocol query probes just fine. I've never seen any issue.

There are other modes of operation pushing the current over just the unused pairs in the older/slower speeds of ethernet. They still generally won't do it (especially in 802.3af, or in cisco pre-standard implementations) until the protocol handshakes are complete (ie. CDP in cisco pre-standard).

Other non-standards based gear might not necessarily follow any rules, and just blast juice down the unused pairs. (ie. older Inter-Tel VOIP phone setups).

Get the standards to learn..

Ramon F Herrera hath wroth:

Most such arrangements have no interaction between the signals on different pairs.

Actually, I've seen DSL without dialtone. It's rare but still possible. The only thing that is shared between the CO (central office) and the subscriber equipment is the copper cable. The low frequency audio (300-3000Hz) is separated from the DSL carriers (150KHz to perhaps 1.5Mhz) with simple low pass filters. It enters the CO at the wiring board, and is immediately separated into the POTS (plain olde telephone service) audio, which goes to the telco switch, and the DSL carriers, which go to the DSLAM (digital subscriber line access mess).

Some 802.3af implementations works quite nicely with Gigabit ethernet. See general description of 802.3af at:

"A "phantom power" technique is used so that the powered pairs may also carry data. This permits its use not only with 10BASE-T and 100BASE-TX, which use only two of the four pairs in the cable, but also with 1000BASE-T (Gigabit Ethernet), which uses all four pairs for data transmission. This is possible because all versions of Ethernet over twisted pair cable specify differential data transmission over each pair with transformer coupling; the DC supply and load connections can be made to the transformer center-taps at each end. Each pair thus operates in "common mode" as one side of the DC supply, so two pairs are required to complete the circuit. The polarity of the DC supply is unspecified; the powered device must operate with either polarity or pair 45+78 or 12+36 with the use of a bridge rectifier."

Note that there are many PoE devices and adapters that will only work with dedicated power wires (10base-T and 100base-TX only) as well as those that will also work with 1000base-TX. There are also adapters that function as a PoE (power over ethernet) device with specific devices, but are not 802.3af compliant. Check the specs first, especially if you mix and match.

Ethernet is transformer-isolated on both ends of the twisted pairs. To do PoE, they center-tap the isolation transformers on both ends and use those taps as a DC path. The PoE hub applies about 50 volts DC between the ct's of the transmit and receive pair (at a random polarity!) and the end device picks it off, rectifies it for luck, and drives an isolated dc-dc converter to get local power. There is a simple dc-current handshake that lets the end gadget inform the hub about its PoE capabilities.

Standard PoE is good for something like 15 watts. There's a PoE+ or some such that uses 4 twisted pairs to get more power.

There are a number of PoE management chips, and their datasheets and appnotes have details. National has some good ones.

As far as I know, PoE works fine. The balanced DC running down each pair doesn't disturb the ac-coupled data or annoy the magnetics.

John

Your analog phone line from the phone company supplies DC power as well, as well as just about every PBX on the market to power the phone at the end. Supplying DC power does not affect the signal that is being transmitted over it.

He's talking about Power over Ethernet - IEEE 802.3af - see the subject line.

Sam

Doesn't happen here in the UK, or at least not that I've noticed. DSL certainly works successfully in many places which suggests it's not fundamentally flawed.

Yep, it works. If you read up on how Gigabit Ethernet works over copper (5-level trellis coded signals using each pair in both directions simultaneously - I believe it was developed by people with experience in modem design) you'll find that adding DC offset is trivial by comparison. may be enlightening.

The standard goes to some lengths to ensure they don't. Non-standard implementations may break things.

Both are options.

Sam