PoE Injectors

[Dave Liquorice]

Hi,

Does anyone know of a PoE injector that doesn't let the magic smoke out if the power glitches? Local community broadband lost four more yesterday that's on top of the 40 odd in the previous six months. That's a significant amount of money at £20 each...

They have used Fairway Electronic Co. Ltd VE20C-48A and SL Power Electronics PW182RE4800F02 with negliable difference between 'em. These are (supposed to be...) IEEE 802.3at (note t not f(*)) devices, IEC mains in, ethernet and ethernet + (proper) PoE out.

(*) That is PoE+ or PoE Plus which can deliver up to 25 W to the remote device rather than the 15 W of IEEE 802.3.af.

[Capitol]

UPS time?

[Dave Liquorice]

smoke

Most small and affordable UPS's are offline and normally the downstream devices connected directly to the incoming mains. Glitches can still reach the connected kit, indeed a couple of PoEs that died the other week were "behind" such a UPS.

With an online UPS the downstream kit is powered by the invertor all the time not just when the supply power is off/too low.

I've yet to see a dead example but the smoke comes out the RJ45 end so I suspect the mains disturbance is causing drops/surges in the 48 V and the controlling circuitry for that dies big time.

[tony sayer]

In article , Dave Liquorice scribeth thus

If its a 24/48 volt POE system Dave some makers do DC ones where you could run the POE's at the headend off 12 volt batteries 2/4 series connected.

We did that with some Ubiquity equipment's saved a lot of bother with UPS's in places that are a bit rough for 'em;)...

[Andrew Gabriel]

Have you had them checked by the manufacturer (or anyone else) to diagnose the fault?

If the wires go outside the building, what lightning protection do you have on them at the building entry point?

[Dave Liquorice]

48 V but I'm under the impression that 802.3at is active not passive. ie the full 48 V 25 W supply is not applied until the two ends have had a little chat and agreed about what is required. I don't know if just feeding a (current limited) 48 V up the right pairs would "just work".

UPS and keeping things up for a short time is nice but not essential. If the power has gone in an area most users won't have power for their end of the wireless link so won't notice that the AP is down as well. B-)

It's the fact that the injectors leak the magic smoke when the mains has a minor whoopsie that is the real problem. It's geographic as well, one village has virtually no problems, the town and a couple of other villages do. Everywhere is fed via various 11 kV lines from the same 33 kV substation. I wondering if there is a fundemental problem in/around the 33 kV substation or 11 kV switching.

From googling about the other day it seems that Ubiquity kit tends to be a bit non-standard. The kit in question is Proxim 5054 and the like.

That looks passive to me and most places only have two or three bits of PoE kit.

[Dave Liquorice]

The manufacturer is just one of a million "generic" PSU makers in China, It's not unknown for them to be DOA. I hope to get a look at a few samples of the dead next week.

None but I'm pretty sure this isn't lightning or static damage as we tend not to get thunder storms up here. But ice/trees on the lines do cause auto reclosers to play. The fault currents may well cause spikes in the distribution.

[Capitol]

I'd put a transient detector with logging capability on the supply to see what is happening. Are you using surge suppression?

[tony sayer]

48 volts and 25 watts?. Seems a lot unless its feeding more than the one unit!.

The max power permitted in 5.8 Ghz band C is 4 watts effective radiated and theres going the be quite a bit of aerial gain in that sort of unit..

Indeed...

Does seem rather poor. Have you/they sent any units back to the makers for comment or inspection at all?..

Dunno thats non standard..

Well in that instance it's providing for a few more but its run of a batteries and a charger method ...

[Andrew Gabriel]

The PW182RE4800F02

claims to be 19W max, and doesn't claim to conform to any PoE standard (specifically says it doesn't conform to IEEE802.3af).

Looks like it just applies power direct to pairs 2 and 3. If that's not the case, it doesn't say what protocol it uses to negotiate.

What kit is on the other end consuming the power?

Or a spike on the LV phase due to damaged 3-phase neutral?

[Dave Liquorice]

48 V 25 W is the spec for 802.3at (PoE+). There is more than just the radio in the unit but all the other electronics for an AP, Bridge, or Customer unit. The Proxim MP.11 5054 units are rated at "20 W max" and no mention that I have found of 802.3at or af. The smaller customer unit 5012 *does* mention 802.3af (top whack 15 W).

Not that I am aware of, I suspect a certain amount of "it's not worth it for a 20 quid PSU, which if it was just one a year would be acceptable. 40 in six months is another story. Unfortunately the old busted ones have gone in the bin...

[Dave Liquorice]

Varies but Proxim MP.11 5054 which states "20 W max". A 19 W max supply is sailing a bit close to the wind IMHO... Bit of a power glitch the unit pulls more power just when the volts is low so up goes the current a phizt...

various

switching.

Now there's an idea. From what I've heard some places have disco lighting even when there isn't a storm. A high resistance neutral would make the phase voltages bounce about as the load distribution changed across the 3 phases wouldn't it?

I've suggested that people complain to the DNO about the rate at which they get through filament light bulbs or the fact that kit "just dies", anything from DECT phones to Sky Boxes.

[Andrew Gabriel]

You could easily be losing 5W or more in the cable too. How long are the longest cable runs?

Yes. Can also cause quite a spike if one of the other phases has a short circuit.

[Dave Liquorice]

In theory, and if I've read right for proper 802.3af/at PoE, the quoted powers should be those avialable at the remote device. Over what cable lengths, not sure but would expect that of the ethernet,

305 m?

I shouldn't think that is a major contributary factor, most runs are less than 30 m.

distribution

Right, now to find someone in the DNO who is prepared to listen and knows the derivation and physical of the distribution network around here... Might be easier to drum up local people with disco lighting to complain. I can't as I'm in an area that doesn't suffer...

[tony sayer]

Don't think that Ubiquity or the older Egenous units ever got warm;!..

I don't remember having a duff one apart from one that got walloped by lightning the other week. There wasn't anything left of the CAT 5 cable so what chance did the rest of it have;!..