Wisp switch temperature

[sirhc]

I have read the grounding. Since this is not a tower, we are grounded a bit differently. In the attic where the MINI is, it is grounded to a ground bar installed by the solar contractors. That is the only ground available up there. It has #6 from the attic to the electrical service ground. The MINI power supply (AF24 POE) is grounded to electrical service via the UPS it is connected to in the basement of the house. No Ethernet surge protectors are in the mix. I don't have the ability to run a new #2 in the house from ground to the attic.

Ethernet is shorter than ground from POE to MINI. Can't run a new Ethernet from basement to MINI but I could put an Ethernet surge protector in place and then add a long service loop to make that longer than the ground line. But I'm not convinced this is a grounding issue.

I am going to likely move the MINI outside the house to the base of the J-Bar (where the antennas are) to help with heat soon as my weather proof enclosure arrives. But, the switch is under 90C now and is having 3v alarms every couple of hours.

I am almost 100% this is NOT a temperature issue.

OK so I can understand correctly what is going on.

MINI is in attic but is powered from a Shield CAT5 ESD wire like ToughCable with ESD end that runs to the basement where the AF24 POE brick in plugged into a UPS that is plugged into the house AC service?

There is a #6 ground wire that runs from the AC service ground at the service panel or the service ground rods to a ground bus in the attic to ground the solar panels. There is a ground wire that runs from this ground bus to the WS-6-MINI chassis.

The one bad thing here is the Ethernet cable run is shorter then the intended ground run so any ground potential difference or surge will follow the Ethernet path. Yes you could insert an Ethernet surge protector and add a service loop to increase this path length but it would have to be a surge suppressor that accommodates the ESD shielding like the one UBNT sells.

However if the unit is already damaged from a ground potential difference or even a surge or static discharge from the solar panels running through the Ethernet cable to get to ground correcting the grounding will not repair the unit.

Once again I do not think this is a heat issue so I would do the following:
1) Increase the Ethernet cable run to be at least 10% longer than the intended ground run possibly by inserting a UBNT Ethernet surge protector ( I hate surge protectors) and adding cable length. AND/OR MAYBE INCREASE THE CABLE LENGTHS FROM THE SWITCH TO WHAT EVER DEVICES YOU ARE POWERING FROM THE WS-6-MINI.

2) Make sure that the #6 ground wire the solar installer ran is correctly bonded to the service ground rods either at the panel or the service ground rods and not another ground rod and if it is a separate ground rod from the service ground rod(s) that could be your problem so make sure they are all bonded together at the ground level.

After correcting any grounding issues you could swap out the WS-6-MINI and there should be no other problems.

[sirhc]

Eric and Dave will be investigating the software tomorrow to make sure we are properly dealing with erroneous readings.

[cwachs]

The switch stopped responding via Ethernet again today. Power stays up but no traffic will pass over any port. Same problem we had with 1.4.2. If I plug a computer directly into the switch, I can not access it until I do a hard reboot on the switch. Then it comes back normal and works for a while.

I have had to remove the switch from service since it will not stay up and this is a mini-pop location that feeds other users.

We did investigate the ground. Solar company used the same grounding rod as house electrical panel so no problems there. The only remaining potential issue is length of cable of ground vs Ethernet from the attic.

We added a surge protector (we use the Mimosa NID but nearly the same as Ubiquti) to the line so when we replace the MINI, we can increase the Ethernet length. I'm having a hard time wrapping my head around that being the problem here, though. With as many of these as we have in service, there are certainly others out there that have longer ground than Ethernet cables - I know of a couple for a fact that are that way. Again, house installs (not towers) and they are working perfectly.

You are accurate about the cable type. This install is UBNT shielded cable with a drain wire. UBNT shielded ends with good crimp on the drain. Cable checks out solid.

We put the switch in question on the bench. Powered by same UBNT AF24 injector. We reset the switch to factory defaults and logged in. No ports are powered - it is default all the way. The voltages on all three boards are doing the same thing they were in the attic - fluctuating by 2+ volts for 48 and 24 and around .5 volts for the 3 volt.

We've had this out of the box for 5 days and just bought it. Do we RMA or wait on some software fixes??

[sirhc]

I would RMA it but I am not sure the fluctuating voltage reading is the issue causing the "apparent" freeze?

I still wonder about the network configuration as the cause of the network issue?

Personally when it happened I would have unplugged one device/radio at a time and seen if it resumed (not rebooting the switch)

Then after all devices had been tested I would have rebooted routers and again not the switch to see if traffic resumed.

What devices are powered by the switch?

[cwachs]

I forgot to mention. We changed the network environment yesterday to be more "standard" on this switch. It was no longer doing double duty in two environments with VLANs. That has no effect on the freeze.

The only "odd" thing we have going on is one port is set to 24v and is powering an older ePMP 1000 AP. This AP requires the pair swap. We built a special cable to handle that. AP works with that special cable. We have a handful of these out in the field (special cables) and don't have any issues with them.

The powered devices were:

Mimosa B5 Lite (48v)
ePMP 1000 AP (non-GPS) (24v)
Ubiquiti Unifi-AP (48v) - this was removed when we "normalized" the VLAN situation.

Will start the RMA process. We have had extremely good luck with these switches and bought all but this particular one direct from you. Since you are out of stock, this one came from a distributor. Warranty sticker is intact but perhaps it was a return to that vendor that got put back on the shelf??

[sirhc]

Its possible I suppose?

Another thing recently discovered is epmp 2000 also "appears" to have a Flow Control issues and requires Flow Control to be DISABLED on the port feeding the epmp 20000 radio or it acts "similar" as the AFX radios and can cause the switch to stop passing traffic until you unplug the radio and then plug it back in.

So anytime you find an issue where the switch appears to stop passing traffic unplug cables one at a time until it resumes passing traffic then try disabling FC to all devices and see what happens. If all is fine then enable 1 port FC at a time over days until it faults again.

[cwachs]

I just read that about ePMP 2000 a couple days ago. I had one connected to a WS-12-250 for three weeks with no issues - though I was NOT using the switch to power the radio in this particular case.

I also have a pair of AF24's in our network, both powered by Netonix switches and flow control has been enabled for almost a year on both ends (since I missed that post as well). No issues and the Netonix is powering the radios.

I guess I'll log in to both switches with the AF24's and turn off flow control now...

[cwachs]

I have applied for an RMA on this switch but have continued testing while I wait. I put 1.4.3rc7 on it. Same issues with fluctuating voltages.

However, I learned something new... I could not get a Mimosa B5lite to link up with the switch. It would power the radio but would show a link down indicator in the switch. I put a standard (non-shielded) indoor CAT-6 cable between the switch and the radio. Powers up and links up just fine. I then put a shielded cable with shielded ends (tested good) between the switch and the radio. Radio fires but no Ethernet link. I can repeat this test with any non-shielded cable vs shielded.

I then pulled ground from the switch to test that. No change. Pulled ground on the Mimosa and the switch (nothing grounded other than the POE injector). Still same results (works on non-shielded cable, does not work on shielded).

What is going on?

[highlands]

WS-12-DC Firmware Version 1.4.2 Board Rev D


I have a unit that has been going off-line between 2 and 4 in the afternoon for about a week now. Also we used 3 more ports a week ago which is when this started. We can not figure out why the unit just goes offline. Currently I am assuming it is temperature related due to checking the power at the unit is good? There is no reason code in the log. And no we do not log server we send to so we can not see if the is a log entry for high heat. We do like units that have a EEPROM for log file so that they are there after a reboot. The enclosure that this WS-12-DC is in is;

http://tycononline.com/Kit-Die-Cast-Enc ... p_227.html

I thought I remembered seeing here in the forums that these units were good to 125c but when I just looked at the specs it states -25 to 55c which is a huge difference than 125c. Here is the screen shot of the temps.

[Dave]

that temperature is ok, the spec -25 to 55c is for ambient outside air temperature, the spec for 125c is the internal die temperature of the main SOC