Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Handhelds Hardware

Power-over-Ethernet: IEEE 802.3af Draft 172

SkewlD00d writes "Want power (~10W) and your fat 10/100Mbps pipe (1Gbps maybe) on the SAME jack. Surf the web and recharge your handheld w/o finding its clunky base/transformer, home automation, and control systems are just some of the possibilities. Using suitable hubs/switches or dongle, power can be provided over unused pairs, or possibly raise the signal offset above DC, X10-style (no link, u know why). See IEEE 802.3af DTE Power via MDI Task Force. Maxim (not that one :o) already has some goodies. They have schematics for 8.25W/12.5W single-output and 14.2W triple-output 802.3af PSU. Try FindChips if you're looking for them; they're in production and are NOT vapor-ware!"
This discussion has been archived. No new comments can be posted.

Power-over-Ethernet: IEEE 802.3af Draft

Comments Filter:
  • by ZorinLynx ( 31751 ) on Monday May 19, 2003 @08:03AM (#5990794) Homepage
    Not sure how standard it is, but many Cisco switches support Inline Power, which is used by their Voice-Over-IP Phones. We deal with this constantly where I work.

    It can be annoying because if you want to run several phones on one drop, they need power supplies if you don't want to use an expensive inline power capable switch.

    • Cisco uses unused pairs; hence, you can't have Gig-over-copper AND POE. Soon, soon...

      You can purchase a power injector from several vendors who provide POE-capable devices -- good for, as you say, phones, and wireless access points.

      I'd like to see a standard accepted for this, because as far as I know, the only products from the larger vendors (Cisco, Avaya) are propritary, so you can't use other POE-cabable devices from a different vendor on the same port.
      • Cisco uses unused pairs; hence, you can't have Gig-over-copper AND POE.

        Actually, they are using the data pairs. At least when you are providing power via the WS-X6348 line cards. I'm not sure that they don't use the unused pairs with the dumb power injectors.
    • by Zak3056 ( 69287 ) on Monday May 19, 2003 @09:07AM (#5991161) Journal
      It can be annoying because if you want to run several phones on one drop, they need power supplies if you don't want to use an expensive inline power capable switch.

      Have you looked at these [3com.com]?

      They're basically four-port switches that you mount in place of your normal wall jack. They cost about a hundred bucks, support power over ethernet, and there's even a managed version.
      • You should read that page more closely. These switches don't provide power over ethernet, they use it. They still require the expensive upstream switch unless you want a power supply for your wall jack.
        • You should read that page more closely. These switches don't provide power over ethernet, they use it. They still require the expensive upstream switch unless you want a power supply for your wall jack.

          You should read the parent more closely. His problem is he already HAS the expensive upstream switch, and runs into problems when he needs to place multiple IP phones at a single drop location. The 3com switches require power over ethernet, yes, but they also support passthrough to whatever devices are con
      • I've looked into those very jacks with an existing home installation I have... problem is, I want to future-proof the place as much as possible by having a clear upgrade path to gigabit ethernet (over copper... no way I'm crawling around in all the fiberglass to rewire the house with fiber).

        PoE uses the unused pairs (in regular 10/100Mb ethernet) to provide the juice... gigabit over copper requires those pairs to carry data.

        Sooo.. you can have PoE and use it to power those jacks (a very sexy idea, I m
    • Yes, Cisco already does this. But they do it in the pre-standard fashion. (I believe this is in the resistive vs. capacitative methods of sensing the presence of a powered device, such as a phone or an access point).

      The newer powered systems (3750 switches etc) are being held until the draft is finalized so that we'll get real inter-operability.
  • by Anonymous Coward
    from here [currenttechnologies.com]

    Ethenet-over-power, short range for home and long range for ISPs, currently being deployed in the U.S. Check it out.
    • by spacefight ( 577141 ) on Monday May 19, 2003 @08:26AM (#5990910)
      Quote from your link:
      "Powerline communications (PLC) technology uses the existing local electric wires to provide digital communication services such as high-speed Internet access, voice over IP (VoIP), video, and in-home networking."

      We're talking about power-over-ethernet here, not ethernet-over-powerline.
    • Ethenet-over-power, short range for home and long range for ISPs, currently being deployed in the U.S. Check it out.

      That's the other way round from the article! It isn't Ethernet over power, it's data over powerlines; this article is about sending electrical power over Ethernet wiring. Both are interesting, but have very different applications; power over Ethernet is already being used to power low-power devices, such as VoIP telephones and wireless access points.

      Power over Ethernet is limited to a matte

  • by account_deleted ( 4530225 ) on Monday May 19, 2003 @08:07AM (#5990817)
    Comment removed based on user account deletion
  • by kriegsman ( 55737 ) on Monday May 19, 2003 @08:13AM (#5990841) Homepage
    Here's a useful how-to [nycwireless.net] on powering wireless base stations using Power-Over-Ethernet (POE). Several wireless vendors already sell POE kits for wireless APs. Here's a POE kit for the AirPort [devdepot.com].

    Power+Ethernet over one wire is a great combination for settting up APs, because many of the interesting places to put a wireless access point (e.g., in the attic, by the pool, in the middle of the ceiling) may not have a power outlet nearby, or you may simply not want to run two sets of wires (AC power + ethernet) to the location.

    -Mark
  • I can just see it -- I fry myself inserting the Ethernet cable in the card. Instructions read: before inserting, go to your power panel, can shut off the circuit breaker for your computer and then...
    • by swb ( 14022 ) on Monday May 19, 2003 @08:40AM (#5990977)
      While overstated, I wonder what the power safety capabilities are in some of the already-available implementations of this.

      I'm not as worried about shock, but about unintentional shorts burning out switch ports or misplugged equipment. It'd be awesome if there were software-resettable breakers on the switches so that you wouldn't lose line cards. I had the misfortune of accidentally shorting a couple of Merlin phone system ports and those ports are dead, dead, dead forever.

      I also wonder what it does to existing wiring codes and cable plants if any significant power starts to become present on data cabling. It'd suck to become dependent on this power only to have data cabling drops become more costly due to tighter code restrictions on where they can be placed.
      • by bdc0 ( 541532 ) on Monday May 19, 2003 @09:18AM (#5991220)
        The hot-swap chip is required by the standard and handles unitentional shorts. From the data sheet: "Provides Safe Hot Swap for +10V to +65V Power Supplies". This also 1) limits maximum inrush and 2) limits steady-stage current to 1) avoid "drooping" the supply voltage and 2) handle faults on the secondary (e.g. 6 V) side. I haven't looked at the standard recently, but there are supposed to be limits on the supply (-48 V) end to handle faults there. The reasons -48 V is used are it is very standard in telecom so low cost parts available for DC/DC converters and it is less than +/- 60 V which is the limit for "low voltage" in existing wiring codes.
      • I believe anything under a certain voltage is exempt from being subject to the electrical code. 35 volts seems to spring to mind. It probably varies from state to state. There may also be amperage restrictions.

        As far as protection goes, a simple thermal breaker would work, and be self resetting.
    • "I can just see it -- I fry myself inserting the Ethernet cable in the card. Instructions read: before inserting, go to your power panel, can shut off the circuit breaker for your computer and then... "

      I don't see why this is a troll. I got shocked by plugging a phone line into a modem once years ago. Felt pretty stupid.
  • >X10-style (no link, u know why)

    No I don't! :-/
    Why??
  • by wario78 ( 572319 ) on Monday May 19, 2003 @08:19AM (#5990868)
    The latest firewire standard can do 800Mb and provide up to 45W of power
    • And of course, with some systems, push IP [slashdot.org].
    • by Anonymous Coward
      I've had a difficult time in the past getting my firewire cable to plug into the rj-45 connector on the back of my access point.
    • Firewire does it over 10m, POE does it over 100m with Cat5 cabling (well at 10W maybe even over Cat3). I know when I was working with the Cisco/Aironet guys that they had a customer who was having some problems with POE, turns out they were trying to use 100m drops of ancient barely Cat3 cable to power their AP's in some location, problem was that this was there wasn't enough power for the boot cycle, if we attached a power module and booted it then removed the power then it would use the POE just fine but
    • The latest firewire standard can do 800Mb and provide up to 45W of power

      But does it work over several hundred feet? Nope.

      Also, are you sure about 45W? That's 9 amps at 5V. You're going to see sparks if you yank that cable out. They would need to have designed a new connector too. I doubt the original firewire connector was rated for that much current. I know a USB connector can only handle 2 amps and they are a very similar design.
  • by stevenbdjr ( 539653 ) <steven@mrchuckles.net> on Monday May 19, 2003 @08:20AM (#5990871) Homepage

    I use this extensively with my Lucent/Orinoco wireless APs. Been doing it for about 2 years now. It's also part of 3Com's Ethernet switch/jack technology [3com.com], and has been implemented for VoIP.

    I've found it to be a godsend for my wireless APs, as I can mount my APs near the ceiling (away from students), and not have to worry about finding an outlet, or having them unplugged by the cleaning crew!

  • Allows power to be distributed where there's no mains sockets (saves on wiring), but also has the potential to damage hardware if a short in the cable occurs.
    • From the post, they are talking about around 10 Watts, Assuming that they are pushing that at 12V (Sorry, not read the spec - it's not worth the $80-odd the IEEE wants for it!!), that comes out at a little over 830mA.

      I can't remember the Power / Voltage / Current ratings of 10/100/1000 Ethernet off the top of my head, but I'd say that a short accross Power to one of the Ethernet conductors should just cause data to 'stop' in the direction of the short (Rx / Tx).

      I say this because I have frequently (ac
      • It seems like an integral part of this standard is current limiting circuitry that will prevent shorts from doing damage.

        At least the Maxim ICs there appear to have a number of safety features.
  • Sweet! (Score:1, Offtopic)

    by mschoolbus ( 627182 )
    If they just made it a little thinner and lighter, we could have wire Remote Control Airplanes! Or even Boats!
  • Interesting (Score:1, Offtopic)

    by EpsCylonB ( 307640 )
    We have always had trouble using power lines for sending and receiving data not beacuse it is impossible but because our current power lines were never designed to carry data.

    Imagine a merging of the current ethernet and power line standards, you could potentially solve the last mile problem and and at the same time open up new backbones.
  • by JKR ( 198165 ) on Monday May 19, 2003 @08:25AM (#5990902)
    Is this going to work with cheap installations which are already using the extra pairs in Cat5? RS sell a splitter / combiner which (as a last resort) lets you use the "spare" pairs in an existing 10/100 base-T run to run a parallel 10/100 base-T connection.

    I can smell burning...

    Jon

    • Is this going to work with cheap installations...

      That's why you aren't supposed to do that.

      Those spare pairs aren't there for such hacks as you (and, admitedly, I) do. They generally are intended for future improvements such as this one.

      Oh, and running two 100 base-T connections in one CAT5e jacket is non-standard... expect cross-pair interference to be a problem if you run any appreciable distance. Yes, we all do this kind of thing in a pinch, and yes, it mostly works. It's when it doesn't that it be

    • Is this going to work with cheap installations which are already using the extra pairs in Cat5? RS sell a splitter / combiner which (as a last resort) lets you use the "spare" pairs in an existing 10/100 base-T run to run a parallel 10/100 base-T connection.

      It MAY work - or at least not fry it. And it will probably continue to work if you don't want to use the power. But you'll need another device to inject/extract power on the "second line" if you want to use it for both power and signals. And the dev
  • by silas_moeckel ( 234313 ) <silas&dsminc-corp,com> on Monday May 19, 2003 @08:25AM (#5990904) Homepage
    OK this is a good idea from my perspective. I have used Cisco IP phones w9ith this and they work great it's a LOT easier than a UPS for every IP phone (look at a sales guy and try to explain why the phone died when they power took a hit phones need to be reliable) I have used it with 3com AP's again it works great and makes it easy to stick the AP on the drop ceiling and other hard to reach places PoE again can provided needed UPS power to the AP it also has the side benifit of the AP isn't up and running unless the network is plugged in (I have seen it happen it's not pretty)

    I'm waiting for a laptop that can suplement/charge itself over PoE I have a subnotebook that only draws 14 watts charging PoE is close to that possibly enough for a trickle charge. It would seem to be a cheap addon to a laptop ethernet. A lot of people in offices hook laptops up to PSU but leave them on the wireless because it's easier this could make a one cable solution. I can see a whole line of terminals as well for POS applications. In reality on a lot of things this could replace the charger for small devices like PDA's and hook them up to the network at the same time.

    Now it seems to be incompatable with GigE over copper since that uses all 8 wires.
    • Now it seems to be incompatable with GigE over copper since that uses all 8 wires.

      The power supply seems to use BOTH halves of each pair for ONE power wire - the "common mode" - allowing you to also use the differential mode for the signal.

      You need a center-tapped transformer to inject or extract the power if you're using the pair for both signal and power. Capacitors are adequate if you only want the signal. (Note that transformers inserted into a GigE run to inject/extract 20ish miliamps will have to
      • Your right I think you could add power to GigE but I dont beleive it's part of the spec just yet. http://www.cisco.com/en/US/products/hw/modules/ps 4 835/ps5173/index.html is cisco's high end blade and it says gige or inline power so I think it's a choice either or. Granted technicaly it wouldent be so hard to put signal and power down the same line (I'm used to phantom power for Mic's)

        Now Legacy devices are supposed to support ring signal on the center pair (yes they designed it to run phone over as wel
  • yawn (Score:4, Funny)

    by SlashdotLemming ( 640272 ) on Monday May 19, 2003 @08:28AM (#5990924)
    Wake me up when they invent wireless power.

    :)
    • by Cee ( 22717 )
      Wake me up when they invent wireless power.

      It's possible to use microwave links... Just don't stand in it's way or you'll get toasted.
    • Re:yawn (Score:3, Funny)

      by john82 ( 68332 )
      "Large powerful orb discovered in the sky. Thousands in New England turn beet red after prolonged exposure."

      Film at eleven.
    • Wake me up when they invent wireless power.

      If you're looking for wireless power, this [arizona.edu] will knock your socks off! :)
      • If you're looking for wireless power, this will knock your socks off! :)

        Oh, now that is suiting.. Nice place to grab info about the sun from... UofA! (University of Arizona)
    • Re: (Score:2, Funny)

      Comment removed based on user account deletion
    • www.splashpower.com, not exactly a long distance solution, but it's the first step and is great for PAN's
    • Some farmers in the great plains discovered that if they coiled barbed wire underneath the massive power lines running along their property, they could power their houses for free.

      The power company was baffled be the unexplained power drain for months.
  • by dfenstrate ( 202098 ) <(dfenstrate) (at) (gmail.com)> on Monday May 19, 2003 @08:30AM (#5990934)
    to implementing power over TCP/IP.
    Read the RFC Here. [ietf.org]
  • TI also has chipset (Score:4, Informative)

    by asmithmd1 ( 239950 ) on Monday May 19, 2003 @08:35AM (#5990951) Homepage Journal
    here is a link to TI's PDF datasheet [ti.com] only $1.25 in qtys of 1000 for the device side, much cheaper than a wall wart
    • This isn't a fair comparison. The $1.25 is only the hot-swap chip. Then you need a -48 V to (for example) 6 V isolated DC/DC converter. See the p. 10 of the referenced TI data sheet for a more details. I expect that a power-over-ethernet solution would cost >$10US since -48 V parts aren't in the huge huge volume of wall warts.
  • So when can we have power over bluetooth? ;)
  • by urrbanlleg-end ( 542756 ) on Monday May 19, 2003 @08:36AM (#5990957)
    Sweet ! Now you can get ethernet over power over ethernet !
  • by crovira ( 10242 ) on Monday May 19, 2003 @08:40AM (#5990976) Homepage
    Imagine getting a kilovolt (at a kiloamp) back from a web site that's being slashdotted.

    Or better yet. Imagine sending THAT as a response to some Spam.

    That would take care of the problem PDQ.
  • No way (Score:2, Funny)

    by Hard_Code ( 49548 )
    No way am I electrifying my dongle again.

    Fool me once, shame on your. Fool me the fifth time, shame on me.
  • by Chriscypher ( 409959 ) <slashdot@metamedi a . us> on Monday May 19, 2003 @08:42AM (#5990987) Homepage
    Number one cause of device failure for SCSI devices: power shorts on SCSI chain.

    Number one cause of device and motherboard failure for Firewire devices: power shorts on Firewire chain.

    Notice a pattern?

    There's good reason to have two cables running to each device: when a cable inevitably goes bad or the user mis-mates a hot connection, the chance for device damage is minimal.

    • Simple. Properly designed hotswap hardware should have current limiting circuitry built in.

      Similar to a fuse/circuit breaker (Good to have those in there too!), but they don't permanently shut down.

      I work in an RF power amplifier development lab - Trust me, current limiting power supplies are a reality and they are a truly wonderful thing. (Yeah, we short out a lot of stuff here. :)
      • Simple. Properly designed hotswap hardware should have current limiting circuitry built in. [or at least a fuse.]

        Yes, properly designed equipment should have these things. And that should have been true for SCSI-I also.

        SCSI-1 had a 50-pair cable, with (in general) one side of the connector carrying signal/power, the other ground. And is was supposed to be set up so:

        * The power was supplied on the middle conductor of the signal/power side.
        * The middle conductor of the ground side was left unconnec
    • by adolf ( 21054 ) <flodadolf@gmail.com> on Monday May 19, 2003 @01:38PM (#5993051) Journal
      Have you tried plugging an RJ45 in backwards (or sideways, or shifted one row to the left, or...)?

      It is essentially impossible, approaching non-issue status. And nevermind that ethernet is specified to withstand multi-kilovolt fuckups between random pins, unlike massively-parallel/easy-to-crossconnect SCSI - ethernet devices don't mind seeing voltage potential where it doesn't belong.

      Here's a larger problem that standardized POE solves:

      User has two cables running to desktop IP phone. User unplugs cables to re-arrange desk, and upon plugging things back in, their IP phone smells funny and their external modem won't work.

      Upon investigation, it turns out that the 12VAC power supply for their Sportster is not interchangable with the 13.6VDC PSU for their IP-600 phone.

      Therein lies the problem with multiple incompatible, anonymous-looking power supplies at the desktop.

      (In other news: Licking light sockets can be hazardous to your health. Also, this just in: Consuming too much water may cause accidental drowning.)

  • More creative ways to toast a computer using a network... This is truly a BOFH's friend. It's a directed etherkiller! Wow.

    :-)

  • I'm surprised I haven't seen a comment on wireless yet.. It's the perfect solution for putting your AP in an 'out of the way' location.

    I was just looking at this item [dlink.com] from Dlink for doing just that sort of thing. A quick search shows it's only around $50 on the 'net.

  • by Anonymous Coward

    Would be great for my Palm Tungsten C (-:

    Seriously, this should be possible in another way: put a magnet in the Palm, have a huge magnet somewhere in the room that changes its poles rapidly which makes the Palm magnet turn... voila!

    Or is that far-fetched? ;-) Maybe some other devices in the room will not like the big magnet ;-)
  • by Syris ( 129850 ) on Monday May 19, 2003 @09:34AM (#5991307)
    Linear Technology [linear.com] makes some Power-Over-Ethernet IC's. Like this one [linear.com].

    The Linear typical app seems to be much smaller, with fewer external components because it's actually optimized for the application.

    The Maxim chips appear to be generic hot-swap IC's not optimized for power-over-ethernet. Using slashdot is an interesting way to publish an app-note and get some publicity.

  • Just download IEEE_802.3af_FULL_REAL_WORKING_VERSION.exe
  • Is it just me (Score:1, Insightful)

    by Anonymous Coward
    or does the synopsis for this article look like it was written by some lame AOL script kiddie?

    No I am not trolling, I think we should edit these synopsises before they go live.
  • Unimpressed. (Score:3, Interesting)

    by dasmegabyte ( 267018 ) <das@OHNOWHATSTHISdasmegabyte.org> on Monday May 19, 2003 @10:13AM (#5991499) Homepage Journal
    I already have tethered power. But riddle me this, batman: when am I gonna get power over WIRELESS ethernet?

    I'm willing accept a little sparking.
  • lets hope that wireless power is next ;)
  • I want to surf the web on my wireless handheld and have it recharge.

  • wireless (Score:1, Redundant)

    by gumbi west ( 610122 )
    I can't wait until they update 802.11b to include this.
  • I have been selling these at BrightNoise [brightnoise.com] for use with IP enable cameras and APs for some time. They work great. In fact Axis is re-badging them as their own.

    They save you a ton of money on electrical costs and general hassle. There is a web enabled version that allows you to power cycle individual ports as well as set critical ports for UPS use.
  • 3com's IP phones support doing this, as does their MBX telephony hardware.

    I went to a 3com IP phone training session about two and a half years ago in which we learned to set up their telephony boxes. Most of the guys there were telecom installers, guys who'd been servicing telephone equipment for years but hadn't used computers more than the average guy, so I saw things from a different perspective. Neat hardware, quite flexible, and ran some form of *nix, but configuration was through a tastefully design
  • I have a Soekris [soekris.com] net4521, and it works great with PowerDsine's [powerdsine.com] PD 6001 [powerdsine.com] (part of their PD 6000 [powerdsine.com] mid-span series).

    They (PowerDsine) have been doing 802.3af since its earliest drafts, and it's been working as designed.
  • by morcheeba ( 260908 ) on Monday May 19, 2003 @12:04PM (#5992303) Journal
    This spec has been out a while now... 2 years ago PowerDsine made a prototype ethernet-powered razor [zzz.com.ru] - here's coverage from more traditional network press [nwfusion.com]
  • The Gibson Magic [gibson.com] spec included 802.3af support - that's how they're getting the power to their new Les Paul [gibson.com] guitar that sends out an 24/96 audio per string. Makes sense - it's bad enough having an RJ-45 socket on your guitar, even if ruggedized, you don't want extra power cables too...
  • Does this mean that RFC 3251 [faqs.org] is just around the corner?

    RMN
    ~~~
  • Who uses 44+ volts DC for anything? It seems like just going with 5V would be better, and less likely to fry stuff with shorts in the cable; if a 5V line shorts to a signal line, the signal is held high, and your cable just doesn't work, without taking anything out of spec. Of course, 350mA at 5V is too little power, but increasing the current a bit would be reasonable with a 10-fold decrease in voltage.

    What exactly are they planning to power with this? Embedded devices and PDA-level devices (like wireless
  • Power over ethernet over power over ethernet over power over ethernet over.....

    With all them layers, you press the light switch and, after 5 seconds (unless you're a low ping b******), it finally turns on.
  • by Hercynium ( 237328 ) <HercyniumNO@SPAMgmail.com> on Monday May 19, 2003 @05:00PM (#5994538) Homepage Journal
    ...mention of the BOFH!!! [uni-wuerzburg.de]

    Another relevant fave found here. [iinet.net.au]

    The whole [theregister.co.uk] shebang. [ntk.net]

  • The submitter put SO MUCH editorial in his submission that no further comment was needed.

    Now you know how to stop the editorializing, stop bitching about it and let the rest of us enjoy it.
  • So which is it, phantom power, which has been around the audio world for practically forever, or did somebody finally notice that for years now people have been installing (and paying extra for) 8 wires to do the job of 4?

If you didn't have to work so hard, you'd have more time to be depressed.

Working...