How does an earth fault find its way back?
The earthing system of an installation says how the supply is connected to earth and how the metal parts of the installation are connected to it. It decides what path an earth fault current takes back to the source, how large that current is and, as a result, which device can switch the supply off in time. The same device that protects well in one system can fail to operate in another, so the earthing system is the first thing to know before choosing protection. This guide explains the letters TN, TT and IT, the one condition each system must meet and the disconnection times.
This is educational material. The requirements that apply to an installation come from the standards and national rules in force and from the manufacturer’s data.
What an earthing system does
When the insulation of a device fails, the metal enclosure can become live. Anyone who touches it and the ground at the same time carries part of the fault current. Protection against this works by switching the supply off fast enough, before the voltage on the enclosure becomes dangerous for long enough to harm a person. How fast the device acts depends on how large the fault current is, and that depends on the path the current takes back to the source. The earthing system is the description of that path.
Whatever the system, one rule is the same: the metal parts of the installation and the metal services that enter the building, such as water and gas pipes, are connected together by main equipotential bonding, so that they rise and fall in voltage together.
Reading the letters
The code in IEC 60364 has two or three letters. The first letter describes the source: T (from the French terre) means that one point of the source, usually the neutral of the transformer, is connected directly to earth, and I (isolated) means that no point is earthed or that one is earthed through a high impedance. The second letter describes the metal parts of the installation: T means that they have their own earth connection, independent of the source, and N means that they are connected to the earthed point of the source. In TN systems a third letter tells whether the neutral and the protective conductor are separate (S) or combined (C).
- TN-S: neutral (N) and protective conductor (PE) are separate conductors from the source to the loads.
- TN-C: one combined conductor, the PEN, does both jobs along the whole route.
- TN-C-S: the PEN of the network is split into N and PE at one point, usually at the supply point, and the two stay separate after it.
- TT: the installation has its own earth electrode, separate from the earthing of the source.
- IT: the source is isolated from earth or earthed through a high impedance, and the metal parts are earthed locally.
TN systems
In a TN system the fault current returns through metal: from the faulty enclosure along the PE conductor (or the PEN) to the neutral point of the source. This path has a low impedance, so an earth fault behaves almost like a short circuit and the fault current is large. A circuit-breaker or a fuse can then do the job of disconnecting.
Zs is the impedance of the fault loop (source, phase conductor and protective conductor), Ia is the current that makes the protective device disconnect within the required time, and U0 is the nominal voltage to earth, 230 V in a 230/400 V system. The condition says that the fault current, about U0 divided by Zs, is at least Ia.
In TN-C-S the combined PEN conductor is split into N and PE at one point. After the split the neutral and the protective conductor must never be joined again, and the PE must not be used to carry load current. The PEN conductor itself is at least 10 mm² in copper or 16 mm² in aluminium (IEC 60364-5-54), and it must be mechanically protected and never interrupted by a switch, because a broken PEN can lift the metal parts of every installation downstream to a dangerous voltage.
TT systems
In a TT system the earth electrode of the installation is separate from the earthing of the source. The fault current has to return through the ground, so the fault loop contains the resistance of the installation electrode, the resistance of the source earthing and the soil between them. Together these are typically tens of ohms, so the fault current is small. With an electrode of 20 Ω the fault current is about 230 / 20 = 11.5 A, far below the 80 A at which a 16 A type B circuit-breaker trips magnetically. The circuit-breaker alone would not disconnect, and the metal part would stay live. This is why a TT installation depends on residual current devices.
RA is the resistance of the installation earth electrode with the protective conductors, IΔn is the rated residual operating current of the RCD, and 50 V is the conventional limit for the touch voltage. For a 30 mA RCD the limit is RA ≤ 50 / 0.03 ≈ 1 666 Ω, for 300 mA it is about 166 Ω and for 500 mA it is 100 Ω. In practice the electrode is kept far below these limits, because the resistance of the soil changes with moisture and with the season.
Every part of a TT installation that is protected by the same RCD is earthed to the same electrode. A second electrode that earths the neutral inside the installation turns the system into something else and must be avoided. The RCD guide explains how the device works.
IT systems
In an IT system the source is isolated from earth, or earthed through a high impedance. A first earth fault finds no low-impedance return path, so the fault current is very small and no dangerous touch voltage appears. The supply does not have to be disconnected, which is why IT is used where continuity matters, for example in operating theatres and in some process plants. The first fault must be signalled by an insulation monitoring device (IMD) and located and repaired.
A second fault on another phase is a different matter: through the earthed metal parts it becomes a short circuit between two phases. The system then behaves like TN (when the metal parts are connected together) or like TT (when they are earthed separately), and the protective devices must disconnect it within the time required.
Disconnection times
The maximum disconnection time is set by IEC 60364-4-41 (clause 411.3.2.2 and Table 41.1). The values below are for a 230/400 V supply with 230 V to earth. The shorter time for final circuits recognises that handheld equipment and socket-outlets are where a person is most likely to be in contact with the fault.
| Circuit | TN | TT |
|---|---|---|
| Final circuits up to 32 A | 0.4 s | 0.2 s |
| Distribution circuits and circuits above 32 A | 5 s | 1 s |
In a TT installation the RCD is usually what disconnects, so its own operating time counts. These are the times of the IEC text. The national text that applies to an installation decides.
Comparison and common mistakes
| System | Fault path | Fault current | Usually disconnected by |
|---|---|---|---|
| TN-S | Through the PE conductor | Large | Circuit-breaker or fuse |
| TN-C-S | Through the PEN, then the PE | Large | Circuit-breaker or fuse |
| TT | Through the ground | Small | RCD |
| IT | First fault: no loop | Very small | Signalled by an IMD; second fault by the protective devices |
- Splitting the PEN downstream and joining it again, so that load current flows in the protective conductor.
- A second electrode that earths the neutral inside a TT installation.
- Mixing two earthing systems in one building without separating them.
- Relying on a circuit-breaker alone in a TT installation.
- Forgetting the main equipotential bonding.
Worked example
TN. A 16 A type B circuit-breaker trips magnetically at up to 5 × 16 = 80 A, and a type C at up to 10 × 16 = 160 A. For the circuit-breaker to disconnect quickly, the loop impedance must be at most Zs = 230 / 80 = 2.875 Ω with type B, or 230 / 160 = 1.44 Ω with type C.
TT. With an electrode of 20 Ω the fault current is about 230 / 20 = 11.5 A, only 14 % of the 80 A the type B needs, so the circuit-breaker does not trip. A 30 mA RCD needs RA ≤ 50 / 0.03 ≈ 1 666 Ω, and 20 Ω is far below that, so the RCD disconnects the fault.
The example uses only the numbers above and neglects the other impedances of the loop and the tolerance of the trip thresholds. It shows the order of magnitude, not a design.