Electrician Talk banner

Power over Ethernet Impact on Electrical Infrastructure/Raceway

4K views 23 replies 11 participants last post by  paulengr 
#1 ·
I am wondering people's thoughts on the effect Power over Ethernet will have on electrical infrastructure, especially in lighting. What conduit/cable products will no longer be needed due to this? I assume some armored cabling will no longer be necessary if buildings switch their ceiling lighting to PoE LEDs. Are there other components (cable, baskets, trays, steel/PVC conduit, etc.) that will no longer be needed? PoE could likely disrupt some of those manufacturers' business. Will be interesting to see how quickly it is adopted.
 
#3 ·
It's all pure Marketing and Sales BS.

Dropping the voltage requires an explosion of copper consumption.

The instant you start pricing out the new scheme -- you give up.

Of course, in time, it will be obvious to even ret ards that putting critical systems onto the Net is INSANE.

Everything will be hacked.

Trucks
Autos
Power

This is straight out of the Terminator SF trilogy.

The idea that America comes out ahead by unemploying hundreds of thousands of teamsters is INSANE.

We'll have VBIEDs roaming the freeways in very short order, too.

Vegas will quickly be forgotten once the carnage begins.
 
#4 ·
I could see it now, 20% of every building will have to be dedicated to patch panel rooms. Instead of stringing a piece of MC to a couple dozen lights, each one needs a Cat6 homerun. Instead of splicing wire, you'll be crimping jacks.

Yeah, that's going to happen.
 
#7 ·
PoE has been around for around 29 years and has not displaced much. You get a maximum of 48 V and mere watts of power on 22-24 gauge wired. It requires about $10 of stuff on both ends to make it work (manufacturing cost) and the Ethernet switch has to be there too going totally unused if it's just a power application. You'd be better off USB which is already in use for these kinds of applications. They have literally squeezed every last watt out of all the #22-24 wires so the wattage will NOT increase.

With PoE limitations it works great for IP phones. I've had at least one PLC input module with thermocouple inputs that ran on it but you can't use it on say supplying power for say 8 4-20 mA inputs.

And even with LEDs, 10 watt lights aren't even "100 watt equivalents". Wiring labor is lower but CAT-5E cable is a lot more expensive than NM-. It has been on the market for decades and won't go lower in price.

The form factor is also all wrong to displace USB applications. RJ-45 connectors are just not designed for constant insertion and removal and the connector is thicker than most phones.

Granted you could make "mini" connectors or just reuse USB ones but you can't alter the size of CAT-5E or the bandwidth is affected and it can't be used for the intended purpose or it wouldn't be CAT-5E, it would be say NM-B.

Finally CAT-5E cannot be used in raceways with standard power conductors by Code. The fundamental issue is that UL sets a lower limit of #18 on power conductors. This is so there is enough ampacity in a fault scenario to last long enough for tripping to avoid a fire. Since signal bandwidth (more specifically capacitance) would be diminished too much with #18 the cable can't be fixed. With standard Ethernet magnetics (1500 V isolation via isolation transformers with heavy notching from D.C. to 60 Hz) technically UL's issue is meaningless as communication cable but it is very relevant as PoE where power is carried via D.C. systems, rendering the isolation specification moot.

Disagree with hacker scenarios. PoE by itself is just a way to distribute power to low power devices with the convenience of avoiding "brick" power supplies. You can buy "midspan" power supplies that are literally just dumb devices that inject DC onto the cable and pass high frequency AC through from one port to the other for retrofits. There is little to no "smart" or IoT anything going on here. That is all done in the Ethernet side.

So you might be able to power a very expensive nest like thermostat with either a converter to standard 24 VAC HVAC signaling or new control boards. Might be able to power most of the 2 wire 4-20 mA devices we have today and basically displace HART with something simpler and cheaper but it requires new infrastructure which is HARTs original selling point, and it comes with a new 100 meter distance limit so won't work in most petrochemical plants which is the target market. Works good on IP phones and maybe (again overpriced) desk lamps, Mot useful for overhead fixtures, running any household appliances, any entertainment equipment such as even a DVR, TV, TiVO, stereo, etc.

So nice try but the whole concept is vastly impractical for 99% of the applications. This is kind of like the whole advertising for WIFI across a manufacturing plant. Sounds great until you get past the network to the devices. Battery power is not that reliable. Got a dead TV remote?


Sent from my iPhone using Tapatalk
 
#10 ·
Paul ... EVERYTHING gets hacked --- if only to hold the user hostage.

Look what's happened to PCs.

They're being hacked merely to fake out advertisers. ( Kovter )

Today's designs will not hold steady.

They will become ever more 'netted' -- and hackable.

The whole idea of putting ones 'everything' out and accessible to the Web will ultimately collapse.

It's like leaving your wallet on the front door step.

For 2nd World and 3rd World hackers, such schemes are irresistible shakedown targets.
 
#13 ·
I rant about this all the time, right now AC to every LED light doesn't make sense but PoE isn't the answer.

IMO PoE won't be an efficient way to power and control LED lighting, the total cost of ownership over time won't be what it could be.

On the other hand putting low voltage power supplies at every light doesn't make sense to me either. The place where there's room for improvement in LED lighting is the DRIVER.

The best efficiency would be to make a bigger, smarter, more durable, more serviceable driver. The wiring from the driver would be similar to speaker wiring, and the driver would have multiple outputs to drive groups of lights that go on / off / dim together. (No more ridiculous proprietary cables from drivers to lights...)

Picture a typical commercial lighting setup where you pipe to boxes then have a bunch of MC coming off to feed the lights. Replace the box with a smart driver and replace all that MC with speaker wiring. If you have a good place for it, you'll probably mount that box in a closet where it's easy to service rather than in the ceiling, but the topology remains the same.

* Cost of wiring obviously goes down - AC power and ethernet to driver, speaker wiring from driver to lights.
* Lights are now dumb LED holders and will last forever.
* Total cost of drivers will be lower - cheaper to manufacture fewer / bigger than more / smaller.
 
#16 ·
LED lighting is somewhat unique in the sense that at least when we get up to higher power LED's just like neon, we need current control rather than voltage control. The voltage drop across a green or blue LED and hence white (which is just phosphor coated blue ones) is around 6 V. Up to a point you can run lots of them in series or parallel until the small differences in devices start to add up.

So ultimately the best distribution system would be to go low voltage DC if the only thing we look at is the LED's themselves. But this doesn't work in practice for two reasons. First problem is that thermal loads are proportional to I^2*R, so we can push more power and get down to your proverbial "speaker wire" only if we accept the idea that we're going to jack the voltages up since this allows us to transmit more power with smaller wires.

Second issue is that going with DC also creates its own problems. Chief among them is that now voltage drop becomes a bigger factor...there is a good reason for jacking up voltages into the kilovolt range for lower losses over longer distances. The same is also true with increased frequency although as we do so the "skin" effect becomes a bigger issue...transformers for instance become very small but wire size has to increase since most of the signal travels only along the surface. Eventually we get to the point of using waveguides (tubes instead of cables). VGA video signals for instance in some versions actually run on coaxial cables which explains why the darned cables were so very expensive. But the smaller transformer aspect is the reason that aircraft often run 400 Hz systems because weight and space is a critical thing on an airplane.

Summing up if we are going to create large converters to drive large amounts of LED lighting, the ideal application would be to run higher voltage but still AC. The LED's are naturally diodes although obviously we take an unacceptable efficiency hit if we don't convert it back to DC so at the end we really didn't get rid of the "power supply at the fixture" design issue after all. Best we can possibly do is shrink it by going to higher frequencies even though this also jacks up the conductor size (which goes back to jacking up the voltage again to reduce conductor size).

That's never mind the fact that there still needs to be a standard to ever get any amount of market penetration. There are lots of clip-on designs on the market to reduce the labor of doing terminations at each fixture but so far none have gotten to the point of becoming an industry standard other than the NEMA plug and receptacle standards, and those aren't for permanent wiring.
 
#14 ·
This is one of the few markets/ applications that will continue to evolve. The problem will be the immediate obsolescence of any proprietary system installed today.

Probably, at present available technology, the best compromise could be a system with a central processor that would be connected to multiple drivers. Each driver could have multiple addressable outputs to individual lighting units. From the drivers, route say a 2 conductor #18 to each light fixture. This arraignment would be similar to a PLC with remote I/O and COULD use standard nonproprietary components.

Above the ceiling could be a real spider web of small cables, even worse than today.
 
#20 ·
This is one of the few markets/ applications that will continue to evolve. The problem will be the immediate obsolescence of any proprietary system installed today.

Probably, at present available technology, the best compromise could be a system with a central processor that would be connected to multiple drivers. Each driver could have multiple addressable outputs to individual lighting units. From the drivers, route say a 2 conductor #18 to each light fixture. This arraignment would be similar to a PLC with remote I/O and COULD use standard nonproprietary components.

Above the ceiling could be a real spider web of small cables, even worse than today.
Was actually just looking at the new DC panels that are available for residential application. Controls everything, potlight, under cabinet lighting, USB receptacles, garden lights, etc.

Believe the company was called Magic Lite or something like that, I will have to get the spec sheet to confirm. ESA is not sure how to handle them just yet, because anything that comes out of it is not covered by the CEC / OESC except for the "bundling requirements".

Looks like the beginning of less 14 (12)/2 and more #18/2.

Cheers

John
 
#17 ·
Sort of along the lines of this, I ran across about the neatest driver idea I have seen yet, it was magnetic, so it just sticks to the fixture you are changing the lamps from fluorescent to led with. And they come with built in Wago terminal blocks so the existing wires out to the tombstones you just cut near the old ballast and plug them into the right colored hole in the driver. So easy pretty soon there won't be any electricians changing them , the janitor will be doing the work cause he is paid a whole lot less, and you would have to be a total idiot to get shocked installing one of these color by numbers led drivers.
 
  • Like
Reactions: readydave8
#18 · (Edited)
Our office lights are PoE (Philips or Lithona I believe). There is a network switch in the back office that runs all of our 2x2 and 1x4 led ceiling fixtures. Not a single 14/2 needed for the whole lighting system.

Oh...and just because it's connected to an internal network doesn't mean it's connected to the internet. No hacking worry there.
 
#24 ·
That’s nice. Here’s an example fixture.


https://poelightingusa.com/product/il-girro-metalo-ceiling-mounted/

10 Watts, boasting 100 lumens per Watt. So this is 1,000 lumens. That’s about the same as a 60 watt incandescent. So far, so good. But I looked everywhere and can’t find prices which is usually a bad sign. As I said just add $10 to an LED fixture.

Here is about the cheapest PoE switch I could find (cost per port), at $8.46 per port.

NETGEAR 26-Port Gigabit Ethernet Smart Managed Pro PoE Switch (GS324TP) - with 24 x PoE+ @ 190W, 2 x 1G SFP, Desktop/Rackmount, S350 series https://www.amazon.com/dp/B07PHVBQVS/ref=cm_sw_r_cp_api_i_nNQ7EbP8J4VE5

So PoE adds $20.

As far as addressable lighting lots of strip lighting has been out there. Just go to any baseball or hockey game. They have full color LED graphics on strips along the various tiers around the stadium. The way this works is there is a 3 or 4 wire connection at one end that supplies power such as 12 or 24 V. The other one or two wires injects a serial signal that consists of 3 bytes of an RGB signal and a tiny amount of overhead. Each “pixel” grabs the RGB bytes then passes everything else through onto the next “pixel”. In this way an entire string of pixels is individually addressable and powered from a simple 3 or 4 wire connection. The chip at each pixel costs about a dime. The strip can be mass produced or pixels can be wired like Christmas light strings. There are simple, cheap power supply/drivers that use Ethernet for communication and include a 2 port switch (for daisy chaining) that run around $30-60. These are used for large synchronized holiday light displays and sports arenas.

This is what this PoE lighting is competing against. Except that LED RGB strips are a real product available NOW. And the prices are a few dollars per pixel. This is because the overhead of Ethernet is gone. And cheap low cost wiring is used except at the “controllers”.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top