You are here

Surge Protection

13 posts / 0 new
Last post
AE6XE's picture
Surge Protection

I'm looking around to select options for cat5 Surge Protection--hardened tower site not easily accessible.   Anyone have any particular cost effect devices to recommend?  I see in earlier posts in our community people have recommend.




K5DLQ's picture
I'd opt for the polyphaser.

I'd opt for the polyphaser.


WU2S's picture
Weatherproof surge protector

The L-com is weather proof and the Polyphaser is indoor only.

An alternative may be the Everstar.

how does it work?

How does surge protection work on an Ethernet link? Ethernet devices use a transformer on each pair for galvanic isolation, and I seem to recall that their insulation is rated for about 1-2 kV. PoE on the Ubiquitis use two spare pairs so its switching power supply would have to maintain galvanic isolation. I don't know if it does this, though.

Also, the Ubiqiti units are specified to use shielded twisted pair, meaning there's a DC path between the unit's frame ground and that of whatever is supplying it power. The shield shouldn't carry any current under normal conditions.

Bottom line, I don't know how this affects surge protection, e.g., from a nearby lightning strike. I'd be curious to know.

AE6XE's picture
Here's the schematic of one

Here's the schematic of one that sheds some light on what these things are doing.



Image Attachments: 
Thanks, that's just what I

Thanks, that's just what I was looking for. It's reasonable to assume the gas tube thresholds are above anything you'd intentionally put on the cable, but do you know what the threshold voltages are on the diode clamp array? Since you're already using them I assume they're better than 24V, but I wonder about 48V.

LCOM offers many models, all
LCOM offers many models, all have data sheets with a max voltage and a clamping voltage. The one Joe link clamps data pins at 15v and POE at 58v, while the model I have is 80volts for all pins and 90 for the SHEILD and is rated at up to 60v POE voltage (hence higher clamp voltage ) while at the same time they also offer 24v models that would NEVER handle 48v's POE. It all depends what you buy (LCOM data sheets are great on providing specs)
AE6XE's picture
I'm new to using these, but

I'm new to using these, but surely if the specs claim full 802.3af or 802.3at standard, it must support 48v.  

K6AH's picture
These are the L-Com

These are the L-Com protectors I've spec'd for Mt. Laguna... a particularly nasty place for the monsoonal storms we get up from the Gulf of California into the east-county mountains.  In fact, the site owner mandates lightening protection at the bottom (or perhaps it's top) of every vertical cable drop and again at the entrance to the equipment building.  I opted for the screw terminal version because the UBNT Carrier ToughCable isn't fun to terminate.


NOTE: Make sure you tie the

NOTE: Make sure you tie the shield of the cable to GROUND because if notice those connectors  of the unit you posted they are NOT shielded connectors

I've gone with the LCOM (I actually went with the model that had the POE injector built in since I was feeding from an external battery bank ) CAT5E models

In addition the unit I went with the ground shield is only pulled to ground at high potential, so I actually used the ToughConnector with an Aux drain wire and took that to the ground terminal as well  because I want the sheild to always bleed residual current build up BEFORE it hits surge potential and Ubiquiti warranty. REQUIRES the shield be pulled to ground always so the small drain handles the small atmospheric build up and the gastube inside will take the main strike on a better course.

Phil: ya 1-2kv isolation isn't much when a lighting strike comes up, its enough to help protect but not enough for strike survival. 

Reason for surge suppression:  Your not putting it in to protect the radio, your putting it in to protect everything else inside the communications bunker, ya its sad I know, ive seen deployments docs where the radio value was 1/4 the cost of the sure suppression gear going in because the gear in the bunker was worth a pretty penny.

Thanks. Do you try to

Thanks. Do you try to maintain a single common ground point? If so, where?


This is going to depend a lot

This is going to depend a lot on your site.

In my case the entire facility (ground, tower, ground radials, DC grounds for radios, coax shields, etc) were all bonded together across the site.  You can't have a single ground point here but everything is 100% bonded. 

This does open an interesting point about ground loops. The shield pins on the CAT5 are tied to the device ground internally and this is in turn tied to connector shields, if your antenna is grounded you very well may have some grounding at the tower (if well connected to the tower and not isolated by paint or something or non conductive mounting)

RF and Datacom camps have different opinions on ground loops,  RF says avoid them, datacom says ground both sides. It's a really fascinating argument that sometimes just plain comes down to code or personal desire.

I heard from somone the other day that one site they wanted actually requires bonding the cable sheild at every 90 degree crossing (which I would think would be ground loop potential central but that's the sites requirement as I understand it)

Back to the actual status this means you may be poorley grounded at tower, what you do at the ground may indeed create an actual loop circuit on ground.

In my case we grounded the SHEILD to the ground pin of the case with the drain wire at the same place the lighting ground was (same mounting bolt) and the case of the suppressor was drilled into metal plate (so entire case is in compression contact with metal base plus drilled holes were a secured bond) where our polyphasers terminated for RF leads which is in turn bonded to the site ground system via 3" copper strap.


I should note some gear pulls the sheild to ground already so the extra static drain wire I put in at the suppressor may not of been necessary as it would of been drained at the device down stream that is attached to the system ground. In my case I intended for it to often run unconnected and to not have an assurance of a grounded device down stream (such as a Raspberry Pi on USB cable from wall wort)  this is another part of the "shielded cable" ground system to think of.






ground loops

Physics is not really a matter of opinion, but there are enough unknowns about some future lightning strike that it's hard to know the right thing to do.

With the kind of current that flows in a lightning strike (upwards of 50-100kA for big ones) there is simply no such thing as a "low resistance" conductor, so there will always be an IR drop, quite possibly a big one. Even if we had room temperature superconductors, the inductive reactance would still cause voltage drops. Resistances and inductances in parallel both go down, so multiple paths usually don't hurt. But that IR (or IZ) drop will never be zero.

So it's simply impossible to keep everything at ground potential during a lightning strike. Even ground won't be at "ground" potential because large currents will be flowing through it. Fortunately, it's not necessary. The idea of a single-point ground is to ensure that all your equipment potentials rise and fall together. Then you won't have harmful currents flowing through them and their interconnects.

You first have to think about where all the currents go. Find every possible path to ground, not just the one with least resistance, because the lighting certainly will. Assume there will be a large, random IR voltage drop along each segment. If voltage differences between any nodes would cause any damage, you're vulnerable.

Also, any conductor carrying lighting current will develop a very powerful magnetic field and enormous mechanical forces that can tear the cables away from their mounts. Parallel currents will attract, anti-parallel currents will repel, and each conductor's magnetic field will try to straighten it. This is why you avoid bends in lightning grounds.


Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer