Earthing and Bonding a Service

Two Jobs Sharing One Green Conductor

An earthing arrangement is the one part of an installation nobody can judge by looking at it. A rod driven to the shoulder, a clamp torqued and taped, a conductor two sizes above what anyone would question — all of it can be immaculate and still be worth nothing, because what the arrangement is actually worth is a number in ohms, and that number depends on ground you did not choose and cannot see.

Two separate duties get collapsed into the one word. The first is the fault-clearance path: the metallic route by which a line-to-earth fault returns to the source, carries enough current to operate a protective device, and does it inside the time the code allows. Where the supply's neutral and the installation's earth are joined by the distributor, that path is copper and steel the whole way home and barely touches the soil at all.

The second duty is the earth reference — tying the installation to the general mass of earth so that exposed metalwork sits near the potential of the ground people are standing on. This is what an electrode does. It gives lightning and surge energy somewhere to go, it holds the system voltage steady with respect to earth, and on a supply where the distributor provides no earth it becomes part of the fault path too, which is the moment its resistance stops being an academic figure.

Confusing the two produces the standard misdiagnosis. A circuit trips, somebody drives a second rod, nothing improves, and the ground gets blamed. On a system whose fault path is metallic, that electrode was never carrying the current in question. Work out which duty you are trying to improve, and only then decide whether more copper in the soil is any kind of answer.

Soil Is the Component You Did Not Get to Choose

Resistivity is the property everything downstream hangs on, and it swings across three orders of magnitude between the ground on one site and the ground on another. Moisture content dominates. Dissolved salts come next. Temperature matters at both ends, since frozen soil behaves electrically like rock and baked clay in August behaves like gravel. Compaction counts as well, which is why a rod driven through loose fill and one driven through undisturbed subsoil a metre apart give readings that look like different sites.

On any job where the answer carries weight, measure it with a four-pin array before designing anything. Four electrodes in a straight line at equal spacing, current injected through the outer pair, potential read across the inner pair, and the spacing swept from small to large so what you end up with is a profile through depth and not one figure. The wide-spacing readings are the ones that govern a driven rod, because a long rod averages everything it passes through on the way down.

Seasonal swing is the part most often designed out of existence. A reading taken in March after a wet winter is the best number that site will ever produce, and quoting it as the site's resistance is a form of optimism nobody will remember making. Design against the worst month the location gets — a late-summer drought, or frost penetrating to the depth the electrode occupies — and treat the commissioning figure as a datum for later comparison.

Made ground deserves its own heading. Demolition rubble, ash, crushed concrete and imported fill are electrically erratic and frequently aggressive to copper. Where the top few metres are fill, an electrode that punches through into native ground beneath will beat a longer one that never leaves the rubble, and the conductor crossing that fill wants protecting.

One habit worth keeping: leave a driven rod for a day or two before taking the reading you intend to write down. Soil disturbed by the hammer settles back against the steel over that time, and a resistance measured within the hour of driving comes out high by a margin that has talked plenty of people into a second rod they never needed.

Rods in Company Interfere With Each Other

Two electrodes close together do not halve the resistance. Each sits inside the other's zone of influence, both competing for the same soil, and the pair behaves like one slightly larger electrode with a modest improvement to show for the second rod. Separate them by at least their driven length and the interference falls away; separate them further and the result creeps toward the simple parallel arithmetic everyone assumed at the start. Which is why codes that call for a supplemental electrode usually attach a minimum separation to the requirement, and why a second rod driven a spade's width from the first — because that is where the trench already was — meets the letter of a rule while delivering almost nothing the meter can see.

Where several electrodes are used, wire them so each one can be isolated. A test clamp or a disconnecting link at every electrode turns a periodic verification into a ten-minute job. Without them, the only measurement available is of the whole system in parallel, and that measurement cannot tell you one rod has corroded off its clamp underground.

Adding electrodes is not the only lever available. In genuinely hostile ground, chemically enhanced electrodes and conductive backfills — bentonite, graphite-based compounds — reduce the contact resistance between the metal and the soil around it. They depend on moisture and they need revisiting, so they suit installations that will actually be maintained.

Electrodes You Do Not Drive

On most new buildings the best electrode available is already in the ground before an electrician arrives. Steel reinforcement encased in the concrete of a footing, bonded through its length and brought out to a connection point, presents an enormous surface in intimate contact with damp concrete — and damp concrete is a fair conductor in its own right. It costs a length of conductor and one coordination conversation with whoever is pouring.

The coordination is the entire difficulty. The connection has to be made and inspected while the steel is still visible, the tail has to leave the concrete somewhere the switchgear can reach, and it needs protecting where it crosses the damp-proof membrane. Miss the pour and the option has gone for the life of the building.

Ring electrodes and radial conductors answer ground that refuses a rod. A bare conductor buried around the building perimeter at a workable depth behaves as one very long horizontal electrode, and in shallow soil over rock it will beat anything driven. It also hands you a route for tying outbuildings, masts and metal fencing into a single system instead of several.

Metal underground water pipe was the default electrode for decades and in older stock it is frequently the only one present. It is vanishing under plastic replacement mains, often with nobody telling the electrician, which converts a compliant installation into an unearthed one in silence. Where a code still permits its use, the permission arrives with conditions on the length in contact with earth and on a supplementary electrode alongside it.

Measuring It, and the Three Ways a Reading Lies

Fall-of-potential is the reference method. Inject current between the electrode under test and a remote current probe, read the potential between the electrode and a probe moved along the line between the two, and plot resistance against probe position. What you are hunting for is the flat portion of that curve, where the two zones of influence no longer overlap. The single reading taken at roughly sixty-two percent of the distance is a shortcut, and it only works when the plateau is genuinely there. A plot with no flat section means the current probe is too close — on a site with a large electrode system that is the commonest bad measurement of all, because getting far enough away means running a test lead across a car park or a paddock and nobody much wants the walk. Short distances always err low, which is the flattering direction and therefore the dangerous one.

Clamp-on testers measure a loop and not an electrode. They work where the electrode under test is one branch of a parallel network with a low-impedance return through the supply neutral, and they are quick and safe because nothing has to come apart. On an isolated electrode with no parallel path they return nonsense with the same confidence as a valid reading, and on a multi-electrode system they report one branch — which is frequently exactly what you wanted.

Disconnecting an electrode in order to test it strips the earth from a live installation for the duration of the test. Where the arrangement leans on that electrode for fault protection, either isolate the supply first or choose a method that leaves it connected. Test clamps exist so this decision does not get made casually with a spanner in one hand.

1. Establish which supply earthing arrangement you have, and whether the installation can safely lose this electrode for the length of the test.
2. Run the current probe out along a straight line, as far from the electrode as the site allows and clear of buried metalwork.
3. Take readings with the potential probe at several positions along that line, never at one position alone.
4. Plot resistance against distance and look for the flat section; a curve that never flattens means the current probe has to go further out.
5. Record the value, both probe distances, the date and the recent weather, since the number means nothing on its own.
6. Reconnect, confirm the connection is tight and the label legible, and re-prove continuity from the main earthing terminal to the electrode.

Supplementary Bonding, and the Conditions for Leaving It Out

Supplementary bonding is local work. It ties extraneous conductive parts and exposed conductive parts together inside a space where the chance of a body touching both at once is raised — a room containing a bath or shower, a pool surround, an agricultural building. The conductors are short, an impedance requirement governs their size, and the whole job is removing a difference inside one small volume.

Modern rules permit its omission in a bathroom when several conditions hold together: residual current protection on every circuit of the location, protective conductor continuity established back through the main bonding, and disconnection times satisfied for every circuit serving or passing through the space. All of them, verified on the day, not inferred from the presence of one device on the board.

Refurbishment is where this comes apart. A bathroom rewired under a partial upgrade, with the supplementary bonding pulled out because the new board has residual current protection, has satisfied one condition and been credited with three. Nobody goes back to test the main water bond in a house whose incoming main was swapped for plastic a decade earlier.

Pools and spas carry their own equipotential requirements, and those are about bonding the structure itself — reinforcement, coping, ladders, handrails and the water — into a single plane. None of that lapses because the circuits happen to sit behind residual current devices.

Loop Impedance Is the Only Proof That Counts

All of the above converges on one measurement taken at each point of the installation: the impedance of the entire loop from the source, out along the line conductor, through the fault, back along the protective conductor and home. That figure, set against the device's operating characteristic, decides whether a fault clears in the required time. It gets measured. It does not get hoped for.

The external portion of the loop is established at the origin and the internal portion accumulates along each circuit, so the far end of the longest run is where the arrangement either passes or fails. Readings are taken with conductors at ambient temperature and those conductors run hotter under load, so comparing a raw reading straight against the tabulated maximum is optimistic by a margin that decides the marginal circuits.

On a supply where the distributor provides no earth, the electrode sits in that loop and the impedance will be far too high for any overcurrent device to clear an earth fault. Residual current protection does the disconnecting instead, and the electrode's resistance then has to be low enough to hold the touch voltage during a fault inside the limit — and stable enough to still be there in August.

A structural lightning protection system belongs inside this arrangement and is not an alternative to it. Bonded in, it shares the same reference. Left as an island, it becomes a second reference in the same ground, and the difference between the two turns up across whoever is touching a downconductor and a handrail at the same moment.

Write the numbers down and keep them where the next person will find them. Loop impedance at the origin and at the extremities, electrode resistance with its date and the ground conditions, continuity of every main bond, and the supply arrangement you established at the cutout. Every future addition to this installation depends on those figures existing, and re-establishing them costs an outage nobody has budgeted for.