Simple enough. I have a RB5009 PoE OUT version running several PoE devices. Input is 54V 3A via the DC jack. If I power 4 devices, everything is fine. If I force on the PoE to a 5th device, it hits overload despite only drawing between 50-60 Watts. If I force on 6, then the first 5 work fine. Force on 7, then 6 work fine, etc, etc.
Looks like a software issue since all of the ports work and it WILL put out the power, just keeps cycling the lowest priority device.
I found the RB5009 UP couldn’t power 3APs with a maximum draw of 11 watts each.
I had to slap a CSS610-P on mine.
Moved the UP version to the “Do Not Order” list.
Was this with the stock 48V 2A power supply the 5009 ships with, or did you try another adapter?
I contacted support and they had nothing for me.
I turned down the transmit power on the APs and the reboots stopped. This pointed directly at the RB5009 POE levels per port. Seeing as I needed 33 watts and the 96watt adapter didn’t shut down the router… I knew I had found another CRS354 level issue.
(Basically you’re F–ked and have to move on.)
By default RB5009 comes with 96W PSU and PoE-Out budget of 76W. There is psu#-max-power setting on the RB5009 for safety not to overload the PSU, The default settings are:
/interface/ethernet/poe/settings/print
psu1-max-power: 96W
psu2-max-power: 150W
routerboard-max-self-power: 20W
poe-out-limit-power: 130W
psu1-poe-out-max-power: 76W
psu2-poe-out-max-power: 130W
Your PSU gives you total budget of 54V * 3A = 162W. You need to change the settings of RB5009 to reflect the new PSU parameters:
/interface/ethernet/poe/settings/set psu1-max-power=162
Where PSU1 is DC-Jack and PSU2 is 2-PIN
You have still not replied to our last question in the mentioned support ticket. In case of overload, all of our PSE devices will instantly drop voltage from cables and in case of ROS, produce overload log entry, that does not happen in your case. As you know, your AP’s simply produce link-down and voltage is never stopped on the cables. What your AP does on it’s side to “determine” that RB5009 has not enough power cannot be answered by MikroTik. Usually such behaviour is related to LLDP where PD expects approval of power-usage and that is the last thing our support team suggested to You: enable PoE-LLDP.
If Gokusj5okazu or gotsprings still have issues after checking steps above, please contact support@mikrotik.com
@Reinis
With reference to Gokusj5okazu’s post, the sum (for the 6 devices connected) of the mA’s is 202+207+209+405+190+179=1392=1,392 A that, at 54V are 75,168 W, so very near to the default 76W cut-off, but if I get right he had issues with less than the listed 6 devices connected, so maybe there is something else going on:
4 devices (ok) 202+207+209+405=1,023x54=55.2 W
5 devices 202+207+209+405+190=1,213x54=65.5 W
Maybe the measurement of the output is not very accurate, though a 1-(76-55.2)/76=27% or even 1-(76-65.5)/76=14% seems to me too much, I would have expected a tolerance in measurements of 5% or so at the most.
Or -more likely - the devices have an initial (when first powered) surge/peak of current needed, in which case something to think about for a future release of Ros could be something like the “staggered spinup” that is (was) used for hard disks, to not overload the PSU at PC startup.
Reinis,
Gary over at Cambium Networks took over testing. I will follow up with him.
Locally we have not seen any issues for RB5009 to power up our PSE Load’s that takes board maximum power from each port in any scenario (cold boot, PSU power-cycle with different delays etc.) All of our 802.3af/at PSE’s have ~1s delay between each channel turn-on for the precise purpose: not to throw too much current surges at the same time as practically all electronics have higher current load briefly. Hopefully Gokusj5okazu will post results or contact support so we can get down to what exactly is happening in his setup.
Good to hear. If they can give us any details to why their PD behaves in such a way when powered by RB5009, that will clear the picture for all of us and what changes can/should be done from each party.
Reinis,
Updated the support portal with the RIF File.
Reinis,
Here is part of an exchange I am having with their support.
“From what I have seen, the PoE power draw on the XV2-21X AP can fluctuate (never spiking really high , just fluctuates) when the radio power is at or near max power and some PoE switches cannot handle that.”
For my install. I had to put a CSS610-8P in to the install to stop the reboots. As i stated before… when we turned down the transmit power on the AP the reboots stopped. After placing the CSS610-8P in the rack… we were able to run the access points at full power.
I used the CSS610-8P because it was available and I have one at my office, and never saw this issue. I also have not seen it since.
I have driven that access point in installs with:
CSS610-8P
CRS328-24P
EdgeSwitch 8-150
EdgeSwitch 24-250
EdgeSwitch 48-500
The PD is marketed as 802.3af which at no point can exceed 15W. RB5009 @48V will hold much more than that. But anyhow, if current-surge were an issue, RB5009 would power-off the port and provide overload log entry, that is not for PD to do. From our standpoint, it is still a black box what the AP is doing and why.
RB5009, CSS610 and CRS328 ar driven by exactly the same PoE, only voltage and some limits changes. The only thing that comes to mind is that the AP might want higher voltage, thus try powering RB5009 with at least 54V and see if that changes anything, if it doesn’t, then Cambium should provide solid information as of what exactly their PD does.
Also, as provided in support ticket, you can easily measure voltage on cables to confirm that PSE does not turn off voltage at any point. To reiterate, link-down != power-down
XV2-21X AP datasheet says consumption is 11W typical. It’s not meant maximum.
Questions to Gary:
The output of the CRS328-24P Driving an XV2-21x
[admin@Smash_Meri_Core] /interface/ethernet/poe> monitor ether1-Office_AP
name: ether1-Office_AP
poe-out: auto-on
poe-voltage: auto
poe-out-status: powered-on
poe-out-voltage: 51.9V
poe-out-current: 105mA
poe-out-power: 5.4W
I have seen a peak of 11.2 while watching.
If I hit the cli on the XV2-21X and ask it to show power
Device > show power
Platform Requirements: 8.00W min to boot, 12.95W max to draw
Physical Layer Power Range: 802.3at4 30.00 to 25.55 Watts at 100M cable length available
This was it 100%
The underlying issue is that I’m using the OUT version, which ONLY has the 2-pin Phoenix connector, no DC barrel. Since there is no barrel, the 2-pin is “PSU1” and therefor stuck at the 76w, for which there is zero documentation about (or at least it’s not made apparent).
Thank you!
Gokusj5okazu,
Does that mean the maximum power budget for the out version is 76 watts?
So that addition Mikrotik made to the product description about using a higher power supply, means NOTHING?
I think the underlying problem here is that you need to change config to allow higher power output.
Default it’s at 76W on psu1.
That’s what I understand from all previous info.
Depending on your model (outdoor), there is no barrel, only 2-pin connector but it is labelled PSU1 and that’s the part which does not seem to be documented enough.
The block diagram for indoor and outdoor makes it clear.
What’s not logical IMHO is that for the OUT version labels are changed for PSU, it adds to the confusion.
See post again from Reinis:
That last line makes it clear he (she ?) was referring to IN version.
But OUT only has 1 PSU. So that’s PSU1.
holvoetn,
So you can set the PSU1 to a higher power? Its not locked on 1. Just making sure
I was curious if this setting exists in other units so I tried that settings command against my RB960PGS-PB.
You get a syntax error as soon as you type sett
Reinis gave you to config code to change it.
But if we think about it logically, I would think psu1-poe-out-max-power should be changed accordingly (new max minus what’s allowed for own operation).
Unless it’s a calculated field ?
Then the next question is… “What’s the actual limit of the hardware? How much PSU can we attach to this thing and not blow it up?”
I asked Cambium to try some other access points on the RB5009UP. The problem didn’t follow.
So it looks like a compatibility issue.
That means we need to continue on with the voltage and power supply question.
Support should be able to tell BUT …
from what I see in the published specs, both RB5009-PoE versions IN (130W) and OUT (120W) allow less output power then what’s being suggested here above by Reinis (change PSU-max to 162W minus 20W for own operation = 142W).
So it’s safe to say that only adds to the complete confusion surrounding this subject …