48v failing on WS-8-150-DC - killed 4 ports

[cwachs]

There are bad weeks and then there are expensively bad weeks... On going issues now with 2 separate WS-8-150-DC switches on a tower. Replaced one that we lost control of twice with a brand new one. Mounted it in the enclosure at the top of the tower. Ran a check of the Ethernet cables via the GUI - all fine - as they should be since this is a simple switch replacement.

As soon as we power up 48v POE connected to a Cambium ePMP AP, the switch makes a "pop" and then watts consumed jump up to 80 for that port. I thought it killed the radio but the radio works just fine on the Cambium POE injector at the bottom of the tower luckily.

So, my first assumption is bad Ethernet. We are using shielded braided cables with drain wire crimped into the connector. I replace both ends and test with a handheld tester and the switch. All good. Power it up in a new port (since the old port is now dead) and "pop"! Same issue. Not an Ethernet cable problem.

OK, now I try a different AP on a new cable. POP! 80 watts. That's three down. Three to go! When this happens, I can unplug everything from that Netonix port and if I turn 48v back on, it jumps back up to 80 watts on that port with nothing plugged into it. Port seems fried.

So, now I am thinking grounding issue. Maybe. We are 75' up a Rohn tower. I ran #4 green up from the base for use at the switch. Two of the three radios are bonded to this ground. The third radio is an ePMP 1000 AP with no grounding lug so it is just floating. The switch is mounted with two small bolts (through the two holes in opposite corners of the switch) to an aluminum backplate in the fiberglass enclosure. Both the switch ground lug and the aluminum plate are both connected to the #4 ground wire. The #4 runs to the bottom of the tower where it goes into a grounding rod. The tower is grounded to that rod as well at the bottom. We also ran #4 from the electrical grounding rod to the tower grounding rod. There is a Netonix in the tower shack (AC version) that is also grounded to this same ground. It's not perfect but we spent some time trying to make it much better than it was.

I then power up a small Ubiquiti radio we use for AirView on the DC switch at the top - 24v. No problem. Take that same cable, plug it into a 3rd ePMP radio and turn on 48v. Pop! Loose the port. 4 down.

I have two Ethernet cables coming up from the bottom AC powered switch that are spares in case of a top switch failure. I plug one of them into an ePMP AP and turn on 48v on the bottom switch. Works perfectly. That begins to rule out my grounding issue since they essentially share the same ground.

The old switch that we just replaced at the top never blew a 48v port and was wired identically. It, however, had other issues where we had to keep factory resetting it. It's on RMA right now.

So, I take responsibility for killing 4 ports of this brand new switch but I can't for the life of me understand why. Currently, my only thought is something on the internal circuit board related to 48v that is right next to the bolt holes is getting grounded out through my bolts holding the switch to the aluminum backplane of the enclosure. Possible??

Oh, firmware on the DC switch is 1.4.7rc18 with a Rev F board.

[Julian]

This is a block diagram of what I know about your setup, going by the past couple days worth of emails. Hopefully it's self-explanatory, what i'm looking for is a confirmation or set of edits that need to be made. I'm not currently sure how you're getting power to the tower, hopefully you can clarify.

CwachsBlockDiagram.png (8.36 KiB) Viewed 4737 times

This is going to be a process, hopefully this makes things easier.

Thanks,

Julian

[cwachs]

That looks pretty good. Blue line from base switch to tower switch is double fiber pairs. As for power, 24v DC is being generated (currently) by a MeanWell AC to DC power supply in the cabinet with the base switch. Everything is on a large APC UPS at the base.

[Julian]

so what we have to be looking for here is unwanted continuity between the blue and green lines. Where it's coming from, well, that I can't tell you.

I'm not familiar with the meanwell converter, how's that hooked up? AC line in, 12/2 up the tower? bonded?

Take a stripped stub of cat5 up with you next time, and measure the resistance of each pair to ground with black on ground, and then do the same with red on ground. Do this for both devices. A low reading in both directions would indicate a fault to my eyes, but that can vary with manufacturer, so i suppose what you're looking for is the old 'one of these is not like the other'.

troubleshooting is a pain, but remote troubleshooting?... well, there are words for that.

[cwachs]

Will do my best to get up the tower tomorrow morning and try this. We have one working 48v port left (I think) and can try your suggestion. I have a few other volt meter tests we want to conduct as well while we are up there. There is alot of FM antenna activity on the tower so it's possible one of those techs have done something but some simple ohm meter and volt meter tests will reveal that.

As for our 12v power supply, it is not bonded to the tower. Is is 12/2 stranded from the output of the power supply at the bottom (which is grounded) all the way to the terminals on the switch at the top - no breaks in the 12/2.

We are replacing that power supply with a 56v 300 150 watt supply (rack mounted) next week - but that will not change anything other than DC voltage supplied to the device at the top.

[sirhc]

Which specific model epmp radio?

[cwachs]

Using both 1000 GPS (the 48v version of the 1000) and a 2000 radio.

[wolfcreek]

We saw something similar on -48 dc tower site last year which caused us to pay to repair a couple of switches. What i found if I put a meter on the shield of the cat 5 cable there was a significant loop current flowing over the cat shield. Soon as we would turn on poe could hear some crackling and then it went pop. The tower is properly grounded(old at@t longlines site) and we have our own heavy grounds back to the rack from the main tower grounds but could not remove the loop current.

Long story short the new neg dc switches should not have this issue but have seen this on two tower sites using neg DC (pos ground) and attempting to use a pos ground dc switch. We were using with an isolated dc to dc converter and still had an issue. Both of these sites have multiple cell company installations and there must be some kind of loop being created in the grounds at these sites. We have other neg dc sites of our own that with a dc-to dc converter have worked fine but they have no other users on site.

[sirhc]

We saw something similar on -48 dc tower site last year which caused us to pay to repair a couple of switches. What i found if I put a meter on the shield of the cat 5 cable there was a significant loop current flowing over the cat shield. Soon as we would turn on poe could hear some crackling and then it went pop. The tower is properly grounded(old at@t longlines site) and we have our own heavy grounds back to the rack from the main tower grounds but could not remove the loop current.

Long story short the new neg dc switches should not have this issue but have seen this on two tower sites using neg DC (pos ground) and attempting to use a pos ground dc switch. We were using with an isolated dc to dc converter and still had an issue. Both of these sites have multiple cell company installations and there must be some kind of loop being created in the grounds at these sites. We have other neg dc sites of our own that with a dc-to dc converter have worked fine but they have no other users on site.

ehhhh, be careful, the WS-26-400-IDC simply has an isolated power supply, like you may have done with a WS-12-DC at some sites and worked fine. With the WS-26-400-IDC we simply combined an Isolated power supply inside the chassis that was 400 watts.

We did this because so many people using the WS-12-DC and supplied power with 2A power supplies and wondered why radios would reboot as it was starved for power. A WS-12-DC should have a 5A power supply in front of it if you plan to power more than a couple devices.

The power supply in the WS-26-400-IDC is isolated and should work for most people, especially those that were using the WS-12-DC with an isolated power supply but the 400 watt or 8 AMP power supply which is capable of 400 watts / 8 Amps sustained with surge spikes above that is a more appropriate size power supply for 24 POE ports of which 6 can be powering AF radios.

I can just see people already hooking a WS-26-DC to a 2A power supply and wondering why things go wrong.