Not sure if 5.19 was supposed to fix the power problem, but it hasn't.

it did not have to fix anything yet,

CoxWireless you say you could have used a smaller capacitor. What size? That level of electrical is over my head, but I was hoping to solder a cap on my 750UP. Is 24V 1200uf reasonable or what would you recommend?

I have another RB750UP to modify. I'll go smaller and see where the breaking point is at.
It will be a few days to complete...

Bravo for 5.20, it seems to work pretty good. Ubiquiti equipment has peaks of about 700mA according to current monitor.

Regards

good to hear that everything is fine, at least for one user.

hi, whith 5.20 on my 750up (poe firmware 2) i'vo got random reboot on ubnt rocket M5. Set "forced on" do nothing.

What to try out??..
log

hi, whith 5.20 on my 750up (poe firmware 2) i'vo got random reboot on ubnt rocket M5. Set "forced on" do nothing.

What to try out??..
log

They added a lot of information here:
http://wiki.mikrotik.com/wiki/Manual:PoE-Out

I haven't had a chance to try anything out yet, so I don't have any suggestions. Report back if you find something there that helps please. The option "ether1-poe-in-long-cable" sounds like it might help if you have a long cable run to the 750UP.

[/quote]

They added a lot of information here:
http://wiki.mikrotik.com/wiki/Manual:PoE-Out

I haven't had a chance to try anything out yet, so I don't have any suggestions. Report back if you find something there that helps please. The option "ether1-poe-in-long-cable" sounds like it might help if you have a long cable run to the 750UP.[/quote]

I run no poe to rb750up, direct to power suply. "ether1-poe-in-long-cable" test to on or off, same result.
I've just 3 ubnt (2 rocket, ether 3 ,4 and Bridge25 on ether2) on that RB and the only one showing reboots is on ether 4.-

if you are powering RB750UP using DC power jack - there is no point in setting ether1-poe-in-long-cable setting to yes. Actually there is difference as described in wiki.

hi, whith 5.20 on my 750up (poe firmware 2) i'vo got random reboot on ubnt rocket M5. Set "forced on" do nothing.

In the "5.20 Release" thread http://forum.mikrotik.com/viewtopic.php?t=64785 , one has problem with poe on port 4 also.
As I understand it, he get's it working by turning ether5 poe off. What are your poe settings on all ports 1-5?


/Paetur

Sent from my iPad using Tapatalk

to Paetur - it is same guy - hagar83 in both topics
to hagar83 - we need more useful information.

/interface ethernet poe print
/interface ethernet poe monitor [find] (on the moment when it is ON and when it goes OFF and when you enable disable)
What power supply do you there (in Amps)?

i don't have any M5 here, but several 750UP boards that i have here behaved properly so far.

to Paetur - it is same guy - hagar83 in both topics
to hagar83 - we need more useful information.

The user in other thread is "sindudas".


/Paetur

Sent from my iPad using Tapatalk

Ok, so far so good.
Changed ether4 (problematic) on ether 2, no reboots since then (half day).
I will try ether4 later just to double check.-

(sorry my bad english)

Had no problems before new poe coontroller with ubnt devices. After upgrade to new poe controller nothing works. The leds are flashing if I put any ubiquiti device in it. The device get power and loosing him asap.

Update: The problem only occurs when eth5 has POE enabled. Without POE on eth5 everything works fine.

He MT,

could you reproduce that problem or is a fix available?

unable to reproduce problems with ether5 and poe-out.

RM5 random reboots, AG27 working.

@Janisk
Maybe there is a problem in protection of the ports.

There was a power supply (24) with only 0.8A for testing at rb. We changed it to the original one. Is it possible that the eth5 is damaged and will pull down every port?

Bravo for 5.20, it seems to work pretty good.

Same here. No unexpected behavior to report so far.

Hi ...

I ordered 2 x RB750up yesterday (1 is spare) expecting to replace a RB450 and deliver power to 3 x RocketM2 ...

Regards

There is a major problem. And this is the second board

1. New RB750UP out of the box
eth1: Ubiquiti Rocket M5
eth5: RB433AH with one 52N
2. "system reset" to factory: ok
3. Update to 5.20:ok
4. Update poe controller: ok
5. Enable poe "auto" on eth1:ok
6. Enable poe "auto" on eth5: CRASH; both poe enabled ports flickering red very fast
7. Take off the Rocket and the RB433AH
8. Reboot: ok
9. Disable poe on eth5 an enable poe on eth4 "auto":ok
10. All devices running!

@MT: Maybe ist is worth to set that in your lab.

ether2 instead of ether1? And how much Rocket uses, maybe i can set another router instead of it? Also, is both of the routers in AP mode?

Rocket has 24V/1A DC power supply. So no more that 1A I guess.

you did not mention - AP or in client mode?

Sure! eth2. My mistake. I disregard eth1 everytime cause we want to have it switched!

Rocket is in AP mode. Before upgrade to POE 2.0 everything was powered fine.

@complete2006

I had found a similar issue as well (new out of the box).
Maybe you guys can duplicate this...

load 5.20 and update poe firmware.
turn "poe force on" ports 4 and 5
turn "poe off" ports 2 and 3

Now plug a UBNT (radio) device into port 3 and watch 4 and 5 LEDS go crazy.

So I've spent some time with 5.20, the new PoE firmware, and the "ether1-poe-in-long-cable" option.
Some great improvements but I'm still having issues.

Instead of the old behavior of reboots and current limit shutdown, with the new firmware I just get reboots.

I've got a 35v 220uf capacitor running now (not the 35v 2200uf anymore) and I can power 4 UBNT devices reliably. I'm going to put this in the field tomorrow for real world testing....

I've got a 35v 220uf capacitor running now (not the 35v 2200uf anymore) and I can power 4 UBNT devices reliably. I'm going to put this in the field tomorrow for real world testing....

Hi ...

Any updates on that?

Regards;

I had found a similar issue as well

@mikrotik You should be able to do this in your lab (this is all your gear):
1. plug in 750UP via 2.5A 24V include supply and computer in ether1
2. load 5.20 and update poe firmware.
3. with default config, only change poe off on ether3:
/interface ethernet 4. plug ether1 of RB750 into ether5 of 750UP
5. plug ether1 of RB750G into ether3 of 750UP
6. watch RB750 on ether5 reboot

@CoxWireless, @complete2006, is that what you were seeing? Is that what you meant by "LEDs go crazy"?

EDIT: I found "LEDs go crazy":
1. plug in 750UP via 2.5A 24V include supply and computer in ether1
2. load 5.20 and update poe firmware.
3. with default config, ether2 poe off, ether4 poe forced on, ether:
/interface ethernet 4. plug ether1 of RB750 into ether3 of 750UP
5. watch LEDs on ether4 of 750UP
6. set ether5 to poe off and the problem goes away

Anyone had any luck getting any model of Axis IP cameras to draw power via PoE on the 750UP ?

5.20, fw 2.0, tried auto/forced and multiple ports, but nothing.
Powering for instance a 750G is working.

I've got a 35v 220uf capacitor running now (not the 35v 2200uf anymore) and I can power 4 UBNT devices reliably. I'm going to put this in the field tomorrow for real world testing....

Hi ...

Any updates on that?

Regards;

I have one running with the 35v 220uf capacitor. System-Health reports a constant 21.8V
My current setup is:
24V @1A Ubiquiti Poe
|
120' of direct burial Ethernet
|
POE y-splitter (splits power from data)
|
RB750UP with capacitor. (power jack from Poe-splitter, data on port 5)
|
POE forced-on ports 2,3,4: NanoM5, LocoM2, Engenius 1650

I was planning to swap the POE injector for the 40w one included with the 750UP, but this is so stable, I may just leave it.

2.5A@24V PSU is over what rb750UP can consume (it is close to that value, so there should not be any issues regarding lack of power).

next problem - 50m - 70m with minimum set resistance for cat5e cable of 0.18Ohm per meter (and this is taken straight out of physics book) so we are talking 9Ohm (50m) and 12.6Ohm(70m) (usually higher quality cables have less resistance) so now calculate the voltage loss over the length of the cable. And then calculate how much power will use connected device when it uses 8W@24V

I have a following configuration and I am little lost here, whether it will work or not.

This resistance of 0.18 Ohm/m is for both directions i suppose?

I have RB750UP powered by DC jack, so as I understood, there should be less problems with such configuration. I want to power RB751G-2Hnd with it on 80m, that is 260ft. The fast Ethernet version of RB751 consumes 7W, there is no data about RB751G. Let's say that RB751G-2Hnd uses 8W of power.

So 0.18 x 80 = 14.4 Ohm loss on cable.
If I take max. current of 500mA and 8W, minimum voltage for RB751 will be 16V.
Loss on 80m of cable in that case is 0.5A x 14.4 ohm = 7.2V. 24V from power supply - 7.2V=16.8V That should be enough.

Will it work?

I had found a similar issue as well

@mikrotik You should be able to do this in your lab (this is all your gear):
1. plug in 750UP via 2.5A 24V include supply and computer in ether1
2. load 5.20 and update poe firmware.
3. with default config, only change poe off on ether3:
/interface ethernet

Code: Select all

set 0 name=ether1-gateway
set 1 name=ether2-master-local poe-priority=10
set 2 master-port=ether2-master-local name=ether3-slave-local poe-out=off poe-priority=10
set 3 master-port=ether2-master-local name=ether4-slave-local poe-priority=10
set 4 master-port=ether2-master-local name=ether5-slave-local poe-priority=10

4. plug ether1 of RB750 into ether5 of 750UP
5. plug ether1 of RB750G into ether3 of 750UP
6. watch RB750 on ether5 reboot

@CoxWireless, @complete2006, is that what you were seeing? Is that what you meant by "LEDs go crazy"?

EDIT: I found "LEDs go crazy":
1. plug in 750UP via 2.5A 24V include supply and computer in ether1
2. load 5.20 and update poe firmware.
3. with default config, ether2 poe off, ether4 poe forced on, ether:
/interface ethernet

Code: Select all

set 0 name=ether1-gateway
set 1 name=ether2-master-local poe-priority=10
set 2 master-port=ether2-master-local name=ether3-slave-local poe-out=off poe-priority=10
set 3 master-port=ether2-master-local name=ether4-slave-local poe-priority=10
set 4 master-port=ether2-master-local name=ether5-slave-local poe-priority=10

4. plug ether1 of RB750 into ether3 of 750UP
5. watch LEDs on ether4 of 750UP
6. set ether5 to poe off and the problem goes away

Does mikrotik have any answer to this?

I have a 750UP powering 3 951 devices. I must run poe forced-on. Because when I run poe auto-on the devices randomly reboots. any fix to this soon?

I have a 750UP powering 3 951 devices. I must run poe forced-on. Because when I run poe auto-on the devices randomly reboots. any fix to this soon?

Even if i run poe forced-on the 951 will reboot sometimes. Any tips?

any news CoxWireless ?

Do I have to modify my as you have done?

Hold on until I post my update later this week, made a change...

Oddly enough I ran into these issues with my RB750UP yesterday when I upgraded the PoE firmware. If I try to enable PoE on ether5 it basically does PWM to the power outputs. Check out my massive power draw, its two solid state relays.

Hold on until I post my update later this week, made a change...

waiting =)

Does mikrotik have any answer to this?

Since Mikrotik isn't updating us on any of this, I reached out to their support email. A couple weeks later, I was told "we are working on completely revised version of firmware."
Hard to say what it will fix or when, but it is in the works and they admit there are issues.

My latest update from support:

new changed poe-controller firmware will be included into RouterOS 5.22. It will
address "crazy lights" when in auto-on mode, weirdness with ether5 port. However
it will not address issues with forced-on mode.

Ive updated to 5.22
My Firmware is 3.0
My POE Firmware is 2.1

I am powering my 750UP over POE. Cable is long about 50m. Ive checked Ether1 POE over long cable.
I get random reboots (power off/on) on my powered UBNT devices. I have 1 UBNT bullet m2, 1 Airgrid M2 and 1 Airgrid M5.
First time i tried doing it over POE it was version 5.17 i think.. Devices would shut down and power up randomly all the time.
I connected Power adapter directly and everything was working for almost a month.

Now i dont have power in the builing where router is and I was forced to use POE to power on router.

It goes like this.... Some random time in day ... random device/s or all of them go offline. After 5-10 min they get back up.

Ive updated to 5.22
My Firmware is 3.0
My POE Firmware is 2.1

I am powering my 750UP over POE. Cable is long about 50m. Ive checked Ether1 POE over long cable.
I get random reboots (power off/on) on my powered UBNT devices. I have 1 UBNT bullet m2, 1 Airgrid M2 and 1 Airgrid M5.
First time i tried doing it over POE it was version 5.17 i think.. Devices would shut down and power up randomly all the time.
I connected Power adapter directly and everything was working for almost a month.

Now i dont have power in the builing where router is and I was forced to use POE to power on router.

It goes like this.... Some random time in day ... random device/s or all of them go offline. After 5-10 min they get back up.

Are you running auto-poe or forced-poe?

Im using Forced POE

Same results , latest upgrades . 5.22 , poe v. 3.0 . 200 ft. / cat5 24v poe to 750up . NBM5 on eth2 , Rocket 2M on eth3 . Random power cycling on both ports. No matter if auto /forced. Long poe checked for eth1. Still same results.



update: Tried Coxwireless remedy. Solved!

I had the same problem and I've been doing some tests. Correct me if I'm wrong, but these are my humble conclussions:

1) CoxWireless' idea about a capacitor over c14 seems to make things more stable, at least for me.

2) I think you just can't feed so much equipment using POE. The resistance on the awg24 wires is just too high. I've measured about 8,75 ohms over 60m of cat5. Mikrotik and UBNT gear seem to use a switching dc to dc converter, to deliver a fixed lower voltage to the rest of the electronic components and minipci cards. So they tend to have a relatively fixed power consupmtion, measured in watts. That means, the lower the voltage received, the higher the current drawn (keeping volts x amperes = constant). So, when the group of devices is powered up, the current consumption produces a voltage drop over the cat5 cable. As the voltage drops, all devices draw a greater current to meet their power requirements (watts). The greater current causes a greater voltage drop over the cable and so on. This iteration can reach a stable condition on a certain volts/ampere combination. But as you add more devices that require more watts, this iteration does not stabilize. Instead, voltage drop grows to a point where devices just power down. No firmware or settings in the world can fix this. An example to explain this: suppose you have 5 devices (a classical setup of a 750UP controlling 4 Rocket M5 with 90º sectors) drawing 4 watts each (I'm being benevolent here, rockets draw more than that)

5x4w=20w
20w over 24 volts = 0.833 A, running through 8,75ohm wires, voltage drop over wire = -7.29 volts, 24v-7.29v=16.7v reaching the 750UP
20w over 16.7 v = 1.20 A, over 8,75 ohm = -10.47 v, hence 13.52v on the 750UP POE in
20w over 13.52v = 1.48 A, over 8,75 ohm = -12.93 v -> 11.06v on the 750...
20w over 11.06v = 1.80 A, over 8,75 ohm = -15.82 v -> 8.18 v on the 750
20w over 8.18v = 2.44 A, over 8,75 ohm = -21.39 v -> 2.6 v on the 750!!

This is what i found when testing that setup in the lab. There is a maximum amount of watts you can serve over a cat5 cable, it depends on the real resistance of a pair of wires over "double" of the cable length. You have the same voltage drop over the (+) pair than over the (-) pair, so for a 60m (aprox 180-190 ft) cable you have to account for a 120m (over 360ft) of 2 parallel wires' resistance.

The solution: I'm running that setup (4xRocket M5 GPS + 1 750UP), with 2 x 1,5mm² wire to feed the 750UP on the tower via the power plug and a cat5 ftp cable FOR DATA ONLY, then I feed all the other devices via the 750 poe outputs. This setup (and others) have been working great on several towers.

One of my towers has 2 x 750UP feeding 1 Rocket M5 @ 42mts heigth, 1 x 411 AH + 1 XR5 @ 48mts, 1 x 433 + 1 x UB5 @ 38mts, 1 x 411 AH + 1 UB5 @ 36mts, 1 x 433 AH + 1 XR5 @ 34mts, and 1 x 411 + 1 x UB5 @ 28mts. There is a 24v DC UPS in a small cabinet below the tower, a 30mts run of 2 x 1,5mm² up to the lower 750UP @ 28mts, and an additional 9mts run of the same wire up to the second (higher) 750UP @ 36 mts. All devices have 39 days uptime, most recent power cycle for maintenance was 39 days ago.

The "long cable poe in" option only avoids the overload protection cutting power to the other devices when voltage drops, but it can never avoid the devices from powering down themselves because voltage received is too low.

Modified unit (cox wireless mod) feeding 3 rockets over 200ft. of cat 5 , using 28 v 3.5 a power supply . 750 poe out monitor very very erratic. Rockets in AP mode and are said to be @ 6.5 watts , but I'm seeing monitored spikes up to and in excess of 12w which is crazy. All new units, the only thing I haven't changed is the cat 5 feeding up the tower.

Modified unit (cox wireless mod) feeding 3 rockets over 200ft. of cat 5 , using 28 v 3.5 a power supply . 750 poe out monitor very very erratic. Rockets in AP mode and are said to be @ 6.5 watts , but I'm seeing monitored spikes up to and in excess of 12w which is crazy. All new units, the only thing I haven't changed is the cat 5 feeding up the tower.

I have been working with Mikrotik support to no avail on this issue.

When powering Ubiquiti Rocket M5 units we are seeing usage spikes upwards of 35W being reported (non modified RB750UP). I have verifed with Ubiquiti that they are indeed 6.5W. We even tried spilting the power (ie. pins 1,2,3,4,6,7 on one RJ45 connector, and pins 5,8 on a 2nd). RouterOS reports power usage across both ports but usage was still high sometimes the ports would individually be reporting >15W and we still had reboot issues.

This problem is extremely easy to reproduce and it occurs on all 4 RB750UP boards we have yet to deploy. We have some RB750UP boards powering equipment without issue but in all other cases we are using RB411AH with Dbii cards as the APs, and NanoBridges/PowerBridges as the feed.

Our final solution in this particular case was simplying bonding all the DC +/- lines together so everything is powered directly from the injector. While we loose out on flexibility this was very simple to do and we haven't had a single reboot in over a month.

Cheers

HEY MT , when are you going to address this proper?

still waiting..........

I haven't really understood some of the problems some guys have!
I'm powering on ports 4eth and 5eth, tow RB435G with DBII radios on them and no problem encoded till now.
My main config though is the RB411AH with DBII radios on them and no problem too.
Only problem I see of the RB750UP is when I "force" its outputs to forced ON, when I rebbot the 750, it cuts down the PoE on the ports for some ms which are enough to give a restart to RB435 and the RB411s.
IF I leave it on AUTO, everything works fine though.

Have you tried to work it with another PSU than its default?
Switching PSUs out there - most of them - are crap.
Try a better one.
The guy's idea with the electrolitic capacitor is a good one because the board avoids by this way the spikes and the noise of the PSU.

I've had a issue with my 4 ubnt rockets rebooting every second, but I changed my PoE setting on my RB750UP from auto to ON, and problem was fixed

So just by interest sake, is this problem solved now? ..I've checking out all the pages on this thread and it looks like some guys really have a hard time with their RB750UP...

Source of my curiosity: http://community.ubnt.com/t5/The-Lounge ... 101/page/2

Here's my latest status (prior posts elaborate my experiences with the 750UP over the past year).
Using the capacitor described by coxwireless on pages 3 and 5, I eliminated the problem of powering the 750UP at the end of a long poe run and then powering devices another 30' from the unit.
I have 3 of them deployed in the field right now. The first one I used the large capacitor (2200uf) soldered from one side of the board to the other (see page 3 of this topic) and it failed after about 8 months. That 750UP is now dead.
On the other two 750UPs, I used the small capacitor (220uf, soldered as on page 5) of this thread and it's working beautifully.
Powered by the stock PSU (24V @ 2.5A), at the end of 150' POE run from the house, I currently have 1 RocketM2, 1 NanoM5, 1 Engenius 2611, 1 NanoM900 and the load of the 750UP itself.
Using the MT POE2.0 firmware, each port supports up to 1Amp and I regularly see the ports fluctuating between 300-600mA. System Health reports 21.5V.
You have to set them to 'forced on'. Auto doesn't work.

Hi everyone, just a note on my experience yesterday with the RB750UP; (AND my fix)

I purchased a unit to run three ubiquiti devices in a switch mode, but then noticed at 12v i may go over the 500ma limit of POE each port. So i wired two nanobridges up (apparently 8watt) with power seperately. But then when trying to run my third device (Airgrid 5 HP) through the mikrotik which should be ok (3-6watt depending on where you look!), it kept power cycling every 20 seconds, 2 seconds, 20seconds... etc, not even enough time to log into it. Some describe this as flapping. Log messages were "link up" Link down" repeating every few seconds ....

This was all before I found that my RB 750UP should be able to output 1amp (1000ma) per port. See here: http://wiki.mikrotik.com/wiki/Manual:PoE-Out

So, long story short, I found my problem was a cheap 5m Chinese Cat5 cable! This tested fine on my Cat5e cable tester! But must be far inferior in design at 5 meters.

Anyway, i'm unsure if the Mikrotik POE current reading is accurate but my Airgrid runs at approx 280mA, but i've seen it as high as 400 mA briefly (at startup i think), and when temporarily testing a NanoBridge it drew about 350mA ?? (450 max). So i will eventually try running all three devices and may then give an update here.

There seems to be 2 different RB750 boards.

Mine doesn't look like the picture.
Anyone know where to solder the cap on the other (perhaps old) model?

Just to add..

I did add the capacitor to 2 different RB750UPs with resetting POE ports, and both are perfectly fine afterwards.

100-140 feet cat5 on POE port. Long cable set to yes, forced on.

Its funny I stole the capacitors from RB411s with dead ethernet ports.

Further update: I have switched all three devices now through PoE on the mikrotik in auto mode.

1 x NanoBridge 5HP (internet in) [ether2]

1 x NanoBridge 5HP (out) [ether3]

1 x AirGrid 5HP (out) [ether4]

Has been running flawlessly for a week at my solar site with temps close to freezing. I've included a screen shot of running currents (approx average screen capture), but they vary as follows:

Nanobridges 296-495 mA

Airgrid 268-376 mA

Much less than I thought, considering everything is 12v there.

On a side note, It's awesome to be able to log into the Tik to see my solar voltages everyday/night! (even dispite reading being low across the range by exactly 0.5v , but at least consistant!).

Hey guys just wanted to supply an update, I'm not sure if anyone is using the RB750UP in this manner or still having problems.

But after a couple of years in the field, I can say that the best mod for me has been this third revision.
As you know, mod1 was a large 35v 2200uf cap, mod2 was a smaller 35v 220uf cap across C14. Both of these modifications worked for me and some are still running, however I have came up with mod3 that works even better than these two.

mod3:
remove the factory C14 cap and replace with the 35v 220uf cap.

mod3 came about because I had a failure in the field after about a year with mod2. All of my mod1 units are still in the field and running strong, but I have found that the 2200uf cap is overkill and unnecessary.

Bravo for 5.20, it seems to work pretty good. Ubiquiti equipment has peaks of about 700mA according to current monitor.