What Is PoE? A Look at a Pair of Switches
$56.88
So today we’ll talk about what PoE is and “how it works.”
PoE, or Power over Ethernet, is a very old technology. I first heard about it about 15 years ago, around the time all sorts of neighborhood networks like “Pioneer-Net” started appearing here. Since I work with all kinds of power supplies, I got interested in it too.
This technology lets you power relatively low-power devices directly over a network cable. These include all sorts of devices: network cameras, IP phones, weather stations, switches, and even low-power computers.
Incidentally, this is exactly the kind of setup that lets you easily get past the 100 m limit for copper cable: you put an additional switch halfway along the line, power it from the cable, and use it to relay data for the next 100 m.

There are different types of local network cable, but two- and four-pair cables are the most common. With a four-pair cable used for a 100 Mbps network, two pairs are left unused; they were originally intended for telephone service (the blue pair, for example).
Under the standard, the blue pair carries positive and the brown pair negative. Personally, I find that a little illogical: it would make more sense to put positive on the brown pair and negative on the blue pair, by analogy with live and neutral in mains wiring. But apparently the designers had a different principle in mind.

The simplest way to send power over twisted-pair cable is so-called Passive PoE, where the cable conductors simply extend the power supply’s output wires. Special kits are used for this.

The external setup looks like this:
The adapters plug in between the cable and the device, on both the Ethernet and power lines, and are essentially just extension cables.
But there’s a problem: the limited range. The diagram on the left says the maximum length is 30 meters, but it depends on several things:
1. Supply voltage.
2. Cable quality.
3. The receiving device’s input voltage headroom.
So with a 5 V supply and a fairly power-hungry device, the range can be significantly shorter. Also, this setup provides no galvanic isolation.

Since you can’t reduce the resistance of the cable conductors, the way to extend the range is to increase the voltage and install a step-down converter at the far end. I’ve built systems like this myself many times.
The maximum supply voltage in these systems is around 48–55 V. They come in several varieties:
1. Single-port or multi-port.
2. Fully compliant with the PoE standard or operating in Passive PoE mode.
3. 48 V or 24/12 V.
Some “alternative geniuses” have even tried simply putting 220 V on the cable. I’ve come across “solutions” like that myself. At the very least, these people deserve to have their arms shortened, preferably up to the shoulders, because this isn’t just wrong—it’s extremely dangerous:
1. Risk of electric shock.
2. An equally real risk of breaking your neck, since you could cut a live cable while standing on a stepladder that’s itself on some stairway in an apartment building.
3. Risk of starting a fire.

There are two main ways to supply power over the cable, A and B. The difference is simple: in scheme A, the same conductors carry both power and data (surprise, surprise); in scheme B, power goes over the blue and brown pairs, which aren’t used for a 100 Mbps connection.
Scheme B is used most often because it’s a little simpler. Scheme A has the advantage that it works with two-pair cable, which is slightly cheaper.
Scheme B also lets you install an intermediate injector, for example halfway along the cable, although there’s nothing stopping you from putting it next to the switch instead.
I used to build systems using scheme B too.

There’s a table on Wikipedia showing the pin assignments and their functions for the different wiring schemes.
Below that, it also mentions:
The IEEE plans to form a group to develop a new PoE standard that will use all four pairs of Ethernet cable and at least double the power limit for powering devices

In fact, a power delivery scheme that uses all the wire pairs has been used successfully for quite some time, and lets you transmit twice as much power over the same distance. Even with a gigabit connection!
That was exactly the approach I wanted to use in my devices, but I got stuck on the need for special center-tapped transformers. I also had an idea to isolate the power the way it's done in TV systems, where both power and signal travel over a coaxial cable. But then I simply lost interest and abandoned the whole thing.

There is also a corresponding standard, ieee802.3at, also known as PoE+.
Below is a screenshot from the document showing the maximum power specified by the PoE and PoE+ standards. Since the cable resistance and voltage stay the same, the only way to get more power is to transmit it over all four wire pairs at once.
The power classes are also shown below. In other words, a compliant PoE device first tells the power source what it needs, and if the source can provide it (as configured in the source's settings), power is then supplied at the output.
For obvious reasons, the receiving device must also support the PoE+ standard.
There is a very good, easy-to-understand description of the PoE+ standard in Russian at this link.

Everything I've described above essentially defines the transmission medium: what we're transmitting, how, and over what. In this case, of course, we're talking about power, not data.
But PoE isn't simply a matter of feeding power into a cable and having everything work. It's a bit more complicated, and I'll explain why below. For now, let's move on to the practical part of the review: a switch with PoE functionality.
People often ask me where it's best to buy cameras on AliExpress. This review features a product from a store where I buy cameras regularly, and I've had only one problem in all that time: they sent me one camera instead of two (I order them in pairs so as not to attract the customs authorities' attention). I messaged the seller about the problem before I even opened the package. They asked for the order number, then replied an hour later: yes, they could see what happened, and asked whether I'd prefer a refund or for them to send the missing item.
The second time, it wasn't really a problem so much as a minor hiccup. I paid for the order, but the seller kept not shipping it. Eventually, the order was automatically closed before it shipped, and the money was automatically refunded.
I wrote to ask what was going on. They replied that there had been a delay with the item in the warehouse (it happens), and that they could ship it in two days.
I said, okay, but I'd bought it with a coupon and got a $4 discount, and had also lost almost a dollar on the money transfer. Could they compensate me for the loss? They said, no problem: place the order but don't pay, and they'll make up the difference. And sure enough, 10 minutes after I placed the order, they lowered the price for me.
So, I got the switch. It was packed pretty well, even though one of the order reviews complained about poor packaging. As far as I'm concerned, it was all fine. The switch itself was in a cardboard box, with another outer box that fit it exactly. This seller consistently chooses outer boxes that keep the product packaging from rattling around inside.

The package includes:
1. Switch
2. Power cable
3. Mounting hardware
4. Manual and warranty card.

1. The manual is entirely in Chinese. It describes the connections and lists all the switch models the manufacturer makes, along with their model codes.
2. The power cable is a standard one, familiar to anyone who works with computers. The quality is average and the cable is thin, but the device's power consumption is low, so it'll do.
3, 4. They also included five screws and a pair of mounting brackets for attaching the switch to a rack. You only need four screws; the fifth is a spare.

The specifications list support for (ieee802.3af) max. 15.4 W / (ieee802.3at) max. 30 W per port, but the maximum power consumption is 150 W. As I understand it, that's the maximum power drawn from the mains, not the total output power.

The switch itself looks pretty standard: a rectangular metal box with a bunch of connectors.
The front panel is 220 mm wide, which, with the included brackets, lets you install it in a 10-inch rack.


The port configuration is 8+2. In this case, that means the switch has 8 PoE ports and 2 standard ports.
There are different configurations, for example, 4 PoE + 1 standard port, 8 PoE + 2 gigabit ports, 24 PoE + 24 standard ports + 2 gigabit ports, and so on.


For example, one of the switches we use at work to power cameras is a Cisco SF200-48P. It has 24 PoE ports, 24 standard 100 Mbps ports, and a pair of gigabit ports.

In this case, though, it's a managed switch, and one of its handy features is remote control of power to the ports, in case a camera freezes (which does happen from time to time).
After one of the summer thunderstorms, I was left with just one of these two switches; the other was completely dead. I'll have to take it home and poke around inside. I replaced it with a 24-port TP-Link.
I think everyone understands that these devices cost a bit more than the one under review. When I bought it, it cost around $600 (if I remember correctly), or $12 per port. The one under review costs $57, or $5.70 per port.

As with “full-size” switches, some of the ports here are set aside for connecting devices—in this case, eight.

The other two ports can't supply power and are usually used to daisy-chain several switches.

There are several different ways to connect them.
1. Distributed (top). The switches are installed where the devices are concentrated, and each is connected to the server and to the others with a single cable.
2. Star. Here, all the cables run to one point, where they connect to a central switch or switches, which in turn connect to the server.
The advantage of the first option is that it uses less cable, reducing the overall project cost, and lets devices operate farther from the server.
The advantage of the second option is greater reliability.
Cascading switches is convenient when you need to cover a large area, since cable runs have a limited length. But it also creates a problem: if one switch loses power (or a cable is broken), everything goes down. With a star topology, only one device goes offline, and it's very easy to diagnose.
Also, you can't keep adding switches indefinitely, because the main link has a bandwidth limit. For example, this switch can connect 8 cameras, while its “uplink” ports run at 100 Mbps. I usually set cameras to a stream bitrate of 3–4 Mbps, so 8 cameras need up to 32 Mbps; two switches in a row, up to 64 Mbps; three... no, three won't work reliably, because you need some headroom. So you can daisy-chain two of the switches under review, or theoretically three, with limitations.
If you have more devices, choose switches with gigabit uplinks (that's what the extra ports are called), and you can expand the system considerably.
Why do I prefer a star topology? It's not very common to exceed the 100–150 m cable-length limit within a single site. So usually you first choose a server room as close as possible to the center of the star, giving you a coverage diameter of 200–300 m, which is pretty good. You also get centralized backup power (and in my case, power from two feeds with automatic transfer switching), which makes the system more reliable.

Somewhere along the way, after everything I've written above, I forgot to mention that this solution is for people who don't use recorders with built-in PoE. If you buy a recorder like that, it can power the cameras itself, but there are downsides:
1. In a worst-case scenario like a lightning strike, there's a good chance you'll lose the recorder too.
2. This solution is somewhat less reliable, since the box contains not only electronics and heat-generating hard drives, but also a fairly powerful power supply.
Here's a photo of a Hikvision recorder with 8 PoE ports, for example.

These photos were actually taken last winter for another review, but I don't think I'll write it. I can only say that, in the end, I ran into such a nasty bug with a couple of Hikvision recorders that I doubt I'll look at them favorably in the future.

The power cable connector is on the back, while almost the entire left side is taken up by ventilation slots.

But on the right, in addition to the ventilation holes, there's also a small switch that ordinary switches don't have. This switch provides a feature that’s more useful for video surveillance systems and similar setups with relatively low traffic but a need for long cable runs.
The switch forces its PoE ports to operate at 10 Mbps, which allows it to reach farther at the cost of lower speed.

There’s just a sticker on the bottom. Unfortunately, this switch has no feet, which is rather inconvenient when placing it on a desk.

Those who have read my reviews before have probably guessed that I’m about to take it apart. No, I will take it apart, just not yet :)
Next, I’ll show you another switch that I also use. I bought it from a local store, and it costs around $40–50.
It also comes in a cardboard box, with an egg-carton-style insert inside.

Since this switch has an external power supply, the package contents look much more substantial, although in practice they’re almost the same.
1. Switch
2. Power supply
3. Power cable
4. Mounting hardware
5. Feet
6. Manual

1. The power cable is a little better here; the power supply is rated at 54 V, 1.11 A (60 W).
2. A pair of brackets, five screws, and rubber feet! I really missed that last item on the previous switch.

The switch is sold under the UTEPO brand.

You may be wondering why I wrote “under the brand” above, rather than “made by.” It’s quite simple. While preparing this review, I was looking for some information and happened to come across this switch. Don’t you notice anything familiar?
By the way, I’ll come back to FoxGate switches later.

This one also has a CCTVDefault switch; the only other thing on the front panel is the power indicator.

On the side are ventilation holes and a Kensington lock slot.

This switch has six ports arranged in a 4+2 configuration: four PoE ports, and two Uplink ports for connecting to other switches or to a computer/recorder.
The power supply jack is also located here.

The two switches side by side.

And now for the teardown. Since it’s not for the faint of heart, I’ve put it under a spoiler.
A side-by-side photo of the two switches. Even at this stage, I can say that the larger switch is more "delicate" and better suited to indoor systems, while the smaller one has better protection.

Now for a few tests.
First, a bandwidth test.
The speed without a switch was 75 Mbps; through the Uplink ports, it was 65–67 Mbps. For some reason, even with a direct connection, I couldn't get above 80 Mbps, although when testing Android devices I saw speeds of 90–95.

Now one computer is connected to the Uplink port and the other to a PoE port.
In normal mode: 61–64 Mbps
In CCTV mode: 9.5 Mbps, with the speed throttled very consistently.

Testing the second switch gave exactly the same results—I’d say they’re within the margin of error of the test itself.

Next, a load test. I decided it would be best to make the test as easy to see as possible, without any complicated equipment, testers, electronic loads, and so on.
The test is very simple: take 8 IP cameras, connect them to the switch, and leave them running for a few hours.

The switch consumed 3.1–3.2 W at idle.
With the cameras connected, consumption was 34 W, with the IR illumination running at about 50% power. Unfortunately, I couldn't get it to full brightness, but in complete darkness it was around 40–42 W. Since there were so many cameras, though, they started illuminating each other.

In the end, I realized this wasn't really a test at all: the hottest spot (45°C) was the cluster of cameras, while the switch looked downright cool by comparison. Its case temperature was 37°C, and this was in summer.

The problem was that although the test above is very illustrative, it's also flawed, because power consumption depends on cable length.
The explanation is simple. Say a camera consumes 6 W. With a short cable, the switch will supply the same amount. But as the cable gets longer, its resistance adds to the circuit. Since the camera's power consumption in watts stays constant, the voltage drop at its input makes it try to compensate for the losses by drawing more current. Since the current in the circuit is the same throughout, the result is that with the switch's output voltage unchanged and the current increased, the power drawn from the switch also increases.
So I decided to compare the power consumption of a camera connected with a short cable and with a cable about 20 m long. But the test showed no change in consumption—the wattmeter simply doesn't have enough resolution.

All right, I thought, let's make the test more complicated. Above, I mentioned that this switch supports the IEEE 802.3at standard, which means power is transmitted over all pairs, halving the cable resistance.
So I took another cable, about 15–20 m long, but crimped it so that only two pairs were connected. I also checked that the switch could power a camera using just the two signal pairs.
Unfortunately, this test also came to nothing: consumption stayed at the same level. Well, it did increase, but the difference is too small to measure with basic instruments.

Ideally, the next test would use a long cable. I do have three full reels at work, but I won't run this test: 305 m is still a lot, and cutting up a new cable would be rather expensive. When I have shorter pieces, I'll definitely give it a try.
That's why I moved on to testing with an electronic load; I had no other options left.
1, 2. I soldered on a pair of wires, routed them outside, and connected them to the electronic load.
3, 4. I set the load current to 2.2 A. Since the power supply's output voltage is 51.6 V, that gave me about 114 W, or 14 W per port, compared with the 13 W PoE rating. The power drawn from the mains was about 130 W.
5, 6. I ran it like that for about an hour, then reduced the load to 80 W, or 10 W per port—plenty for surveillance cameras.

Now let's measure the temperature.
1. After half an hour at 114 W
2. After another half hour at 114 W.
3. After an hour of operation at 80 W. I didn't let it cool down first; I simply reduced the power and started another hour-long run.
I can say that 114 W is the limit for this power supply. At 104 W (13 W per port), things would of course be a little better, but in practice it's best not to exceed 10 W per port, which is more than enough to power surveillance cameras.

And then, for some reason, I wanted to make sure the switch really does comply with the IEEE 802.3at standard. For this, I once again removed the switch board itself and started examining it closely.
It turned out that things weren't quite as straightforward as they seemed. The photo below shows the section of the board with the traces for four of the eight ports. You can follow the traces from components D21 and D79–81 to the RJ45 jacks.

But I was interested in the circuitry right by the jacks themselves.

To make it easier to understand, I highlighted one of the connectors in color, matching the colors of the Ethernet cable wires plugged into it.
Logically, the brown and blue pairs should also carry power, but you can see the jumpers below (components marked 0), and not all of them are populated. The jumpers to the contacts marked blue and brown are missing. A measurement confirmed that there is no voltage on these contacts while the camera is operating.
This design is acceptable and normal, but it isn't IEEE 802.3at. So, even if switching to 10 Mbps helps increase the range, I still wouldn't go beyond 100–150 meters for power delivery. That said, it depends heavily on the load's power consumption.

I was even more surprised when I decided to check the second switch, which proudly says PoE+ on the front panel.
It turns out there are no traces leading to the jacks from the bottom at all.

From the top, you can clearly see that power is supplied over the data pairs (1,2 and 3,6). Just to be sure, I connected a camera and checked the power: nothing, only on the contacts mentioned above.

In my review, I mentioned FoxGate switches and that the 4-port switch I showed was the same model, just under a different name.
I came across an interesting switch from the same company, FoxGate or Utepo (I don't know who copies whom), which had been tested.
It has an interesting feature: buttons for briefly cutting power to individual ports. In my opinion, though, that's overkill and marketing. Where losing the connection when the whole switch is powered down would be critical, people usually install managed switches. Where it isn't critical, it's easier to unplug and reconnect the cable or power-cycle the whole switch.

Anyway, the guys got these results: Cu means copper, CCA means copper-clad aluminum.

Now I'll try to put all the information together and answer some of the questions people ask me and that I see in discussions. I can't guarantee 100% accuracy, but I'll explain it as I understand it.
1. PoE really is convenient: you can power equipment over the same cable that carries the data.
2. There are two types of PoE:
a. “Standard” PoE, which uses a dedicated controller and a high voltage of up to 57 V.
b. Passive PoE, with no controller—power is simply fed into the cable.
3. “Standard” PoE supplies a low voltage to the cable with current limiting when there is no load. You can cut or crimp a cable connected to the switch without damaging anything. Nothing will burn out even if you mix up all the wires in the connector—I tested it myself, though of course it's better not to take chances. When compatible equipment with the same “standard” PoE is connected, power is supplied automatically.
4. Equipment designed for “standard” PoE will also work from a regular power supply rated at 44–57 V if you apply power to the blue (positive) and brown (negative) pairs. That said, PoE equipment should have a bridge rectifier inside, so polarity shouldn't matter much.
5. Equipment designed for passive PoE most likely won't work when connected to a PoE switch.
6. “Standard” PoE comes in three types:
a. Power over the same pairs that carry data.
b. Power over the brown and blue pairs.
c. Power over all four pairs. In this case, it's the PoE+ standard. Gigabit equipment with PoE works the same way.
7. Remember that, unlike “standard” PoE, passive PoE can also use 12 V or 24 V, so check first or you could damage your equipment. The voltage for “standard” PoE is strictly specified, so this can't happen.
8. Besides PoE switches, there's plenty of other equipment, such as external injectors with full PoE support, video recorders, and so on.
9. How can you tell whether equipment supports “standard” PoE? At a minimum, it should be labeled PoE; ideally, it should list the standard numbers: IEEE 802.3, IEEE 802.3u, IEEE 802.3x, IEEE 802.3af/AT.
10. When installing cameras outdoors, regardless of the type of PoE, protect the connector from water. Otherwise, it will fail very quickly, and that will be a real pain.
What can I say about the switch reviewed here?
I have no major complaints about its operation. It works reliably and is perfectly suitable for simple setups, though with a few caveats.
I don't recommend drawing more than 10 W per port; it will overheat. What I found disappointing is that it can't power a load over all four pairs. On the plus side, it can supply power over the data pairs, so if needed, a single cable can carry signals from two devices at once.
There's a switch that reduces the speed on the PoE ports to 10 Mbps, allowing you to run longer cables to the equipment.
One safety concern: there's no insulating shield over the power supply.
That's all from me. I hope you found the review useful. As usual, I'm happy to answer questions or hear your comments on the topic.
$56.88
This page was found when searching for:
камеры poe для видеонаблюдения распиновка, схема подключения витой пары по poe, 7, питание по лан кабелю, 2, почему коммутаторы роемежду собой соединяются двумя dac двумя кабелями, пое коммутатор на 4 порта для видеонаблюдения еак поключить к пк, на коммутаторе для чего переключатель s-e w, мощный коммутатор от которого будут питатьсяя другие коммутаторы, отдельный poe свитч, КоммутаторFS1024 максимальная скорость на портах uplink в режиме«Extend», 6, для чего в свитч коммутаторе питание, свитч для интернета с поддержкой poe описание, подключение порта с poe к порту без poe, 7-port Combo Hub (4-port USB2.0 3-port IEEE 1394) Б.п., 1 poe подключить два пк, подключение коммутатора через poe порт, zte zxp102028e выход poe зачем, poe коммутатор с переключателем wlan что это, соединение свитчей с пое схема, вход poe in на коммутаторах, uplink порты на poe свитчах, wan поддерживает poe, что такое uplink и poe, PoE коммутаторы, коммутатор Switch PoE, оборудование для видеонаблюдения, сетевой коммутатор, коммутатор PoE портов
камеры poe для видеонаблюдения распиновка, схема подключения витой пары по poe, 7, питание по лан кабелю, 2, почему коммутаторы роемежду собой соединяются двумя dac двумя кабелями, пое коммутатор на 4 порта для видеонаблюдения еак поключить к пк, на коммутаторе для чего переключатель s-e w, мощный коммутатор от которого будут питатьсяя другие коммутаторы, отдельный poe свитч, КоммутаторFS1024 максимальная скорость на портах uplink в режиме«Extend», 6, для чего в свитч коммутаторе питание, свитч для интернета с поддержкой poe описание, подключение порта с poe к порту без poe, 7-port Combo Hub (4-port USB2.0 3-port IEEE 1394) Б.п., 1 poe подключить два пк, подключение коммутатора через poe порт, zte zxp102028e выход poe зачем, poe коммутатор с переключателем wlan что это, соединение свитчей с пое схема, вход poe in на коммутаторах, uplink порты на poe свитчах, wan поддерживает poe, что такое uplink и poe, PoE коммутаторы, коммутатор Switch PoE, оборудование для видеонаблюдения, сетевой коммутатор, коммутатор PoE портов
Questions and answers
What is PoE, and how does it power a device?
PoE sends power over an Ethernet cable. The article describes passive PoE and standard PoE, where the device tells the power source how much energy it needs.
How much power does the author recommend drawing from a PoE switch port?
The author advises against drawing more than 10 W per port because the switch overheats.
What drawbacks did the author find in the PoE switch?
It cannot power a load over all four pairs, and there is no insulator above the power supply. However, it runs stably and is suitable for simple setups.
The figures from this review are in the measurement database: Other devices
Discounted products
You may be interested in
Discounted products
Support the project
All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!
Comments2
Discuss on the forumHave a question or some experience to share? Discuss the article with the author and readers: the forum is a convenient place to share schematics and photos.
Go to the forumOnly registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.
Log inRegister





























Comments are published in their original language.