Something struck me: to protect people, the neutral of the supply is connected on the supplier side to Earth, so that leakage through ground is detected when for example someone touches the phase (residual current devices do that). It doesn't protect people who would touch neutral+phase at the same time though.

But wait a minute, if the neutral wasn't connected to earth (which basically is the case when we isolate the mains supply), there would be no path through earth so no chance of leakage. So why is the supply made like this then, what do I miss?

Figure to illustrate - the dashed line separates the supplier side and the user side. enter image description here

  • \$\begingroup\$ Probably because there are other bad things the output of the 'pole pigs' can short to that would make a mains voltage shock look like a tickle. \$\endgroup\$ – Spehro Pefhany Jan 23 '15 at 23:39
  • 2
    \$\begingroup\$ In any large electrical network there is always leakage to ground. \$\endgroup\$ – Hot Licks Jan 24 '15 at 3:40
  • \$\begingroup\$ @HotLicks When I say leakage, I'm more interested in leakage through people. Spehro Pefhany: I don't really understand (the fact that I don't know what's a pole pig does not help...), could you expand? \$\endgroup\$ – user42875 Jan 24 '15 at 11:20
  • 1
    \$\begingroup\$ As I said below, imagine that, in your second diagram, there are two resistors of relatively small (low ohms) value connecting the two legs of your transformer secondary to ground. This is what you get in practice, if the secondary side of the transformer connects to a network of any substantial size (as with your standard neighborhood pole transformer). (If nothing else, imagine that it's raining.) This creates a substantial (though unpredictable) shock hazard should you touch either line. \$\endgroup\$ – Hot Licks Jan 24 '15 at 13:56
  • \$\begingroup\$ @user42875: google.com/search?q=pole+pig&tbm=isch \$\endgroup\$ – Fizz Jan 24 '15 at 20:20

The electrical system we use was designed close to 100 years ago, and was modernized very little. Grounding and bonding was needed to blow fuses in case of short to an enclosure, or ground. The grounding was a reasonable option then, and it is unacceptable today.

The change to make it safer is not going to happen, because the misconception, that grounding is good an essential. It is too entrenched. Electricians are horrified of any change, and for years they were taught that grounding saves lives.

Grounding and bonding is actually the biggest cause of electrocutions.

But not all the circuits are grounded. Industrial 3 phase delta in North America is not grounded, and is potentially safer.

What we should have had by now is:

Circuit isolated by a transformer, grounded through a high resistance, and monitored for any leaks. This system already exists, but the main intension is not to protect people, but eliminate phase to ground arc.

For example currently the surface of an electric stove is connected to a bonding conductor, which is connected in the panel to a neutral, which goes to a transformer. The surface of the stove is connected to a leg of a transformer. The neutral is grounded at the transformer and at a building, but those grounds are not very good. Typically at least 5 to 10 Ohm. The solid connection is to the transformer.

This is why touching the stove and phase can be lethal.

Here is how it is done now in USA and Canada enter image description here Here is a better solution. A ground fault circuit breaker would be used. A person touching a ground and a live wire would have a limited current running through them, and the circuit breaker would disconnect the power. This system doesn't protect from line to line shock. enter image description here The resistor sensing works on the principle of a voltage drop, the resistor also limits the fault current to a safe level. The ground fault works on a principle of magnetism and it compares current in the incoming wires. If one current is higher, it shuts off.

  • 3
    \$\begingroup\$ +1 I agree. The problem is however not conservatism but cost. Changing the total electrical infrastructure in the way you describe costs tons of money. That's why the existing concept of earthing the system is still the safer one and remains in existence until today. \$\endgroup\$ – Ambiorix Jan 24 '15 at 2:53
  • 2
    \$\begingroup\$ Ambiorix, You are right. Currently new set of rules is applied in North America, making all circuits to be changed to arc fault protected. This will cost billions, and make billions for the manufacturers of those circuit breakers. There will be a limited benefit, but the money would have been better used for modernizing the system. This is already a law for all new construction, and it is a step in a wrong direction. \$\endgroup\$ – sparky Al Jan 24 '15 at 3:34
  • \$\begingroup\$ I have doubts about the general validity odf this answer but have not spent time looking in to it and cannot do so immediately. Note that marine power systems are not (AFAIK) hull grounded. Needless to say. on a boat hull grounding of one or other line happens easily enough. When it does the circuit is lethal from one or other leg to hull. There is a lot of "hull" attached stuff on a boat. Toss a coin to know which leg will kill you. You can shut off the power in that circuit and find the fault or trace it hot - and FAST - ALL the connected circuits are affected if run live. ... \$\endgroup\$ – Russell McMahon Jan 24 '15 at 3:42
  • 1
    \$\begingroup\$ sparky Al: Thank you for your answer. It is not very clear though what should be done and what is currently done, could you update your answer to make it clearer? A "good" schematic VS a "bad" schematic would help, which includes high resistances to ground and leakage monitoring equipment. Because I don't quite get it... \$\endgroup\$ – user42875 Jan 24 '15 at 11:24
  • \$\begingroup\$ There's also the small issues of: (1) getting other countries to manufacture compatible products Reasons for grounding the secondary include lightning and physical damage to, or failure of, the utility transformer. IIRC the utility transformer is being fed around 1,250V. \$\endgroup\$ – Technophile Jan 24 '15 at 17:49

The vast majority of people only get in contact with electricity via an appliance, tool or other device. If you connect the neutral to the earth and earth all the devices a circuit breaker will likely trip if the earth resistance is low enough. If you'd isolate the neutral you would not know unless you touch the neutral or the neutral makes contact with the earth accidentally, in which case you get the shock, not the circuit breaker. The 1st option is still the safer one.

This is also the reason why in many countries the earth resistance needs to be below a certain value before you get certification.

The only exception are double insulated appliances and tools which are build in such a way that the enclosure can never become live. These device are not to be earthed in any way because that would again increase the risk of shock if the earth wire would become live through another equipment that would accidentally leak.

  • \$\begingroup\$ Thanks. But connecting the isolated neutral to the chassis of equipment would make breakers trigger too if there was a fault in isolation, so is it really like it sounds: something wrong that is still used nowadays because it's too costly to change it? \$\endgroup\$ – user42875 Jan 24 '15 at 11:40
  • \$\begingroup\$ That would increase the risk of electrocution tremendously. Imagine if for some reason the phase wire would touch the earth somewhere in the network, and that chance is considerable seen the size of the average area one distribution transformer feeds, the chassis of the equipment would become live without anything tripping. You wouldn't even be even able to solve the problem locally because this earth to live shorting could be located anywhere. You'd figure, ok then you earth the neutral at the feed transformer. That wouldn't work either as you would require plugs to be polarised. \$\endgroup\$ – Ambiorix Jan 24 '15 at 13:32
  • \$\begingroup\$ The only improvement would be the proposal of Sparky Al. This requires an isolation transformer and earth monitoring system in every building or house which is a considerable cost. It also introduces some new problems as someone already pointed out. In fact all these configurations already exist. If you google TT, TN, and IT systems you'll find plenty of documents that explain the pros and cons of all these different network configurations. \$\endgroup\$ – Ambiorix Jan 24 '15 at 13:32

By not connecting earth mains to neutral, you're allowing the wall's line voltage to "float". I.e., although the hot and neutral line will always be 120VAC with respect to each other, there's nothing stopping them from rising above ground potential. The stick figure on the right will get a nasty shock if the line is floating far above earth ground--which is certainly possible (it's happened to me! although the mechanism is different, see my question here: Connecting center-tapped transformer to earth ground; or, why am I being electrocuted? ).

edit: the figure is touching the live wire! I didn't see that. He's going to get zapped even if you connect neutral to Earth. But if you don't connect neutral to Earth, touching it can zap you. Furthermore, most of the things you plug into the wall are "grounded" via the neutral pin. If that's floating, you can get electrocuted just touching your toaster!

  • \$\begingroup\$ If the supply is floating, isn't there going to be only an equilibrating current until the touched wire is at earth potential? It should be like a static shock without much harm done shouldn't it? \$\endgroup\$ – user42875 Jan 24 '15 at 11:26
  • \$\begingroup\$ Interesting link you have there, I don't understand though how there can be a non-temporary current, especially that high! \$\endgroup\$ – user42875 Jan 24 '15 at 11:30
  • \$\begingroup\$ Regarding your edit: even if the figure is touching the live wire there will be no shock because of the supply voltage on that wire, because there is no loop for the current to be closed. As you noted correctly the wires are isolated (floating) therefore no current will flow. (Except e.g. if one person touches the live wire and another thouches the neutral, have fun then... \$\endgroup\$ – Andreas Wallner Jan 24 '15 at 12:30
  • 1
    \$\begingroup\$ @AndreasWallner - You are missing the point that in any electrical network of non-trivial size inherent leakage currents will prevent the system from "floating" in the way you imagine. Imagine two fairly low-valued (but not necessarily equal) resistors running from the two legs to ground -- that is what you get in practice. \$\endgroup\$ – Hot Licks Jan 24 '15 at 13:53
  • \$\begingroup\$ Consider lightning, pole-transformer failure or damage, and falling trees and branches knocking distribution wires onto feeder cables. Transformers, wires, etc. are built from available materials. Insulators fail. What stops the (IIRC) 1,250V distribution voltage from making it into your appliances? Ground the secondary center-tap. \$\endgroup\$ – Technophile Jan 24 '15 at 17:45

You are right that isolation would prevent a shock from directly touching a live wire, but brings with it other problems. This is also something that is used in applications where one needs additional failure safety in case of shorts to ground (no immediate need to switch off in the case of a single fault), but with care that the problems below are handled correctly. Cases this is done are e.g.:

  • in ICUs in hospitals, where most devices are completely insulated anyways.
  • in the chemical industry where a power outage might even be dangerous (e.g. exothermic reactions)
  • areas that need special protection against explosions (e.g. coal mining)
  • on ships
  • sometimes by fire brigades when using portable generators

It is also not done as simple as you drew is, e.g. metal cases are still grounded, and the secondary side of the transformer has a connection to earth, but only in the range of a few kΩ.

Such supply nets are normally limited in area, since finding a fault is relatively hard. Most of the time those are also permanently monitored for isolation (with a insulation monitoring device) as to detect single faults before they cause problems.

The way you drew the system you could have multiple problems (not an exhaustive list):

  • You can not use RCDs to prevent from touching e.g. cases of devices where the live wire is shorted to the case. Which would not be a problem by itself, but becomes one if you touch anything that is grounded (water pipes, heating, being in your garden, maybe even your living room - depending on your isolation to earth in that case) or e.g. another device where neutral is connected to the case (either by design or also by accident).
  • You have to monitor that your net is never connected to ground at any point (either by design (made for a different country) or by accident).
  • You would have problems with the shields of e.g. network cables. You have to connect them somewhere, but you can't connect them to ground anymore. Maybe connect them to neutral? Then you get a compensation current over the shield, because of the different neutral level of each device (because of the return current and the ohmic resistance of the neutral wire)

Your view also ignores capacitive coupling between the power lines.


Every object that can carry a charge is at some potential referenced to earth. Thundercloud tops, for example, can reach a surface potential of over a billion volts. Choosing zero volts as the mean potential really is the safest option; any other voltage would be higher and thus the potential for an electric shock would be worse.

For some related trivia, nowadays three phase electric power is carried to houses preferably in a cable which has five conductors: Line voltages L1, L2, L3; the Neutral and Earth. But back in the old days at the countryside when copper was scarce, there would have been only four: L1, L2, L3 and a much smaller in diameter Neutral, to save material. And Neutral would have been literally connected to the earth, using a copper nail of at least one meter in length, driven into the soil at the side of your house. To my knowledge, this is not done today. But it gives some sense to the term "earth" in electric wirings and sheds some light to the practicality of using earth.

When the neutral is tightly connected to earth, there's another safety plus: when electric cables do break, by the action of a digger or a falling tree, the earth gives a conducting path for the current, which can be detected and fault relays at the power station can switch off the power in the cable. Similarly, if an electric appliance of yours breaks down, there's a good chance that it will blow a fuse in your fuse box if the shield in your appliance is earthed.

  • 1
    \$\begingroup\$ Maybe in your house, but definitely not in mine (in the USA). Here we get three wires: L1, L2 and N, and the ground MUST be connected to a stake at the service entrance to the house, and the neutral wire is also bonded to ground at this point. \$\endgroup\$ – Dave Tweed Jan 24 '15 at 0:13
  • \$\begingroup\$ Oh, ok. I don't live in the US. All new electric connections have been three phase here for the past couple of decades or so; impossible to get a new single phase contract anymore. Damn shame, they were much cheaper... \$\endgroup\$ – PkP Jan 24 '15 at 0:20
  • \$\begingroup\$ @DaveTweed You may be referring to 1 1 phase feed with centre pat - 180 degrees apart vectorially. His is a true 3 phase fee at 120 degrees. \$\endgroup\$ – Russell McMahon Jan 24 '15 at 4:07
  • \$\begingroup\$ @PkP where DO you live. Not NZ as we still have 3 phase. Maybe UK? You may wish to discuss your bizarre affection for solar panels offline sometime. Generators are of interest to me but less relevance. \$\endgroup\$ – Russell McMahon Jan 24 '15 at 4:08
  • 1
    \$\begingroup\$ @DaveTweed " ... you may be ..." in the Benjamin Franklin advised sense :-). I had minimal doubt that you knew what you had but you did not make it clear to others, I think. L1, L2 in the context being used is suggesting "proper" phases, 'which you have not got'. It's 'probably' in the interests of general learning to somewhat spell out such differences. \$\endgroup\$ – Russell McMahon Jan 24 '15 at 12:02

It is very desirable to construct devices such that when they are switched off or have a fuse blow, no part of them will be able to feed current into ground, even if there exist faults within other devices on the same circuit. This effectively requires that all power leads which could source or sink significant current when their potential substantially differs from ground must always be switched or disconnected simultaneously. It is much easier and cheaper to have a hot wire fused and a designated neutral wire unfused than it would be to guard hot and neutral wires with a "double-pole fuse" constructed such that an over-current condition on either wire would disconnect both. Although it's possible to construct fuses in such fashion, such designs are much more expensive than independent fuses.

In the absence of a designated/forced neutral, an appliance which developed a short between one of its power leads and ground could cause the shorted power lead to behave as "neutral" and the other to behave as "hot". If another appliance on the same circuit only switched or fused the same power lead as the one on which the first appliance developed a short, the entire second appliance would be live (relative to ground) even when it was switched off--a potentially dangerous condition.

If all appliances switched or fused both power-input leads, it wouldn't matter which was hot and which was neutral. Having a designated neutral, however, makes it possible to have appliances safely switch or fuse only a single lead, rather than having to ensure a simultaneous disconnect of both leads. That's a pretty big win.

  • \$\begingroup\$ (+1) That's a very good and clear point, thanks. \$\endgroup\$ – user42875 Jan 24 '15 at 22:46

The electrical system is, as you say, old. It is not unacceptable though. (I am in Australia so it is a little different than the US, but not that much in principle.)

The neutral being connected to the earth potential is, as has been explained, so that if there is a short circuit to the ground/chassis/earth in a piece of equipment, the fuse will blow.

Fuses were indeed the only protection we had for many decades. Now, in addition, we have residual current and earth leakage current devices that literally measure the difference in instantaneous current in the active and neutral conductors.

If there is a difference, it means that some current is going through the equipment to ground, like in the case of a refrigerator where moisture is causing tracking of current through dust and dirt to the frame of the fridge. The other possibility is that current is going through a person to the ground through their shoes... and maybe their heart!

That is why this relatively recent innovation is so important and operates in addition to normal circuit protection in the form of fuses and circuit breakers.

In Australia we call RCDs (Residual Current Devices) "Safety Switches". They are exactly that!

That ends the lesson.

  • \$\begingroup\$ Thanks. I know what RCDs are, the thing is that if the neutral is not connected to earth in the first place, we don't need those? Connecting the isolated neutral to the chassis of equipment would make fuses blow too if there was a fault in isolation, so is it really like it sounds: something wrong that is still used nowadays because it's too costly to change it? \$\endgroup\$ – user42875 Jan 24 '15 at 11:39
  • \$\begingroup\$ Grounding the neutral does make the fuse blow. It also means that a short of a hot wire to ground does not raise the other wires to a lethal voltage. \$\endgroup\$ – Technophile Jan 24 '15 at 17:59

Unfortunately none of the answers on this page seem to mention one rather important thing about high-voltage transformers, which makes them fairly cheap/affordable: graded insulation. From what turns out to be one of my favorite books (in answering questions here), Grounds for Grounding (by Joffe and Lock) here's the relevant figure:

enter image description here

Without earthing the neutral, you can't use graded insulation. I've only included here the figure for the initial high-voltage transformer at the generator site, but the same graded insulation applies to all the substation transformers on the distribution network toward the consumer.

EDIT: I guess I shouldn't have left this dangling. Since you can't (cheaply) get rid of earthing on the high-voltage distribution, when you couple an ungrounded low-voltage supply system to it, you have capacitive coupling as mentioned at the end of Andreas Wallner's answer. The effects of a ground fault in a large ungrounded system are more severe than one (leaky) home transformer experiment may suggest. According to IEEE Std 141-1993 you can see transient over-voltages five times the nominal voltage in such a case when a phenomenon called arcing ground occurs. (The same info is repeated in http://www.hv-eng.com/2010IASGrounding.pdf) There's a somewhat intuitive explanation of arcing ground phenomenon at http://www.electrotechnik.net/2011/05/arcing-grounds.html I'm not really sure about the reliability of that latter source, but it says basically that the arc forms and breaks up many times that's why you get the overvoltages (said to be 3-4 times the nominal in that source). There are ways to prevent those, but an ungrounded low-voltage distribution isn't as simple as an isolation transformer in a Class II wall-wart.

EDIT2: The IAS presentation actually explains the overvoltage due to arcing more scientifically in terms of a resonant circuit that forms. It's actually a very good presentation, but rather long and my EE time today was rather limited... Towards the middle (of the 100+ slides) it gets to talk about high-resistance grounding (HRG) which is a way to get some benefits of no grounding (reduced risk of shock) without much of the downsides, but alas it's only applicable to three-phase loads; HRG can't be used with single-phase loads. So we probably won't see it in residential distribution.

  • \$\begingroup\$ My answer covers another aspect which I didn't see mentioned: it's extremely important to ensure for safety that any wire that could source or sink significant current to/from ground be fused, and fairly important to ensure that the act of disconnecting any such wire will disconnect all of them. It's possible to construct double-pole fuses (such that when any leg blows, all will disconnect) but they're a lot more expensive than ordinary fuses. \$\endgroup\$ – supercat Jan 24 '15 at 20:26
  • \$\begingroup\$ (+1) Thanks, very informative links in particular. \$\endgroup\$ – user42875 Jan 24 '15 at 22:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.