Lightning Suppression

Good old mother nature swept thru and destroyed some of my gear.
Most noticeably a Arecont Outoor IPCam and the POE port on SG300-10p and both uplinks.

Ironically I have all my gear now on battery backup and suffered no power equipment loss.
I have much of my gear double covered UPS'd and have a nice size extra pack on my rack.

But I never spent a dime on ethernet surge protection. I see the ethernet suppressors are not cheap.

What do you use?

I am going to switch to fiber for uplinks for my SG300-10 POE routers, I know that.
Isolate from my top of rack SG550X which was not cheap.
 
For cables that start and end within the house: Nothing. We have a metal roof, so stuff below the roof is unlikely to get a lot of stray fields, unless it is a direct hit (in which case electronic is the least of your concern).

For the phone cable (which carries DSL), the phone company has their standard surge arresters where it enters the house. I then add a surge arrester that's made by TrippLite.

I have one ethernet and phone cable that goes from the house to a "garden shed" (which contains pumps and well control, we have our own water supply system). Distance is about 25m; the connection is completely underground, but in plastic (not metal) conduit. Both buildings are really well grounded (ufer + ground rods), the ground are well interconnected, and both ethernet and phone have TrippLite surge arresters at both ends. I really want to pull the Ethernet cable out though and replace it with fiber ... not for the speed, but for lightning protection.

About 15 years ago, we got a direct lightning hit, but not on the house: On the 12kV power line that feeds our house, second power pole from the house. The hit damaged the 25kVA transformer that provides electricity to our house (had to be replaced), and cut through one of the two 12kV lines (had to be patched and re-hung on the pole). Inside the house, there was very little electrical damage, except for the variable speed control for a 2kW electric pump motor, which was blown out.
 
I am a loyal 30 year APC user. First place I went was Amazon search for PNET1GB. I always pull up the one star reviews. I feel it gives a glimpse of the worse case scenario.

When I looked at the one picture in the bad review I am blown away. Components not touching solder pads?
Granted being so popular probably means lots of fakes. Especially Amazon.
When I look at the rest of the reviews I can help but feel that 20% of them are junk.

So I had to look elsewhere for my personal fulfillment.
$500 worth of gear gone in one brief storm. I need to act with confidence.

I bought a pair of these for testing.
They recommend a separate lightning rod for each box. Keep line to ground short too.
 
The components in the CITEL box look reasonable. But it is strange that the incoming and the outgoing cables share one "entrance" into the plastic box. Any surge on the shielding (if there is any) will likely bypass the PCB and "jump" from one cable to the adjacent one. Regarding humidity one "entrance" is easier to manage. But from the ESD point of view a straight PCB with input and output at opposite walls of the box should be better. May be it could be an improvement to drill additional holes at the left side and the right side of the box to separate the incoming and outgoing cables.
 
Alot seem to use the NID (network interface device) style box with input and output and ground within close proximity.

 
Yes, may be for the sake of simple installation. The second product gives me a link to ebay where the mechanics is "straight trough".
I hope the link works when opened from a computer different from mine.

If you arrange the installation with a single "entrance" to the housing it might be sufficient to separate the two ethernet cables by the ground line. Then any surge should prefer the path to ground. That ground cable should be of low resistance and inductance. Something like a copper braid should be fine as a solid ground line and as a shield between the ethernet cables.
 
You are not wrong. Infact the first thing I wondered is that grommet rubber or plastic.
I was thinking I could make a plastic grommet custom fit for my cables to get more watertight.

That the discharge cable is that close to protection was concerning.
 
I think my POE requirement makes some arrestors not ideal. They have to allow 56V to pass.
This should be ok with the cmj8-poe-a-c5e. They claim a pass through voltage of 70V. But you are right, especially lots of low capacitance ESD diodes are designed for a lower clamping voltage.
 
But it is strange that the incoming and the outgoing cables share one "entrance" into the plastic box. Any surge on the shielding (if there is any) will likely bypass the PCB and "jump" from one cable to the adjacent one.
If a voltage spike "jumps" a cm or two through the air, we're talking dozens of kilovolts. If there are dozens of kV on a grounded shield ... it's game over, the ground connection has thoroughly failed (most likely overwhelmed by a lightning).

My wife worked (a long time ago) at an observatory on the top of Mount Haleakala, in Hawaii. Haleakala is infamous for its lightning storms. And their observatory was right next to a secretive defense installation, which had a megawatt laser they shine into the sky, and that makes all manner of electrical noise. All the sensitive / valuable / custom-built electronics was in a smallish room (perhaps 10x15 feet), which had a handful of electronics racks. That whole room was shielded on all sides with mesh fencing made out of copper wire, which then had solid aluminum sheets placed over it (the copper for good conductivity, the aluminum to be solid without gaps, and to have low inductance). All signals entered and left through overvoltage absorbers using coax cable; the sensitive (low-voltage) absorbers were built using sacrificial varistors, and were known as "bambi boxes" (because they have the same fate as bambi, in the "bambi vs. godzilla" movie).

One time they had a direct lightning hit. It unfortunately hit the power transformer on the power/telephone pole, so all the 120/240V lines had huge megavolt spikes on them. They had an old AT&T desk telephone, which was not recognizable: nothing but plastic melted into the metal parts; it left a burned spot on the wooden desk. Their computer (a PDP-11/70) was not in the shielded room, and it was simply thrown into the trash and replaced: All the serial cables to terminals and modems went right into it, and many chips had blown apart. The power transformer itself (it's a can, about 30cm diameter and 80cm tall) was heavily deformed, with a puddle of melted "goop" underneath it. It was amazingly lucky that the building didn't catch fire. The amazing thing: All the sensitive electronics in the Faraday cage survived undamaged, after the bambi boxes were replaced.

Most people are not in an area that is this lightning prone, so full Faraday cages are usually not required.
 
Here is the old version of PNET1
It's build with 1n4006 (800v) Diodes for ESD protection.

20220404_152414.jpg

20220404_152428.jpg
 
Its obvious that the Amazon review picture is fake. Nothing looks similar.
Nice rugged thruhole components. Just like you would expect from APC.
I guess that is the bad thing about shopping by reviews. You just can't trust them all.
I think it was a fake though because board was marked APC like yours but used all surface components.

I did see a comment somewhere mentioning the APC did not use the glass tube doo-hickey...
That is apparent from your picture.

Do they work with POE?
 
Do they work with POE?
Hard to tell. One would need to examine the circuit diagram.

The picture shows overvoltage absorbers, which are pretty standard Teccor P2600E. Those clamp at +-260V, so the typical PoE voltages (about 48V) should have no problem. Alas, it only shows 6 of these clamps, and an ethernet cable has 8 conductors. That's already weird. It also shows 18 or 19 diodes, and the identification as 1N2006 seems plausible from their size. What do those diodes do? If they are really bidirectional clamps, then they'll hold every line to +- 0.7V, which would kill Ethernet traffic. I suspect that they are not 1N2006, but some sort of zener instead.

To really think this through, one would have to draw a full circuit diagram.
 
Dear VladiBG, thank you for taking the picture.

Q1...Q5 seem to be P2600 from Littlefuse. https://www.littelfuse.com/products...rs/baseband-voice-ds1-protection/pxxx0e_.aspx. Ok, now ralphbsz has already posted :). I have just seen that Teccor has been acquired by Littlefuse.

At least the designers have taken care about signal integrity for some extend. There are at least two traces with zig-zag pattern to compensate the delay of the tracks. But maintaining symmetry with that kind of through hole components is very difficult. The huge single ??5KE?? might be a (special) diode, too. But I have no idea about its purpose.
 
No it's not working for PoE as it's only for 100Base-TX. This is the old version which doesn't include 1000Base-T.
The six thyristors are in series to provide the 0.25kA clamp.
The TVS diode is 1,5ke75a (1500W)

Edit: all 8 pins are protected.
1649142350027.png
 
Placing the ports close to each other is to reduce the crosstalk between RX/TX as it shorten the traces on the PCB. IT's just a new version with SMD sop4 diodes.
 
I ended up with a batch of these for cheap:
Tycon/Laird ESP-100-POE
My cams are only 10/100 so these are fine. EOL but fine.
I have a new digi camera. Pictures in order?
I don't want to litter the forum with junk but the inside view of gear is nice.
 
Well as they say; Pictures or it didn't happen...
 

Attachments

  • POE.JPG
    POE.JPG
    457.2 KB · Views: 89
  • POE2.JPG
    POE2.JPG
    493.3 KB · Views: 90
Back
Top