Hello. I have two PowerBox Pro (RB960PGS-PB) that I am using in an emergency to power some equipment until my Netonix gets repaired.
FW 6.45.9
Connecting directly to an 18v battery bank, through the DC in jack, so this is not a power supply issue. BOTH PowerBox Pro have this issue, so it is not a damaged unit. This does not happen with an old 10/100 PowerBox, but I need the Gb ports ...

Powered devices have between 15m and 20m of FTP cable, so the cable lengths are short.

PD1 consumes <500ma, priority is 10
PD2 consumes <500ma, priority is 20
PD3 consumes <100ma, priority is 30
PD4 consumes <250ma, priority is 40

In this configuration, the log reads " detected poe-out status: overload " for PD4.
If I swap the priorities between PD2, PD3 and PD4, the lowest priority PD always gets the error. I cannot test swapping priority for PD1 because I will lose connectivity to the remote site.

According to the specs, at 18v there is a max 2000 ma total limit, and a max per port limit of 1000ma.

Does anyone know what is happening? Rush currents? I don't have many options, since it will take weeks to get a replacement Netonix to my location during the pandemic..
Thanks in advance for any insights -

can you run 24v? also check that the dc jack isnt loose. ive seen a few that dont engage well and cause problems.

make sure you go into routerboard settings and upgrade the bios as well. i have seen a version or two of routeros that worked weirdly with poe out on the powerboxes.

Thanks you for your reply changeIP-

All DC connections are tight, but I'll replace them tomorrow just in case.
Bios is upgraded to same version as FW, 6.45.9. I always prefer to use "long term" FW, but I also looked through the change logs of the stable FW versions and did not find any changes related with POE out.

The problem occurred last night when there was a power issue and the battery voltage dropped from 19.4V to 18.6V. It was not occurring while the batteries were being charged and floated. So you are right that at slightly higher voltages an overload is not detected, and therefore 24V should probably work. But the current only went up aprox 5% due to the voltage drop, and they are still considerably less than max specs. In fact, the total amperage is at <60% max amps, and individual port currents range from 5% - 45% max port currents according to specs.

I can run it at 24V, but it would add complexity and some work. It is a very remote site designed to run on 18V so as to simplify the design. I used a Netonix WISP switch to increase the voltage, but it recently got fried, and it will take many weeks to get replacement. If I cannot get this solved remotely, I will have to go onsite and will probably just figure out some workaround, but the question really is why does this happen when the power consumption is within specs? It should not be falsely detecting a power overload... Like i said, it is happening on two different units, so I doubt that the issue is a loose connection. More likely some weird inrush current and it decides to cut off the lowest priority port... And it is not happening on the regular non-pro version of the PowerBox, with higher amps.

Perhaps I should send this one to support@mikrotik...

Looking at the POE-out stats for hours, the "overload" condition seems to be detected when the total POE-out (sum of all ports) is aprox 1.1 amps. This value is low compared to the total stated limit in the specs of 2amps, but it seems pretty close to a +10% margin for 1amp.

Is it possible that there is a bug in the FW that enforces the >30 volt POE-out limits on the Power Box Pro even when the unit is being supplied <30 volts?

Has anyone else had a similar problem and/or discovered a solution?

this is kinda what i observed too so if we try to run 3-4 prisms on a powerbox pro we have to redo things and run 48v stepdowns.

At 4 prisms it makes sense. Each one would use close to 500mA peaks ... but I am seeing this behavior with 1 prism, 2 AC lite, and an M5 (all ubnt rockets). Total amps barely fluctuate between 1000 and 1100 mA

And I also see it on another PowerBox Pro running 2 prisms, an M5 and a simple DC relay. Also at about 1000 - 1100 mA

I just tested that if I lock the data rates on one of the Gb ports to 100Mbps full duplex, the mA used by the device drop by about 60mA, and that seems to be enough to stabilize the situation. But obviously I need Gbps speeds.

The fact that a 60mA change makes the problem go away means that at 1000-1100 mA total POE-out, a certain threshold condition is being triggered...

I have emailed the supout file to support@mikrotik. Hopefully they will investigate and include a fix in a future release, or let me know if there is a different problem.

Summary:

Voltage: 18v (Real range: 19.4v to 18.6v)
Total current consumption: 1,1A

When: 1,1A(Total) @ 18,6v (Overload detected on lowest PoE priority) Right?

I prefer the values in Watts to have a better idea. I also have a PowerBox pro, but I use 24v and a max current of 200 - 210mA per port (I never had problems because obviously are low currents).
I want to know which is the total watt consumption when the overload occurs.
I see that with a -60mA the PowerBox stay stable, but, what is the difference in Watt?
When you have exactly the same power consuption on old PowerBox occurs the same?

Best regards.

The important item is the current (amps), since watts stay the same. As the voltage decreases from 19.4, the current increases proportionally. The voltage range is actually 22 - 17, from equalization to almost completely discharged batteries. But the problem occurs at 18.6V or less

The electricity was re-established today at the site. I rearranged some radios to do some testing. At 19.5 volts, a Cut and paste from the CLI command
interface ethernet poe monitor [find]

I cut the power remotely and let the batteries drain, and I get the following at 18.6V, after the lowest priority POE power cycles
Obviously the values change, and I only reported an single moment, but it more or less gives you an idea of the relative amps. The watts don't change when the voltage changes, because
Power=Voltage*Current

When I put lock an interface to 100Mbps FD instead of letting it auto-negotiate Gbps, the current drops by approx 60mA at 18.6V
So theoretically
1.116W = 18.6V * .06amps

And in practice I also see approx a 1 Watt drop, also, which seems to confirm the theory.

I have not tested the EXACT same load on the power box 10/100. But I can tell you than I never saw this problem, and I have run a few prisms on them, with no problems. I do not have enough old Power Box to test, as I am using all of them now in production...
.

Can you test with a power supply at 18v @ 2A (or more Amps)?
And with 12v @ 3A (or more Amps)?

Regards.

Thanks for you reply krafg. But I can assure you that the limitation is not with the battery bank, and the cabling is oversized for less than couple of amps. I could power my home for a day on that battery bank. As soon as I can replace the Netonix, I will pull the Power Boxes and test the POE outs with a high amp 12V power supply and some known loads connected to the poe outs, maybe LED light strips, to see if I can determine the exact nature of the problem... I don't have any extra units to test, since they are all being used right now in production. But i really think that this is a FW error, that it is using the wrong comparison values to detect an overload condition, or that it is displaying the values incorrectly.

hi,

you do not say how far away the power lead is from the battery and powerbox (length) or what the actual wire diameter it is.

A cable that can "handle 2 amps" is not what you are looking for, what you need is a cable that does not droop under instantaneous high current spikes.

The latter will not show on the average numbers on the UI.

I would suggest making sure the power feed lead is as short and has as much copper as possible.

another way to resolve this and something we used to do for the 750UP when it's overload detection code was flawed was to attach a large electrolytic capacitor to the power input right next to the router. This would smooth out transient current draw and lower the effective input impedance.

In some cases we fitted small super capacitors inside the router case itself on the input power line.

Later firmware revisions largely fixed this and of course the "long poe in" option helped too which works on both inputs as they are just diode or'ed.

We also used to link out the diode on the input and the diodes on the poe outputs internally to reduce the volt/power loss in the more power critical off grid installations.

cheers

Bill

i have this "bug" too and if i put one poe with the priority 11, i get the overload only on that port, even though that port doesnt pull any power! so for me its a bug, and sill not fixed with firmare 6.48.3