Neighbourhood Wiring and Safety Issues

So strange kamaguza ethernet hubs/switches/RJ45 things do indeed have excellent isolation, on every input/output.

Providing the difference is not major, the isolating transformers in the network cards should handle it - they're tested to around 1kV if I remember correctly, and there's often a 500-1kV spark gap too. Lightning is considered "major" though ;-)

I worked in the heavy electricial industry for years. Lightning means overtime. If lightning hits the ground, around one time out of three it goes to ground through a high tension power line. If it hits a power line, it knocks out a bunch of high tension fuses. So you spend hours driving around replacing high tension fuses. Easy work and lots of it.

Most 11kV/400V transformers have these spark gap arc horns to divert lightning. The lightning jumps across the air gap and goes to ground rather than going through the transformer. It's reasonably rare that lightning takes a transformer out, although the HT [high tension] fuses come before the spark gap, which is why they blow. So while you definitely get spikes on the low tension side, they tend to be of very short duration, though the voltage can be up to six or eight hundred volts.

Now, if you get a lightning strike on a low tension line [the segments which go to your house], then kiss your appliances goodbye. I've seen it. Everything in the house that is plugged in gets shagged. And your insurance company says "Act of God, tough luck mate." (Always a good idea to check to see if you're covered for lightning strike.)

High tension lines also have the occasional lightning arrester, which when blown apart seem to be made up of lots of disks stacked up into a cylinder.

Lightning does raise the ground potential at the instant of strike, and is something to be aware of. There are cases of cows being electrocuted because lightning struck nearby and the difference in potential between the ground at their front legs and the ground at their back legs was enough to electrocute them.

So we have this situation where you string a cable from one house to another, is it safe or not?

All right. If the two houses are on different supply transformers then don't do it. The reason is this: your cable can carry current caused by differences in ground potential between the two sites. Each transformer has its own earth mat, but the only electrical connection between the two transformers is the ground itself. Your CAT5 can be a better conductor than the ground. Under normal conditions you may get away with it, under fault conditions your cable may heat up until something melts.

OK, two houses on the same transformer are already connected together - by the neutral conductor. So stringing a cable between them is a bit safer. You are paralleling a big fat wire with your little thin wire. However, you can run into trouble like this:

In one of the houses the hot water cylinder gets replaced. The plumber disconnects the main earth conducter from the copper pipes and doesn't reconnect it, leaving the house with no bond to earth. A tree then falls through the overhead line, taking out the neutral, but not the phase. So electricity can get into the house, but has no return path. If you had strung a bit of coax between this house and the next for your 10base2 network, then you're in trouble, because the shield on the coax will try to carry the whole return current for the house. (Actually, you're in trouble anyway, because every metal appliance in the house would bite you if you touched it.)

Now your 10baseT network should actually put up with this problem because of the 2kV isolation. But if it breaks down at one end then you get the full voltage at the other end. The NIC at the other end may put up with this, but it is not good for you if you unplug the cable and touch the end.

Under normal conditions, stringing CAT5 a short distance between 2 houses on the same distribution transformer would be fine. Under abnormal conditions expect trouble. Plan for the abnormal conditions.

> How much of your work was "normal conditions"?

That's a bit of a moot question - if you're working 'faults' then abnormal conditions is what you see all the time. A better question may be "out of a hundred houses, how many will have a serious problem regarding electricial supply in a one year period?" Not many.

Working under normal conditions at the moment, seen two abnormals in the last month - one was a burnt out mains cable where it terminated at the isolator on the meterboard, the other when working on a switchboard with old cabling with the bare earth wires and seeing enough difference in potential between the earth wires to cause little arcs where they touched.

> How many of the abnormal conditions could not have been foreseen?

I would say 99% of abnormal conditions can be foreseen, and the way they can be foreseen would be to figure out all the things that can go wrong: cars running into poles, trees falling through power lines, snow snapping lines, high wind banging lines together^*) etc, etc. The problem is that such conditions are statistically unlikely and so are not really taken into consideration. And then they happen.

^*) I remember one time when it was a bit windy and the power to the whole town on Amberley kept going out - got sent up to check it out - had a look at the Amberley end, looked ok, drove up to Waipara, looked ok. So I say to the guy I'm with, "Shall we follow the lines across country?" He says, "Nup, we'll sit here for a bit," and pulls out the paper and starts reading it.
Couple of minutes later there's this enormous ZAAAP! and the sky to the right lights up. "There it is," he says. The wind was blowing the 33kV line up into the earthing conductor that was strung along the top to attract lightning away from the lines.

Generally the CAT5 cable will not present a risk under abnormal fault conditions IF it is plugged into a NIC at each end AND by fault conditions we mean 400/230 volt supply.

For high tension faults you can figure out the volts per meter back to where the HT is earthed (where it is supplied from, which is normally lots of kms away) and get a rough idea if it is safe or not. E.g., if this 11kV line comes down, it has 6.6kV to ground, it is supplied from the substation 4km away, so that's 1.65kV per km, so 300 meters of CAT5 will have 495 volts difference between each end IF each end were connected to ground (which it isn't). You also have to take into consideration the breakdown voltage of the insulation on the CAT5.

When thinking about lightning, well if lighting wants to go somewhere, it goes somewhere.

Personally, I've seen 11kV get through onto the 400/230V once. 11kV to ground is more common. Lightning strike onto HT is common. Lightning strike onto 400/230 I've seen once. Lightning definitely hits things like wires and trees and the ground.

Ooh! This happened to someone I know. He was talking on the phone, with the phone against his ear during a big thunderstorm. He was leaning against the chest freezer in the shed. Leaning with his hip. A shed with a wet floor. So the chest freezer was ground.

The voltage from the lightning strike went into his ear and out his hip, leaving him deaf in one ear, and with severe burns on the side of his head and hip.

If he'd been leaning against the chest freezer with the hip opposite to the ear he had the phone held to -- the zap would have gone through his heart, and most likely would have killed him. It nearly did anyway.

I knew the boy, and the boy's mother. The nurse that took care of him in the burn unit was my aunt.

During a thunderstorm, I answer the wireless phone! Safer!

I've had more than one UPS fuse zapped by lightning -- better the UPS fuse than every computer in the house!

At the mining site where I work, we run a COAX ethernet LAN which passes from the factory to the office via a 4" steel pipe, buried at a depth of approx 1.5m. Cable run underground is in the order of 10-15m.

Being on top of a hill and surrounded by iron-rich rock, we get more than our fair share of CG [cloud to ground] strikes up here and have had two close strikes in the last 3 years that have affected the cable run.

Strike #1 was approximately 1km away. Induction(?) popped both NICs and smoked the mobo's at either end of the buried line, all other PC's (with unfiltered power then) on site were fine.

We fitted Black Box (no endorsement to be implied) coax arrestors at either end after this event.

Strike #2 was a direct hit (<500m) Both surge arrestors copped the lot, but we only lost one NIC, we did however lose multitudes of other gear.

$25,000 later we now have site-wide UPS.

My personal machines run CorCOM surge suppressors. Two per machine, in series. We have had several CG's within 1km of the house over the years and have escaped damage. The latest was yesterday, with a strike just across the road within 150m of the power and phone lines. I lost 3 out of the 4 suppressors, but the machines are still intact. The strike was hefty enough to cause arcing in the domestic wiring, detonating a lightbulb and leaving that heady aroma of scorched electronics and ozone that says "missed you, this time". Only loss was a modem which died in my arms as I pulled the Telecom plug - Having felt the current go through as I yanked the wire, I now have relays fitted - that was one scary sensation!!

The power surges and spikes caused by thunderstorms are not necessarily limited to direct lightning strikes. Water can get into the neighborhood transformer, wet tree branches or bamboo can blow onto the lines shorting out the whole valley, and lightning+water in the transformer can cause it to explode. That's the green flash that lights up the whole valley prior to it going totally black. The power surges are spectacular during these events.

Even a surge protector with a resettable fuse (many don't have that anymore!) is better than nothing.

If you're short on cash, get one of those first, because you should use them anywhere you're using electronic gear -- your stereo and TV for instance.

The deal is, when the voltage goes way too high, the fuse blows, cutting off the circuit entirely before the zap gets to your gear. Most surge protectors just clip out the highs, but don't really do signal conditioning -- they don't protect you from erratic signals or *under* voltages (brownouts) either.

A good UPS does a whole lot more. Basically, a UPS keeps a battery charged up all the time, and leeches a nice conditioned power signal off the battery. So it should be providing superior signal conditioning well as a little bit of time to flush your buffers and take down your computer systems gracefully during a power outage. If it gets too big a zap, it blows the fuse where it's plugged into the wall similar to your surge protector -- but the signal conditioner should continue to pull nice clean power off the battery during this event.

When you've finally got your UPS you can move the old surge protector to your monitor, and leave the monitor off the UPS, for example. This just means you're drawing less power off the UPS during a power outage, resulting in less downtime on the CPU. And providing *some* protection for your monitor.

A good UPS can also aid you in power managment. You can connect the UPS to the CPU via the RS232 port, and program it to tell the computer to shut down nicely when it sees that there's an outage. This is known as APM, Automatic Power Management.

The old apmd on linux actually destabilised the kernel on SMP machines, but there are newer methods and utilities out there (another topic!).

You can also get additional battery backup to add to your basic UPS if you've got several machines, and have a master CPU running the apmd that can then tell all the other machines to shut down, rather than having a separate UPS for every machine.

In a small town in the New Mexico high desert at ~2300m I was walking down the street (like lucky La Rue) and one of those sudden thunderstorms kicked up. I was in town, not out in the middle of a golf course, so I figured it would be safe to put up my umbrella. Saw a few lightning strikes nearby, so I hold the plastic handle and note that I'm wearing rubber-soled shoes.

Should be safe, right? Wrong!

Hair stood on end and was suddenly surrounded by a ball of St Elmo's Fire that seemed to go on for several seconds.

The umbrella ribs were a great conductor, especially as they were streaming water down to the ground.

What could I do? Take the umbrella down, and have it go through my *head* instead? I took short but very quick steps to duck into the nearest doorway--*then* took down the umbrella.

A workmate who saw me from a passing car said I looked like I was in an electric blue birdcage.

I had gotten a sensation of mild electric shock through my whole body, but I was basically unhurt.

Unfortunately, I did not emerge with super-powers.