I’m trying to get 12V power to a non-PoE router, to improve VOIP resilience (it’s a TP-Link EX230V that has a the Voip phone socket on it).
Can a redundant MikroTik RBPOE Passive PoE Injector be re-used on a Cat 5 circuit from a UPS powered PoE switch to power the router with a short Male>Male barrel lead to the router, or is there something unique about the wiring of a MikroTik Passive PoE Injector?
So if I understand it correctly, you want to Power the TP-Link EX230V via a PoE-Switch
and use the MikroTik RBPOE Passive PoE Injector as a PoE-Spliter ?
Actually it can work quite well. (I’ve been in a similar situation and it was solved with good results.)
So… Mikrotik injectors are nothing special, they work like the ones from other manufacturers. Some things to consider:
Thanks for that - lots of useful info to get my head round.
I thought that the new PoE switches were “smart” using “active” PoE (802.11af)and could negotiate power requirements with their “end user”, so would automatically cater for 48V-12V step-down.
I’ve found a TREDNET TPE-104GS that has a power selection switch to select 12V, so it looks as if that might provide the required “step-down” from 48V to 12V. The PoE source switch is a MS105GP with 802.3 af/at 65W total (TP-Link EX230V is nominally 1.5A at 12V = 18W), so in theory should negotiate 12V anyway?
Standard 802.3 af/at/bt/… PoE specifies voltage around nominal value of 48V (depending on particular breed of 802.3 PoE standard it can be anything between 38V and 57V).
And then there are a few variants of “passive PoE” … where vendors went ballistic with different options … starting from voltages. So when you see something “PoE input voltage 12V-57V” you can safely assume it’s a variant of passive PoE (but can be combination of passive + 802.3 PoE).
As to different voltages: 802.3 af/at/bt PSEs (PoE switches) don’t convert voltage … they simply have built in power supply which outputs appropriate voltage (e.g. 57V). PA might provide lower voltages, but those are then used only to power switch’s electronics. Passive PoE switches mostly don’t convert voltages either, they output same voltage as they are powered with (e.g. when using 24V PA also PoE out will be 24V). Rare specimens (also MT has two or three models) feature dual-voltage power adapters (when built in) or dual DC-in jacks … and in those cases it’s possible to select voltage for PoE out (passive PoE only if voltage selected is less than 48V) between voltages provided to switch.
So far I never heard of a PoE switch which could actually regulate PoE out voltage (e.g. software defined) from single (high-voltage) power source. The TRENDNET TPE-104GS you mentioned is a “power splitter” with adjustable voltage out selector … for powering single non-PoE device … and yes, it does voltage down-conversion.
As I’m unlikely to be expanding my requirement for PoE retro-fitting beyond VoIP resilience, the learning curve to get my head round all that looks a bit steep. Sticking with a box with a voltage selector like the TREDNET TPE-104GS is probably my safest option.
Having said that, the TREDNET TPE-104GS was bought to include a future upgrade to FTTP in the VoIP resilience plan. I’m in UK & I understand that a BT ONT needs 12v 0.5A so basically max use 6W. The TPE-104GS is rated at 12 V (1.2 A), so enough for a fail-safe solution for the intended ONT, but not for the TP-Link EX230V — nominally 1.5A at 12V = 18W. If I want a simple fail-safe solution for a router, I’ll need to look for a different switched splitter box with a higher output, like the D-Link DPE-301GI, which I think is has a switched user select for 12V output and is rated at 30W, giving a nominal 2.5A for a router.
At least with all you guys filling in the blanks for me, I now have a better idea of what to look for next.
JFYl, there are el-cheapo splitters around with 802.3af/at that include a step down converter, the voltage on the ethernet cable is the 38-57 V, that is split and then converted down to 12V.
They are common to power non-Poe devices, typically security cameras.
So the more common ones are 12 V 1 or 1.5 A, and 100Mb, while you want a 1.5-2 A one with 1 Gb speed, examples:
https://www.amazon.co.uk/Splitter-Compliant-Surveillance-5-5x2-1mm-PS5712TG/dp/B08HS4NT13?
https://router-mods.co.uk/product/gigabit-poe-splitter-48v-to-12v-2a-5-5x2-5mm-dc-ieee-802-3af-at-standard-power-over-ethernet-splitter-adapter-for-security-ip-camera-voip-phone-ap/
The D-Link DPE-301GI seems to be an injector, not a splitter 
.
https://www.dlink.com/uk/en/products/dpe-301gi-gigabit-poe-plus-injector
The new version of the Trendnet (v2) is 1.5A@12V:
https://www.trendnet.com/langit/products/PoE/Gigabit-PoE-Splitter-TPE-104GS-v2
so, depending on how realistic are the 18W it may be enough.
Yep. If this is a one-off, you’re likely best off with a splitter.
As others have already insinuated, the nominal power requirements for these devices are usually way exaggerated. (Or to be a bit more blunt/common: include some “chinesium” factor - many wall wart power supplies you get from less reputable sellers aren’t actually able to meet the specs printed on them. I’ve encountered the exact same device being sold with different power ratings printed on it.) It’s best to measure it yourself, add some factor (15-25% maybe) and if you’re within the range of the splitter, just go with it.
Yep, but in this particular case it could be that a correcting opposite factor exists.
TP-Link doesn’t state the actual power consumption for this EX230V things, it says that it ships with a power supply 12V 1.5A.
If you check (say) the Mikrotik hex lite, It has a stated power consumption of 2W, the hex 5W/10W, the hex refresh 4W/10W, and all three are shipped with the same 24V 0.38A, so 9.12W.
It is likely that the the actual power need of the TP-LINK is more than 1 A at 12 V, but less than 1.5 A, let’s imagine it really needs 1.25A and the engineers already considered some slack and/or chose the next “available size” of 1.5 A.
So a 1.5A splitter, even if not really-really capable of 1.5 A due to the “chinesium” factor, may actually work, provided that this factor is not particularly large.
@jacklaz You misunderstood, I’m actually saying the same thing. I think in actual operation the device doesn’t draw more than 0.5-0.8A. (But OP should not rely on my guesswork, but his trusted multimeter) Shipping a larger power supply rather makes up for a lack of quality control from their suppliers, or standardization of parts, etc.
A good way to quickly and easily assess these things is to look at the thermal solution. Actually dissipating 12W (1A @ 12V) requires some larger heatsinks and a large well ventilated case (as an Mtik reference, the rb5009 dissipates less than 10W in normal operation - look at the thing’s thermal design) From photos, that tp-link is designed for about 5-8W dissipation without having excessive internal temperatures. 18W is plain out of the question - it would last hours/days before self destructing.
So, all in all we can have a power supply ( or a splitter/converter) that does not actually deliver the current on the label but that can power a device just fine because the device actually needs less than expected.
Yes, we are saying the same things, numbers in these matters are in the best cases a rough approximation of reality and the only way to know for sure is to measure directly what happens.
The voltage and current used by the device can be easily measured with a multimeter, as you say, the tricky part is to understand if the power supply is working at (say) 70% of its max capacity (good) or (still say) at 110% (not good).
Here also the thermal aspect is useful (provided that the power supply doesn’t let out some magic smoke immediately) the only thing that we can observe empirically is the temperature, if the thingy gets warm too fast or too much it is a telling sign that it is underpowered or however it is working over its intended load.
Thanks for all the useful information, which has given me a much better understanding of the pitfalls of PoE.
On the basis of “safety first” I have opted for a splitter that has a 12v dip-switch and left the “smart” side of the PoE Swith to work out required current - which seems to be working.
As the next step on the “resilience” I’m looking to hang a non-Mikrotik WAP off the back of the Mikrotik LTE MODEM (an SXT6).
The SXT log is showing “ether2 detected poe-out status: on” and “ether2 detected poe-out status: wait_for_load”, so PoE “out” seems to be there.
The WAP is rated at 12V 1.5A = 18W(?), but I have a memory that Mikrotik limits PoE out to quite a low level before over-load cut-out?
The PoE switch has more than enough capability to power both the SXT6 & the WAP, but that won’t be any use if the SXT6 trigger cut-out at the 1.5A rating of the WAP hung off ether2.
Where can I find the relevant information to check the sums, as climbing a pole to the SXT6 to “plug & play” the WAP to do an iterative test is not much fun?
The SXT6 has only one port?
https://mikrotik.com/product/rbsxtg_6hpnd
Ahh, wait, maybe you mean a SXT LTE6, this one:
https://mikrotik.com/product/sxt_lte6_kit
It has passive only PoE (both In and out) in a very wide range of voltage, 12-57V , it depends at which voltage you power it (through its PoE in port, ether1):
Details
PoE-out ports Ether2
PoE out Passive PoE up to 57V
Max out per port output (input 18-30 V) 600 mA
Max out per port output (input 30-57 V) 400 mA
Max total out (A) 600 mA
At a “low voltage”, common in “old” passive setups, 24V0.6A=14.4W
At a more “recent” common 48V, 48V0.4A=19.2W
The access point 12V*1.5A=18W can (barely) be powered using 48V but not at 24V, considering that there is likely some losses of voltage (much higher at 24V than at 48V) due to the length of cable between the power supply/injector and the SXT and there will be some dispersion (10-15% is a reasonable amount) in the converter of the splitter.
If needed the losses can be approximately calculated:
http://poe-world.com/Calculator/
An Access Point that actually needs 18W is actually a rare beast, so it is very likely that your device uses much less, you can measure it to make sure, but at 48V the setup should work.
Thanks for that. Yes it is SXT LTE6 with ether1 & ether2 bridged.
As far as I can tell the SXT LTE6 needs 11W and the “wifi n” Mikrotik WAP that previously run off ether2 rated at 4W. The PoE adapter previously in use was only rated at 24V (0.5A) so should have only provided 12W, so either the SXT LTE6 or the WAP doesn’t use as much as rated.
The replacement PoE Switch has a power rating of 65W as an ax-WAP was always going to require more power (rated at 12V 1.5A = 18W), and using the link you provided, a 64W switch should push the required power (SXT@11 & WAP@18 = 29W) at least 50ft - and the SXT is only hung on the outside wall of the house.
It looks like the ax-WAP won’t power up on ether2 giving error message:
ether2 detected poe-out status: current_too_low
I’m not fluent in Mirotik, but as far as I can tell, ether 1&2 are bridged on the SXT, so, at the weekend when I can take time “off line” I may try re-connecting things so the ax-WAP gets the PoE-in direct from the switch and it then powers the ether1 on the SXT from ax-WAP PoE-out. Can’t see that upsetting the Mikrotik logic loop on bridged ether1&2 and should give the power hungry ax-WAP first bite at the cherry?
I think you’re probably correct that the ax-WAP doen’t need 18W. Although I lack the technology to easily measure it, it seems to boot fine from a PoE splitter into the 12v port (TPE-104GS rated at 12 V (1.2 A)) - so that’s a 14.4W power switch. The PoE injector provided with the ax-WAP is itself rated 48V 0,5A (12W), so that’s probably a better indicator than the printed spec.
Well, you did not state whether your switch is using 24V or 48V as output PoE.
Loosely (and empirically) the max power stated in specs for Mikrotik devices is much more than needed, you can approximately consider that real world power needed is 60-70% of the specs, but there can be peaks of 30-40% during boot or if CPU(s) go at 100%, and the power suppiy itself will work better and last longer if it has a similar 30% margin.
The theoretical (as it could have some built-in margins) 24V*0.5A=12W power supply managed to power two devices, 11+4=15W and 12/15=80%, it seems perfectly plausible.
The 14.4/18 is also 80% so it seems plausible as well.