• We will be performing upgrades on the forums and server over the weekend. The forums may be unavailable multiple times for up to an hour each. Thank you for your patience and understanding as we work to make the forums even better.

POE Lighting

Status
Not open for further replies.

Rock86

Senior Member
Location
new york
Occupation
Electrical Engineer / Electrician
Anyone doing POE lighting systems? I'm just starting to look into them and although the concept is very straight forward, I'm hung up on how you'd be able to handle an open office space worth of lights on a Cat5e cable.

Do they work well? Is the output equivalent?
 

cvillej17

Member
Location
Pittsburgh
Occupation
Technical Advisor
Anyone doing POE lighting systems? I'm just starting to look into them and although the concept is very straight forward, I'm hung up on how you'd be able to handle an open office space worth of lights on a Cat5e cable.

Do they work well? Is the output equivalent?
Rock86, did you find the info you needed? I'm new to the forum but I've been in commercial lighting controls for 10+ years and PoE lighting for more than three of them. I'm hoping I can share some insight and help clear up confusion surrounding PoE lighting.
 

Rock86

Senior Member
Location
new york
Occupation
Electrical Engineer / Electrician
Rock86, did you find the info you needed? I'm new to the forum but I've been in commercial lighting controls for 10+ years and PoE lighting for more than three of them. I'm hoping I can share some insight and help clear up confusion surrounding PoE lighting.
I'm open for additional input, so please feel free. if it is a lot and you want, PM me.
 

cvillej17

Member
Location
Pittsburgh
Occupation
Technical Advisor
Rock86,
Like any control system, there can be a lot to discuss, but to start, check out Platformatics.com. It is one of the lighting controls manufacturers in the PoE lighting space. They have an infographic on their website that might help you understand their system.
Igor, another manufacturer, is the market-share leader with systems all over the US and in 30 countries, globally. They have case studies on their site that show how PoE lighting can be implemented. Their website is igor-tech.com.

As in any PoE system, AC is converted to DC at the network switch. There are no AC drivers PoE fixtures. Instead, the LED arrays inside the fixtures are powered directly by a DC driver (node). In general, each network switchport provides up to 71 watts of power to a node. That node can power up to 4 fixtures, depending on the system and the power requirements of each fixture. Since an LED is a constant-current electronic device, the node is custom-tuned to match the mA draw of the LED array in the fixture(s). Local controls like sensors and wall switches also connect to the nodes and use the network cable for communication to control the system.
For Platformatics, each driver requires a homerun to a network switch. Igor's system allows network-daisy-chained nodes to power multiple fixtures, but there is still a power limit on each homerun.
In both systems, the node's homerun is connected to a network switch, which is connected to a central server. The server is used for system commissioning, monitoring, and occasional maintenance. The system scales like any network; more lights=more nodes=more switches.

For an open office area, multiple nodes would be necessary to power a room's worth of fixtures. HE Williams is a manufacturer that specifically sells PoE-ready lighting. They list nearly 1000 different SKUs, but because all their fixtures are LED, they can convert nearly any fixture they offer to a PoE-ready fixture. To accomplish this, they simply remove the AC driver. HE Williams owns Platformatics, so there's an obvious incentive there to offer PoE fixtures, but they also work with Igor and others. As long as the AC driver can be removed, nearly any LED fixture can be converted to PoE and dozens of manufacturers have done it.

I hope that's a good baseline without being overwhelming, but I can go into more details. Let me know when you'd like more info. As long as it might be beneficial for others here, I'm happy to post about it on this thread.
 
Last edited:

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
In general, each network switchport provides up to 71 watts of power to a node.
So that's up to 71W per cat5 cable? That makes it a Class 3 circuit?

A 15A 120V circuit can provide up to 1800W, which could power a lot of LED lumens, and so control on a finer scale makes sense to me. Going down to 71W per cable seems low, a couple hundred might be a better balance point. But I see that PoE fits better into the existing standards without having to redo everything to get a higher power.

Cheers, Wayne
 

cvillej17

Member
Location
Pittsburgh
Occupation
Technical Advisor
So that's up to 71W per cat5 cable? That makes it a Class 3 circuit?

A 15A 120V circuit can provide up to 1800W, which could power a lot of LED lumens, and so control on a finer scale makes sense to me. Going down to 71W per cable seems low, a couple hundred might be a better balance point. But I see that PoE fits better into the existing standards without having to redo everything to get a higher power.

Cheers, Wayne
The latest IEEE 802.3bt standard (2018) allows up to 99w per network port. Most switch manufacturers advertise a "90w solution", meaning 90w available at the port. Factor in some line loss over distance and a few watts for the node and you're close to 71w at the fixture, itself. More might be available if it's a short cable run, but the standard calls for 71w available at the fixture.
My understanding is that it's class 2 because each port is limited to 57vdc max, but since it's a global standard, PoE lighting follows ISO/IEC 60950. I also believe that more than 100w per port is unlikely for the foreseeable future.

Network switches will be available very soon with 90w per port on 24 ports, so that's 2160w/18A current on a single circuit. That's a lot of light - upwards of 75 2x4 troffers.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
My understanding is that it's class 2 because each port is limited to 57vdc max,
If I'm reading 2017 NEC Chapter 9 Table 11(A) properly, above 30V Class 2 is limited to 5 ma. So it would be Class 3, which allows up to 100 VA.

Network switches will be available very soon with 90w per port on 24 ports, so that's 2160w/18A current on a single circuit. That's a lot of light - upwards of 75 2x4 troffers.
So is the idea that the network switch be installed near the centroid of the lights, rather that in a central location?

Cheers, Wayne
 

cvillej17

Member
Location
Pittsburgh
Occupation
Technical Advisor
If I'm reading 2017 NEC Chapter 9 Table 11(A) properly, above 30V Class 2 is limited to 5 ma. So it would be Class 3, which allows up to 100 VA.
PoE lighting is DC so you'll want to look at Table 11(B)
So is the idea that the network switch be installed near the centroid of the lights, rather that in a central location?
It can be, but most installations I've seen have the network switch stack for all the lights in one area in a data rack in the IT closet. Typically those IT closets also contain networking gear for other systems, as well.
Cisco (and soon others) makes a plenum-rated 8-port switch intended for a "distributed" architecture. Cisco's is called a CDB (Cisco Digital Building) switch and it's designed specifically for this application, though it's only rated at 60w per port at this time.
The thought is one AC run and one non-PoE network run to a switch near the lights could power more fixture with less overall cable, thus lowering installation cost. You could argue that the shorter network cable runs from that switch would allow more power at the fixtures bc of the lower line loss (or at least the power transfer would be more efficient). It hasn't caught on, but it hasn't been abandoned, either. I can't say that it's better or worse, just different.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
PoE lighting is DC so you'll want to look at Table 11(B)
Right, thanks.

So on the topology, what wire size is common from the network switch to the stack, and what length is allowable to ensure 71W at the end from a 90W port? Seems a bit odd to put a 19W heater in series with your light, but that would be the extreme case of maximal length and maximal power draw by the light.

Cheers, Wayne
 

cvillej17

Member
Location
Pittsburgh
Occupation
Technical Advisor
Yes, and likely a bundled group of heaters at that.

Standard networking protocols apply to PoE lighting. - 328ft. from switch to node. 22ga wire is required with an -LP rating of .6 or better.
A node will consume a few watts for computing power, but if you have 15w power loss over 300+ ft, you have to have the right -LP rating on the cable to make sure the cables don't degrade over time in large bundles due to heat.
I know Superior Essex has made a cable specifically for PoE lighting for a while now, and I'm sure others like Belden and CommScope do, too. Generally people point to Cat 6A, but it can be Cat 5e. The category of the cable doesn't matter - PoE lighting systems talk at 10/100 mbps. What matters is 22 ga. and -LP of at least .6.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
So let's see if I got this right:

802.3bt Type 4 allows up to 100W of power at the supply end via using all 4 pairs with a maximum current per pair of 960 ma. I assume that means 480 ma per wire. 22ga copper has a resistance of 0.01614 ohms per foot. So the maximum I2R losses per wire per foot are 3.72 mW, or 29.7 mW per cable per foot. At the maximum length of 328 ft, that's 9.76 W.

Am I off by a factor of 2 somewhere, since earlier you mentioned 90W / 71W?

Cheers, Wayne
 

Hv&Lv

Senior Member
Location
-
Occupation
Engineer/Technician
I’m really dumb regarding PoE lighting, but we’ve discussed it a little (very little) in meetings..
Isn't 90W using all four pairs to transmit that much?
 

cvillej17

Member
Location
Pittsburgh
Occupation
Technical Advisor
So let's see if I got this right:

802.3bt Type 4 allows up to 100W of power at the supply end via using all 4 pairs with a maximum current per pair of 960 ma. I assume that means 480 ma per wire. 22ga copper has a resistance of 0.01614 ohms per foot. So the maximum I2R losses per wire per foot are 3.72 mW, or 29.7 mW per cable per foot. At the maximum length of 328 ft, that's 9.76 W.

Am I off by a factor of 2 somewhere, since earlier you mentioned 90W / 71W?

Cheers, Wayne
I don't think you're missing anything, but 100 watts is the maximum available inside the IEEE standard. 90 watts is what manufacturers typically advertise as available at the port. 71W is the minimum available for the PD that was determined by IEEE for their standard. 71w takes into consideration both power loss over distance and power usage by the driver. As I said earlier, more might be available at the light, but 71 watts is the minimum required for the IEEE standard. What is exactly available at each port could be measured (or calculated by the switch), but I'd guess it's generally 90w.
The point is moot, though, as most fixtures don't require exactly 71 watts or 75 watts. Even if they do, LED arrays inside a light fixture change (even if it's the same light fixture sku and the same production run), which means the required drive current output from the driver has to be adjusted. Lots of times it's more of a trial and error output setting that starts with a drive current that's close. As much as I'd like it to be, it's not an exact science.
Prominent commercial fixtures like 2x4 troffers and 6" downlights are in the 12w-40w range each. When I layout a PoE lighting system, I design for 75 watts available for the light fixtures. If it's a 25w 2x4, I'll plan for 3 per driver. If it's a 20w downlight, I'll plan for 4 per driver knowing that 18.75w on that fixture won't look any different than supplying it with 20w, while keeping in mind that 75w is arbitrary anyway.
If you can drive a fixture (or set of fixtures) at exactly the current limit of every driver on a project, you're maximizing wattage per dollar which is good. It's not typical, though, even if it's the goal.
 

petersonra

Senior Member
Location
Northern illinois
Occupation
engineer
So let's see if I got this right:

802.3bt Type 4 allows up to 100W of power at the supply end via using all 4 pairs with a maximum current per pair of 960 ma. I assume that means 480 ma per wire. 22ga copper has a resistance of 0.01614 ohms per foot. So the maximum I2R losses per wire per foot are 3.72 mW, or 29.7 mW per cable per foot. At the maximum length of 328 ft, that's 9.76 W.

Am I off by a factor of 2 somewhere, since earlier you mentioned 90W / 71W?

Cheers, Wayne
Each pair would seem to be 960 mA going out and coming back.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
Each pair would seem to be 960 mA going out and coming back.
Well, the pairs are labeled DC- or DC+, which would suggest that each pair has no voltage between its members (other than whatever is used for data signaling on 2 out of 4 of the pairs). And if there are 4 wires carrying 960 mA that are ~50V different from the other 4 wires in the 4 pair cable, that would be 200W, not the 100W described in the standard.

So as I understand it each wire only carries 480 mA.

Cheers, Wayne
 

petersonra

Senior Member
Location
Northern illinois
Occupation
engineer
Well, the pairs are labeled DC- or DC+, which would suggest that each pair has no voltage between its members (other than whatever is used for data signaling on 2 out of 4 of the pairs). And if there are 4 wires carrying 960 mA that are ~50V different from the other 4 wires in the 4 pair cable, that would be 200W, not the 100W described in the standard.

So as I understand it each wire only carries 480 mA.

Cheers, Wayne
I misread what you wrote previously.
 
Status
Not open for further replies.
Top