When water incursion shorts out a 48VH cable

[Brough]

In the early hours of this morning (during rain), a two-year old but now damaged cable caused the PoE power to an AirFiber 24 radio to short out. This knocked out the WS-12-250AC switch that powered that AF24, and both of the site's routers, causing the site to go down. Port 1 of this switch is likely fried, but once a tech got there, identified and then unplugged the shorted cable, he was able to reboot the switch and the rest of the site came right back up. We've scheduled more permanent repairs for Friday.

The switch is board Rev F running 1.4.7 firmware. What data I have is attached below. It appears port 1 went down at 2:13am, the switch went down at 2:24am, recovered briefly around 2:28am and then went down permanently at 2:29am.

I'm assuming the power supply shut down or went into current limiting as a result of the short. Does this make sense?
Is there per-port fusing or per-port current limiting? or should there be?

This is a site that has grown over time without a plan. It now has 12 radios, 2 WS-12-250AC switches, two routers and a UPS. With even a little planning we could have remotely recovered all but the one dead radio. So, when we rework the site, we will dual power each router from each Netonix switch. Then, if one switch goes down there are still paths into the site by which we can hack in and reboot. And of course a careful inspection and rework of any marginal cables should help. Ah, the benefits of hindsight...

Syslog messages from the switch

SNMP responses from switch

SNMP queries.PNG (11.89 KiB) Viewed 3985 times

Smokeping with the switch

WS-12-250AC

[sirhc]

Our current models use Polyfuses same as the UBNT ToughSWITCH.

Polyfuses are thermal breakers, to learn more about them Google Polyfuse.

Polyfuses are there to prevent fires, since they open too slowly they do not protect the switch port which often gets fried. However our LAB testing showed most ToughSWITCH would completely fail and never boot again from a dead short cable whereas most times the WISP Switch would only lose that port which can often be repaired for $50.

I suspect that the switch shut down because water leaked into the switch sef through the Ethernet jack? I would pull that unit and RMA it before corrosion causes more damage and requires a complete board replacement which is still cheaper than a new switch.

We are working on what we call active over current protection for our next series of switches which will protect against fires and protect the switch itself.

As I stared above the Polyfuse is oversized to provide .75A per pair @ 55C environment and since the switch is often in cooler environments so the polyfuse provides more than 1A per pair and there are no Ethernet Transformers available on the market to handle more than 1A per pair so a shorted cable burns out the Ethernet Transformer and sometimes this damages the Transformer in such away that it conducts current across the magnetic coupling into the PHY which is in the SOC permanently damaging the port.