This guide provides comprehensive explanations of the elements in selecting and using Power Over Ethernet with IP cameras.
Inside this report we cover:
PoE vs Low Voltage
When to Use PoE, When Not
PSEs vs PDs
802.3af vs 802.3at vs 802.3bt
Nonstandard PoE Implementations
Power Consumption vs Specification
Calculating Power Budget
PoE via Switch, MidSpan or NVR
The Top 5 PoE Misunderstanding
PoE vs Low Voltage
All cameras need electrical power to operate.
'Power over Ethernet' (PoE) uses a single cable to connect a camera to both the data network and a power supply. In most cases, powering cameras before the advent of PoE meant using low voltage power using separate power supplies and dedicated power wiring. PoE eliminates the second cable / supply.
I have found that a lot of devices which are PoE will also have no issues with a 12v supply connected at the same time as PoE. (only from own checking of devices)
Whilst it's not really needed, and why should it be, perhaps there is some value in having maybe a critical device have a fallback supply if the PoE is accidentally turned off. or some nugget unplugs a device by mistake from the Switch.
At least you won't need to wait for the device to come back online with the PoE re-instated.
Startup time for old IP cameras use to be quite long, this has improved with newer solutions, but it does seem to take like more time when you are waiting for it to appear on the NVR display!
I seem to remember earlier Arecont cameras, circa 2010, had a dip switch to set for either POE or 12V power, and they warned against having 12V powered cameras connected to POE, but my memory may be fuzzy.
UI3 - The first two are standards, but the last is yet to be ratified. High PoE and PoE+ have been used to describe anything over 15.4 for years. A more specific name for 60w PoE is PoE++ (or UPoE for Cisco gear)
Yes, have just started to notice the PoE++ designation for Hi POE. My experience is in the Axis world which was always POE+ for 30w and Hi Poe for 60w. Going by memory which is getting more and more suspect. Good information to know...
Can a cable plugged into a port, but not a camera electrocute me or be a safety hazard?"
Answer: No. Due to the initial negotiation process, PoE power is not actively issued unless a connected device requests it. This means that a cable connected to a PSE is not 'electrified' at all until plugged in to a PoE device and will not present a safety danger because of incidental contact.
In the most common of implementations, this “negotiation” consists of nothing more than the PSE sending out a short pulse and measuring the load for a signature (passive) resistance of the device.
Pro Tip: Just keep your body’s resistance over 25 Ohms ;)
In some cases, powered devices may handle passive PoE without issue. However, those which are not specified to use may be damaged. Because of this, we do not recommend using passive PoE unless the device explicitly specifies it.
Passive PoE is most common in wireless equipment, such as Ubiquiti or Mikrotik, but not common in IP cameras.
Also, passive POE @ ~48V can be found on many (older) midspan injectors. Since they provide an 802.3af level voltage, (sans negotiation), they can be used without fear of damaging actual 802.3.af gear.
Paige Game Changer cable (22 gauge instead of 23 Cat6) is supposed to send data up to 200m... does that mean that PoE running on the same data cable would travel the same distance due to the larger gauge? We are looking in to using them for an upcoming video surveillance contract and want to make sure we’re specing this in properly without having to using injectors, creating another potential point of failure.
My 2 cents, ask the cable manufacturer. I doubt they are following an official certification spec because I never heard of one that goes beyond 100 meters for 8 conductor twisted pair Ethernet, so I would put it on them to say if it would work or not. After all, they are the ones who has to stand behind it. I think it'd be pure speculation on anyone's part here unless they have used the product in the past. Even if it they have, I think (unless I'm wrong) it's still a not based on an industry spec.
NOTE: I'm not an electrician but I play one on the internet.
Whenever you increase the size of your conductor you decrease resistance, so a larger gauge cable will carry current further and have less voltage drop. However you're only going up one gauge size here. Every three gauges your wire doubles or halves in cross sectional area (i.e. 20 ga is twice as thick as 23 ga).
23 ga wire has a resistance of 20.3074 Ω/1000 ft while 22 ga wire has a resistance of 16.1045 Ω/1000 ft. You're definitely gaining something with the extra gauge size but you'd need to work out the power draw of the device and the voltage drop of the cable over that distance. Google it, calculators for this sort of thing abound.
Pro tip: the +3 gauge size also works when simply twisting conductors together. If you have 24 ga CAT5 and you combine two conductors together, you now have the electrical equivalent of a 21 ga wire. Put four conductors together and it's the equivalent of an 18 ga wire. Twist all four pairs into one giant conductor and it is equivalent to a 15 ga wire. In a pinch, you can use various wires this way to effectively carry more current than they were intended for at the expensive of conductor count. If you have the ability to run the proper spec cable however, obviously do so.
NOTE: I'm not a comedian but I play one on the internet.
I am teaching my newly GF electronics-101 and she can attest that the increased size of a 1 conductor does not decrease the resistance, she does however agree that the larger gauge carries more current without the voltage drop.
that's a lot of writing about the gauges and ohms. The formula i fairly simple, we learnt it in grade 9 if I'm not mistaken: 1/R = 1/R1 + 1/R2 + .... + 1/RX, where RX are resistances of your conductors and R is a result for twisted cable.
So, if you have two 23ga 1000ft wires (you said it's 20 Ohm and I believe you), your twisted wire would be 1/R=1/20+1/20. R=10 Ohm. If you use four of those, your R=5 Ohm.
Just measure the single-wire resistance, and calculate how many cables you need to get the acceptable total
CAT6A from certain manufacturers (Belden being one) tends to have 22 gauge conductors. We're having one hell of a time finding male RJ45 connectors to fit. Yes, I know many say to terminate the camera end with a jack and use a patch cable to the camera but that is just not possible in many situations. In particular, we're installing some cameras outdoors on 4x4 metal fence posts. There is absolutely no way to run a terminated cable either through the post or through the 3/4" conduit between the pull boxes and the posts.
Who makes connectors that can accommodate heavier gauge wires, especially the direct burial CAT6A used on this job?
Little known irony, a 22 gauge conductor at gig Ethernet speeds, carries almost all its current on the very edge of the cable...
This is misleading in the context (POE) of this discussion, since although it is true as pertains to the AC high-freq data signal, the POE current is DC and is carried equally thru the conductor. U4 should be ashamed of himself ;)
Coming from a microwave internet background, I have a question about reversing polarity.
Are there any camera manufacturers that do reverse the polarity on their POE?
In the microwave radio world there are a few manufactures that do. So I had a chart of model numbers and pinouts to keep from burning up devices. I've been searching and can't find anything, so I'm hoping they all use the standard in the camera world.
The chart that is shown for just 802.3bt PoE Classes is wrong.
The power listed for each class is for classes 0-3 and NOT 5-8. I know that this is an old document but that just means that it has been wrong and no one has corrected it for over 3 years. Copy and paste still needs proof reading.
Now that it’s fixed, it might be prudent to remove Jeff’s capture of the previous incorrect info, or at least put a big ❌ over it, as people might inadvertently reference it, or it could be returned in image searches without explanatory text yet still containing the IPVM imprimatur.