CAT6 STP Ground Loop Input

[clanderson]

I'm very eager to hear everybody's thoughts on an issue I'm working on. I work as a WISP engineer for a company who recently merged with a much larger company.

Our bread and butter involves one of two scenarios:
A.) Deploying equipment on a structure not designed to be host to such things and consequently is not conducive to protecting these sensitive devices.
B.) Deploying equipment on a structure designed specifically to host microwave equipment but which consequently introduces massive amounts of EMI/RFI.

The equipment I speak of consists mostly of several wireless radios at the top of a structure and an ethernet switch at the bottom. We've been having a monstrous amount of equipment failure surrounding scenario A lately, and I suspect it's due less from near or direct lightning strikes and more from static build up/discharge and ground voltage from improper grounding. More specifically potential differentials introduced through grounding loops.

Our microwave equipment topside is chassis-grounded to the tower (or grain leg, silo, pole, grain bin, etc.) which almost certainly has an incredibly high resistance path to earth. They then run CAT6 STP down to our enclosure which houses a switch. This switch is chassis grounded to the electrical earth ground (usually nothing more than a single solid #4). At the insistence of many different vendors, my company has always grounded both ends of our CAT6 STP. This is counter to my understanding of EE, which suggests this will do nothing more than create a ground loop, considering the other adjacent paths to ground.

We are in need of some serious procedural reorganization, and I'd like to be very clear with my expectations. I believe we first need an isolated low resistance path to earth ground run up the tower, to which our radio chassis can be bonded. Second, I believe we need to eliminate the CAT6 STP grounding topside, and let the CAT6's shield serve as a drain for EMI/RFI, grounded only at the bottom switch enclosure, bonded with the switch chassis ground. Third, I believe we need a significantly lower resistance path to ground at the enclosure level, which should then be bonded to the electrical service ground.

Am I correct in my assessment?

[tepalia02]

I think you can get answers if you post your topic to a microwave Engineering related forum like this: https://www.microwaves101.com/
Here is also an old thread related to STP cable. Don't know it is of any help to you: https://www.eevblog.com/forum/networking/preventing-ground-loops-on-shielded-twisted-pair-(stp)-cabling/

[madires]

Yes, a proper ground/earch connection is important. Any electrically floating metal structure asks for trouble. However, there are two things to differentiate. One is grounding/earthing meant for lightning protection and the other is grounding/earthing for electrical safety and EMI suppression. Based on the specific installation they can be simply the same (using one ground wire) or different things. It's more complex than most people think. And for Ethernet I'd recommend to switch to glass fiber.

[nightfire]

Grounding concepts are also often misunderstood in datacenters, and cause a lot of trouble.
Basically you have to possibilities to do that:

star or mesh topology.

Star has the disadvantage, that you really need to be able to control what is going on regarding grounding, and if there are any connections between grounded devices, you have nasty loops.
Or everything is grounded with short and low impedance paths to each other to prevent the buildup of high currents that can (and will) flow.

Another idea could be the use of fibre cabling to those devices, to have galvanic separation of the data line. Downside: No PoE power supply available from the switch.
I already used this technique to network some office buildings for about 200 meter distance to prevent (because of different earthing points for the buildings) currents to flow.

Most switches can hold some gigabit/10gigabit GBIC or SFP, and on the side of your microwave equipment you could use some external fibre-to-RJ45 converter. One more single point of failure, but eventually something to prevent other nuisances.

[PKTKS]

SUGGESTED TO MOVE TO NETWORKING

Brazil is perhaps the top place for lightning storms..  this is a recurrent post here

https://under-linux.org/forum.php

Yes grounding is the top concern on such deploys and there is no silver bullet for that

The problems just get worst with collocation tower crowd

You can try a search on these there.. you will find a lot of similar issues and ideas

Paul

[ejeffrey]

Generally STP should be grounded to the device chassis at both ends a shield grounded at one end is not a shield but an antenna.  If ground loop problems prevent the grounding at both ends then it's often better to just use UTP or (much better) switch to fiber.

Ordinarily, ground loops via STP are not a big problem.  The signal wires themselves are differential and transformer coupled, and for indoor runs of moderate length the loop currents do not amount to enough to cause a major problem.  However for this kind of harsh environment it can be a problem and then fiber is the best option.

It's also good practice to measure your ground leakage currents with a DC/AC clamp meter.  If you have high currents flowing through conductors that are supposed to be shields that may indicate a problem that should be fixed