What is exact POE standard for the ePMP 180, 1000 or 300-16?

Trying to spec a comparable replacement for the Ubiquiti NS-5AC. I’ve tried looking at the datasheets for the ePMP Force 180, ePMP 1000 and the ePMP 300-16, but they don’t make it clear on this critical question: can any of them be powered by standard 802.3af or 802.3at POE switch [edit: without] an injector?

We don’t want to use anymore radios that use none standard POE that requires an injector.

Force 180 is not 802.3 af/at, it is powered up with cambium poe (we are solving that with eth2 on e5xx where eth2 is configured for cambium poe, but I am not very happy with that

Mixig, thanks for the reply. Cambium and other manufacturers are really missing the boat here. No one wants to mess with injectors anymore, that’s we’re increasingly looking for access points and bridges that natively use standard 802.3af power (and not with an inline converter). Makes rebooting WAPs remotely much easier and less components to have to deal with.

1 Like

Yep, it will be much easier when all they product are e.g. 802.3 af/at. In our standard installation / setup we can power force 180 via cnpilot eth2 cambium poe, but force 130 is using 24V passive poe :smirk:.
It will be much easier to put poe switch and power all device from that switch.
In most cases we have only one eth cable on the pole and few devices, only solution is poe switch with one eth port with poe in. In that case only devices with same poe standard come into consideration to be pluged to poe out switch ports

the use of poe injectors is a mixed discussion, especially if you are using a sync injector like Packetflux. Plus the lower and wider voltage requirements is nice for sites that run off batteries. 802.3af/at all require power to not exceed 56v, but the charging voltage of a 48v battery can be as high as 62v during equalization whereas a 24v system has a 29v equalization.

This is all besides the fact that the radios use 5 and 12v internally regardless of the supply method. The higher supply voltage is to reduce the current over the wire and the devices have a converter to drop the voltage and thus increase the available current. Still have to keep below the max wattage of the wires though.

Douglas, those aren’t bad points, but after years now of the IT field having to deploy and support wireless access points, I think it’s become pretty evident most in the IT community much, much prefer 802.3af/at compliant radios than having to use injectors with 24V radios. In the community forums where this has come up, almost everyone agrees. Especially for edge devices in remote or off building purposes. You can’t reboot a WAP remotely when it is hung up and on an injector, unless you go through the extra expense of putting in an IP enabled PSU, too. And while in a perfect world field techs should know not to plug 24v passive POE radios into active 48V POE switches, experience has born out this is far from a perfect world.

I can see where the power requirements for 24V Passive might be advantageous for some applications, like solar power and mobile vehicle power applications. But ironically, that would be niche use case that would be a small % of what is desired for the majority of applications.

So what we’re seeing after decades of experience now is 24V Passive should be more of a niche use for a smaller number of deployments while the market prefers more native 802.3af devices. Ubiquiti seems to be the only manufacturer really taking this seriously. They were one of the first to offer in inline convertors so you could attach 24V radios to 48V POE switches for the entry level radios. And one of the first that I saw to start offering entry level radios with native 802.af support. It’d be nice if Cambium and other manufactures could ramp up their competitive offerings so we don’t have to always be circling back to Ubiquiti.

the Ubiquity converters work on Cambium radios (except in a specific set of older radios which are not polarity agnostic).

If your techs are doing things like not checking the power type before plugging in, then that is an education issue. After 20+ years in the IT world as a professional and the past 16 working with Ubiquity, Motorola/Cambium, Early Mimosa, Microtik and the old AeroNet and WaveRider systems, education and understanding is paramount.
They make 48v solar panels too, these are industrial grades. Same panel just configured differently.
The problem comes to powering other devices. Most devices are actually 5, 12 or 24v inside their cases so it makes sense to not add an additional component to the device if you dont have to. That 802.3at converter has a fairly high cost as it is a licensed technology, this would increase the cost of the radios.
The dedicated APs that Cambium puts out are 802.3af/at compliant. So right product for the right need? Using a SM in AP mode is not the standard but the exception to the designed use. And since the epmp line is not a WAP, its a bridge (ptp or ptmp, but still a bridge), comparing to a WAP is wrong.
Now for the cnPilot line, these are WAPs and should be powered by 802.3af/at and should be able to be powered on passive 24vpoe. These may need to be two separate product lines for the same radio. It is nice being able to put up a solar wifi spot in a park connected to the epmp network and only have one voltage to worry about.

IP enabled PDUs are nice but expensive. You can get IP enabled injector with local system monitoring and autonomous action that works on your chosen system voltage (packetflux site monitor system). Yes these are expensive too, but when on battery run time is important and allowable battery bank dimensions are restricted, double converting your battery voltage to standard 120vac and back to 48v (56v) DC just to be reduced again to 5, 12 or 24v is very wasteful and reduces your available runtime from days to minutes. This is why UPS systems do not run for hours until you get into datacenter size UPS systems.

And while we are quoting “standards”, it is apparently standard to go with whomever has the feature of the day for hardware. Running to Ubiquity because they have the one thing you want regardless of the actual quality (still can believe they still dont properly shield their radios) is a personal preference. You still prefer Ubiquity so they are your go to.
My main competitor is a UBNT shop, yet I keep getting their customers switching to me whom is 99% cambium based (I still have a few non-cambium Wifi WAPS in place and am not changing them until they die). Dont get me wrong, I have my troubles too, I come on here and see whom else is having the same or similar and complain when and where needed to. So far my complaints have been addressed or in the case of a couple, they were incorporated in the next edition of hardware.

I am not disagreeing with you and your stand point, it is valid for some of the product line. Which is why Cambium has made it possible for those products, but they have also seen what we the WISP community has been doing and how. This also drives what they offer. Deciding something s niche just because your experience says it is doesnt reflect the reality. Then there is the regulatory side of things, technically any cable that is to be carrying a voltage over 30v in Canada must be installed by an electrician with a permit. All connections must be made by said electrician too. Now there are some exemptions in place for data and telecom, but you cant rely on them as at any time an inspector can come in and order a stop work plus fines for non-compliance.

Like most things, there is always more to the issue than what we normally see and experience. So take this as just what it is, my point of view and nothing more than engaged conversation.

1 Like

Hi Douglas;

Would these type of 802.3AF/AT PoE converters work? Since I thought the ePMP series automatically senses and works with any of the 24V Passive POE. Since, in essence you can use a Mikrotik, Ubiquiti and/or Canopy PoE brick/ inserter with any of these CPE radios (such as the ePMP Force 180 and up). That way, you didn’t have to enter the customer premises to do a power supply swap. PoE agnostic.

There are cases where, say at your tower site (WiPOP), you have a cluster of GPS synch radios (or you just want to use standard managed PoE switches) to power all of your APs and PTP (ePMP backhaul) radios. It appears that these new ePMP Force radios are all using passive POE injectors. So having them plugged into and powered by managed PoE switch seems like an option one would want to use.

It does appear that all of these do the following:-“They accept data-in and 802.3af/at PoE to a shielded RJ45 Jack and provide data-out and 18V or 24V PoE on the shielded RJ45 output jack. They work by supplying power to the remote device on the unused Ethernet pins 4,5(V+) and 7,8(V-).”

Whereas, the information from Cambium ePMP POE power requirements (link above) has the following:- “Compatible with Cambium PoE pinouts (V+ = 7 & 8, Return = 4 & 5)”

So the voltage on the pins are reversed. However, if the ePMP is POE agnostic (based on comments found here and elsewhere), would this still matter if you used one of these 802.3af/at converters? Maybe Cambium support can also comment on this. Though I know they were saying that their new POE switches is supposed to work 24V POE devices.

So something like these may still work?:-

Tyco’s 802.3af/at/bt to 24V Passive PoE Converter

Tyco’s 802.3af/at PoE in, GigE 18/24V Pasv PoE


Ubiquiti’s Instant PoE Converter, Indoor

[Part Numbers here for quick reference (Google search)]

  • POE-CONV-4824G
  • CM-AF-24-13-I
  • TP-POE-1824
  • TP-POE-1824G
  • POE-165S

Be careful, not all versions of epmp equipment is passive poe polarity agnostic. See the admin and user guides to help you determine this.

The e3k APs are 802.3af/at (this is a wattage difference not a pin/polarity difference) compatible with an emphasis on using 802.3at as the AP can use more than what 802.3af can support.

We use packetflux sync injectors which have two power source inputs for different voltages. We mostly run our systems on 24v as it does not require converting from battery voltage to whatever is needed. Where we are running poe switches with poe output, we run 48v dc and place a voltage regulator to keep the battery voltage under 56v while charging (equalization can be as high as 64v).

If you plan to use a 802.3at to passive converter, you will need to know which harware revision you are using. But in short they could work, but so does just using a passive injector with a dc feed and a buck converter.

Personally I would be evaluating why I am using a 802.3at poe switch in the first place. What do I need 802.3at for? If you have a ip camera, then a small dc to 802.3at injector may be more economical and easier to support.

“Personally I would be evaluating why I am using a 802.3at poe switch in the first place. What do I need 802.3at for?”
Being able to use a 802.3af/at switch is about simplicity of setup and remote control. So you don’t have to have a separate dumb injector that you can’t remote into to reset the wireless equipment. With a managed POE switch you can remotely reset your wireless equipment, cameras, VoIP phones, door controllers, etc. You can setup a ping watchdog to automatically bounce the port, too, if a device hangs and stops responding to a ping. Things you can’t do with regular injectors.

They make managed injectors too, see packetflux’s website. Yes its a more complicated setup but there are massive gains not converting to 110vac just to convert back to 3, 12 and 48v inside the switch. Your backup time on an inverter is less than 1/12th the run time on battery direct. We used to have the same mentality of one box to do it all, but found the run times to be abysmal. A battery bank that would give us about 6hrs run time using an inverter now has us running 72+hrs which means no one has to run out to a tower on a weekend just because the power dropped.

You could also use netronix boxes, these are very capable and you can cycle a ports power independently but no common sync, do not rely on the internal gps, its a good backup but go with a sync over power solution.

Personally I do not see the need of limiting to a poe switch just for a small feature of power cycling a port since there are network controlled relays that are more than capable of interrupting power on an injector based on ping or any other network metric. Network-wise your cable passes through two devices, an injector and an LPU, so one more thing to add to the diagnostic path but is a lot cheaper and quicker to change than a whole poe switch.

In the end it is your choice though. From a tower point of view the e3k AP supports 802.3at power as well as the latest generation of f300-25 radios.

For what it’s worth, they do make 12V and 24V 802.3at POE switches, also. But I’m not educated enough in electrical to make a guess in how much run time efficiency there is in between 24V POE and 802.3af/at 48V POE.

However, we may be talking apples and oranges in different worlds. Your world seems to be more dependent on high availability remote locations and maybe solar. In my world not so much, so battery run times are not as critical.

My company provides internet and cloud services. Down time can make clients leave for other service providers or encourage a service provider to encroach as you create a market for them. We have a 99% uptime standard and most of the time we are in the 99.99% range with a couple towers hitting the 99.999 and 99.9999% ranges. This is very important if you have clients that rely on your network for their living, which they gladly pay a premium for but expect to receive the up time too.

Solar aside, our power distribution grid uses a ground return system on anything over 1000v this saves cabling costs by only requiring one high voltage cable and no return down the pole, but an a simple HI-ESD ground strike can knock out a cross connect leaving a site without power. So its not uncommon to have no power at a site for 12 to 24hrs and generators are nice and all but are very expensive to place and must be serviced, tested and cared for more often than batteries. Batteries on the other hand, if setup correctly, require a site visit once per year to add water (flooded cells, but a good charge controller and lithium can be used) and are easy to get practically everywhere should one be failing. Having a large reserve also means that night time visits are reduced to near none and a daytime trip to a site also reduces hazzards. The one time we had to take a generator to a site was due to the power lines being missing after a storm. The site ran two days on battery and we recharged the bank with the generator in about 4hrs for another 48hrs of battery time before power was restored. There is a point to having large battery banks.

Battery run time is simply the Wh rate of the battery bank divided by the Wh rate of your equipment. If you dont have the Wh rates for your battery bank then use the current draw of your equipment and divide the Ah rate of your bank by the current draw and divide by 2 (50%DoD). This is for new batteries only, every year remove 5% or if you completely drain a battery remove 8% immediately (they dont like to be dead).

Efficiency is a large topic but very simply your general UPS uses 24v and has about 15mins runtime at 50% load of whatever the VA on the box says. It doesnt get much better as you reduce load as there is also the galvanic properties to take into account which need a minimum amount of current flow to cancel out the losses.
If you have a nice APC 48v UPS, these are more efficient but they still have switching time, albeit in the ms range, but its enough to reboot equipment. The best you will get is 80% capacity as you still loose power in the step up process and batteries are a time based current source. The higher you step up the voltage the more current you pull and the less time you get on battery.
Once you go to dc powered devices the efficiency starts at 80% of available energy and depending on your choice of devices this can reach 98%. This is equivalent to doubling the backup time.
Most devices are actually 3v, 5v or 12v inside and have down converters that provide this from some higher voltage. Most are buck-converters that are designed to be as efficient as possible with little to no waste heat. Telcom network switches that run -48v are no different than a enterprise/datacenter switch but has a voltage converter inside it. If its a poe switch it has a filter network and regulator before its passed to the 48v rail. These are about 95% efficient and are good choices. If you use a 24v switch it will have a voltage booster for the 48v rail, but it will only boost enough current to supply what is being drawn. Still in terms of standby time you will get between 90 and 95%. This is due to the higher current on the supply leads as compared to 48v. Basically you double the voltage to halve the current but keep the same wattage. So If you are drawing 1A at 12v (12w) but your cable can not support 1A you can double the voltage to 24v and only draw 0.5A and still keep your 12W of power. Once you know all of your worst case wattage requirements, pick a voltage, then divide watts by volts to get current. Now if you are using an inverter/UPS you need to know the battery voltage and output voltage. Eg 12v battery 120v out is a step up factor of 10. Take your current and multiply by this factor to get the battery current. Add 10% to this to get a very close actual battery current and then take the battery AH rate and divide the AH rate by the current draw.

Other things to consider:
Cheap home or SMB UPSs are not designed for long runtimes or extreme temperatures.
Enterprise UPSs are expensive but are rated for long runtimes. But you are limited to how much battery you can attach as the UPS will not charge it within normal parameter times and you may not get a choice of battery bank voltage which may require an electrician to service due to your local regulations (here anything under 60vdc is good).
Not all UPSs are good for this type of equipment and the switching surge may reboot some or all of your site unless you opt for an online UPS, but that is way more expensive.
Battery cable size: the higher the voltage the smaller the cables can be and thus cheaper to install. 8awg cable carries 200a at 60v (use 125c rated cables, you will thank yourself for it) but only 50a at 12v.

But also consider that we use 12awg 60degree wire for our 24v systems and could (we just stick with 12awg for simplicity) use 16awg on 48v. Our 24v sites with 4AP sectors and two BHs (the AP doesnt use too much more power) a cisco 2900s class switch (ciscos non poe switches are just 12vdc so we add a buck converter for it and the cameras) and a couple packetflux injectors with syncbox jr. and 2 IP cameras (we get these nice 12v ones) draw 9.68A at 24v, even 12awg is overkill. We place a voltage controlled power supply in the box that is twice the current requirements (we have 500w power supplies set to 27.6v and a reverse current diode). Yes the 500AH battery bank takes 5 days to fully recharge but we usually have the time and when we dont, we have a large capacity(100A) charger and a generator to make sure the site stays up.
Our one remaining 48v site is also a disaster recovery site. It houses our backup servers and cold NOC. We also use it for off-site client data backup services. The online UPS there just needs to run on battery long enough to let the big generator start and cut in. Then the batteries are recharged by generator power until utility power is restored.
Most of the ramblings are to point out that there is most definitely more than one way to skin this cat and each has its advantages and drawbacks. We are very comfortable with dc systems and have no problem copying what the telephone companies have been doing for years (yes your land line is -48vdc powered): power supply to battery to device. If you look, you will find what you need to do the job, usually very affordable and easily integrated.

So… yes.

Just want to weigh in as I spent a fair amount of time debating the Inverter vs all DC plant thing and, for now, have gone the all in one Charger/Inverter/Transfer Switch route.

All DC with no inverter is better in every way except 1 when you have mixed polarities and voltages, complexity (if everything was -48v I wouldn’t even think twice about not using an inverter).

While we have some subcontractors for installs, tower climbs etc… I’m pretty much a one man show when it comes to 3am outages (or any outages). If I hope to ever be able to take a few days off then I need a system that is simple enough that I can remotely walk someone through replacing stuff in the cabinets. If everything in the cabinet was -48v or hell even +48v then going all DC and no inverter would be a no-brainer but when you have +24v and -30v and -56+ and +48v devices it’s just not simple any more.

Seriously, I have trouble with guys on the tower plugging the RED-BLUE-RED striped cat cable into the MAIN port on the little box with S1SE in big letters on the back… You’re 10 minutes trying to figure out why cable tester isn’t seeing the dongle "Umm… you sure you plugged Red Blue Red into that ? " and tower guy is like " Yeah. Well, Umm, REd ? I don’t see a Red I think that cable is like maybe, I want to say yellow ? on it " or the one that has SE painted on the back in big letters points South East and the one that says SW points South West and then later your trying to figure out how people to the SW of the tower can’t see S0SW but they can see S1SE. at least that doesn’t burn up $1,000’s worth of hardware. So yeah, Inverters for now.

I feel your pain, even with box diagrams in every box, everything completely labled in clear english I still have trouble getting apprentices to do as needed, especially when I am on another site 10 miles away.

Basically what everyone is saying is stick with what is comfortable for you and plan for a 3am walk-through to the least technically inclined person you know. That way you might have a hope of having a vacation (whatever those are).

The difference between a -48v device and a +48v device is simply which side is grounded, which isnt always earthed though should normally be. Else if you have a floating 48v system an a pair of 60v MOVs and 68v TVL diodes on each side to ground, then it does not matter unless lighting hits both devices at once. Btw this is how 802.3at is protected, it is a floating 48v system.

You can power many devices with various voltage needs off a single voltage using buck/boost converters. This is what is inside almost every 802.3at device you can find. Thats how common it is, af devices may use simple drop regulators as power requirements with af is gennerally really low.

Just pick a single, most common and higest load voltage and build from that. That way its simple PSU > fusing &RCD > battery >more fusing and breakouts -----> common voltage devices, converters to whichever voltage you need.