Page 1 of 1

POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:08 pm
by WirelessRudy
For months we had a mini powering two Netmetals from Mikrotik and sometimes a SXT. The power feed of the mini comes over a 48V POE-in supplied by an airmax power adapter over a ftp cable some 70 meters of length.

Last night we has to disconnect the power adapter to change the 220V socket block connected to a UPS to make room for a new router supply.
After the powering up we checked the working of the netonix and all was fine.

This morning we checked and found one of the ports of the mini didn't supply power to one of the net metals anymore since only some 15 mins after my check last night. (Log)
We rebooted the unit this morning, switched PoE off and back on for that port (4) to no avail..
Maybe a bug? I don't remember exactly what firmware version but maximal some weeks old (1.3.6 - 1.3.7?) Anyway, decided to upgrade to 1.3.8 and after that things spiralled downwards. Everytime the unit rebooted it itself reverted back to a state with the last config.
The unit became accessible over my network only for some secs. just long enough to disable all poe-out settings. Then the unit cycled and came up normally and stayed up.
The moment we tried to enable one, or two of the ports with the two Netmetals, the unit lost connectivity and rebooted back to the last working condition.
We went to the tower and tried the netmetals. They both work fine running on a small battery. We could connect them with direct (mobile battery pack) power on their ports of destination and just ethernet cable connected to the switch and all worked fine. Traffic flowed etc. The battery delivers 18V to the netmetals where the PoE-out of the mini was set for 24V.
We checked now the port 4 & 5 to see if it would power up the SXT radio, which it did on both ports without any problem. So on a SXT the ports work fine.
We replaced the mini for another off the shelf, same results. It does power the SXT on each port, but not the Netmetals on any.... (this one runs 1.3.7)

We finally displaced the mini for a MT-Switch that we used before (but removed because in gigabit it had issues) and that MT switch now powers the 3 units (2 x Netmetal + SXT) without any problem. This MT-Switch (rb260GSP) is fed by a normal non standard 24PoE that comes over the same cable and delivers the same 24V voltage to all units and works normal. Both Netmals are functioning.

Now I have the both tried mini's on my desk and they both seem to work fine. The are powered by the same airmax power adapter and both can power up 2 netmetals and a SXT and even another unit.
We also did test both mini's on the desk this morning similar but the moment they were put back in place at the tower couldn't power the NetMetals anymore???

The only difference between the tower and my desk is the length of the 'feed' ftp cable. This is a very static cable and even in the process of swappign units or its power there is little movement needed. On the present connected 24V PoE MT switch (half the voltage, thus double the amps over now only two pairs of wire) this cable works fine with stable (no package losses) Gigabit connection. So the cable is fine....

On the desk we obviously have short cable to feed the mini. Only 1 meter.
Tonight I will make a very long cable and just put that on my desk to feed the unit. See what happens. Apart from that we are clueless what might have happened that the units we put much confidence in and that ran for 7 months now suddenly don't do their job anymore!
If we can't fix the issue we are in despair since we planned several more similar switches to deploy in our presently ongoing network upgrade. Now these plans are halted until we know what happened.

Await your answers/ideas.....

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:16 pm
by sirhc
WirelessRudy wrote:For months we had a min powering two Netmetals from Mikrotik and sometimes an SXT. The power feed of the mini comes over a 48V POE-in supplied by a airmax power adapter over a ftp cable some 70 meters of length.


Is this POE adapter supping power on all 4 pair or just 2 pair? You say airMAX which means 2 pair power not 4 pair power like the AF24 adapter.

Try using a recommended POE adapter like the AF24 POE adapter that supplies 1.2A on all 4 pair.

You can not use a standard POE adapter that only supplies power on 2 pair and expect it to get enough amperage. You may have damaged the switch trying to use a stadard POE adapter.

POE.jpg

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:25 pm
by WirelessRudy
Sorry, my mistake. Wrong wordings. The power adapters supplying the 48V to the mini's are the ones usually used for the Airfibre units. Not airmax.
I read the spec sticker now and indeed it says +(1,2),(4,5) pins. -(7,8),(3,6) pins.
It was recommended to use these....

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:29 pm
by WirelessRudy
Oh yeah, it also reads 50V 1,2amps...

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:32 pm
by sirhc
Not sure then.

I would run cable diagnostics on the cables to make sure all 8 pins are in contact otherwise not all wire would be carrying power?

Also I would bench test the WS-6-MINI units at your shop first before I would re-deploy them.

Test procedure is located in this thread: viewtopic.php?f=6&t=1259

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:35 pm
by WirelessRudy
When we had the mini connected I'd remember they showed sort of 46-47V input where the adapter is suppose to deliver 50V. A drop of 3-4 volts.
The present MT switch shows around 18Volts where the 2,5Amps/24V adapter should deliver 24 volts. A drop of almost 6 volts....

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:38 pm
by WirelessRudy
The cable test we cannot perform right now. To much traffic over the present MT switch to just stop that for replacing and testing.
But like said in the post, we did test both the minis on the desk, like I have one running now. With 3 units attached and no problems.
All we can do tonight is a purpose made new 70mtrs long cable to see what happens.

Tomorrow when my assitent is here again we can than fit the mini again and do some cable tests...

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 12:43 pm
by lligetfa
If the port is running at gig, then all 8 wires have to have continuity, hence PoE over all 8. Also the input voltage shows on the MINI so you know what line losses your are facing.

Re: POE issue occurred on long cable

Posted: Thu Jan 28, 2016 1:54 pm
by WirelessRudy
Yeah, we will take a look again tomorrow, to see what the line voltage is and if the cables are connected fine.
Its still weird that a unit that worked for months now suddenly doesn't any longer after a power cycle.
It's even more weird that similar, new, unit shows exactly the same problem.
It's also weird that both boxes on the desk seem to have no issues.
And its weird that a MT poe switch that in fact is supplied with less voltage, thus higher currents to supply same watts, on two wires work fine.

But ok, maybe one of the connector pins of the cable is damaged or not good anymore. Only a test with a unit will tell us. We can only do this tomorrow morning when its light and not a lot of traffic. We have to disconnect an AP and a backhaul link.

(I'm thinking? This night, in my basement where the central is located, I could supply power over that ftp cable to the MT Switch by the mini. The power input will have a short cable then where also a short cable will have to supply 24V on 2 pair only will supply the Switch. At least I can than already test the cable?)