Earth resistance measurement: what should the test report say?
7 min read
You accepted the equipment room on the strength of a report that says one thing: earth resistance 1.8 Ω, pass. Five years later a fault upstream raises the earth potential, two units burn out, and the loss adjuster asks by which method that number was obtained, at which point, with the probes how far out. The report does not say. Nobody had asked: not the specification, and not whoever signed the acceptance.
The number depends on the method, and every method has conditions
Recommendation ITU-T K.107, Method for determining the impedance to earth of earthing systems, edition 1.0 approved on 29 November 2015, in force and freely downloadable from the ITU catalogue, starts from the risk run by whoever operates the equipment: the earth potential that rises in the power network and is transferred to the telecommunication one, “potentially dangerous to the people working in the telecommunication plant; damage to the plant itself can also occur”.
The first warning concerns the very word printed on the report. Clause 7.1: “The measured impedance value is often called resistance, even though it contains a reactive component. The reactive component can be very significant for large or interconnected earthing systems.” Clause 6.2 says when it stops being negligible: “Determination of the reactive component is necessary when the analysis involves surge or impulse currents.” That is, in the face of surges and impulse currents — the very case the earthing was built for, and the reason lightning protection is a chapter of its own.
The 62% rule holds only if three things are true
The classic field method is fall-of-potential: current is injected between the electrode under test and a current probe, the voltage is read at a potential probe, placed in practice at 62% of the distance to the first. K.107 lists the conditions under which that 62% means anything: “a fairly uniform soil”; “large distances between the earth grid under test and reference electrodes so that all the electrodes can be assumed to be hemispherical”; “the electrode under test has no external earth connections”.
The third one, in a central office or a data centre, is almost never true: cable screens, the power system neutral, armouring and pipework are all external connections, and the value read is not that of the electrode. The minimum field check is to move the potential probe: “Two to three consecutive constant resistance readings can be assumed to represent the true resistance value” — two or three steady readings, not one. And the distances, which K.107 carries from EN 50522: “The distance of the probe PP from the earth electrode under test should be at least 2.5 times the maximum extension of the earth electrode under test (in measuring direction), but not less than 20 m; the distance of the auxiliary electrode CP must be at least four times the maximum extension, but not less than 40 m”. With a test frequency that “should not exceed 150 Hz”.
The clamp is convenient, and its field of use is narrow
The clamp-on meter spares you stakes and distances: you close it around the earthing conductor and read. It rests on an assumption — a multi-earthed neutral whose impedance is negligible against the electrode under test — and K.107 says where it collapses: “The method is not applicable to a multiple-connected earth electrode system, such as a substation earth grid, multi-connected pole, or structure earths”. These instruments typically work between 1 kHz and 3.4 kHz: at those frequencies the test-circuit reactance can distort the reading whenever the inductance is not negligible. One line is about management rather than physics: “Clamp-on meters require frequent calibration to assure proper operation” — and calibration is a date on a traceable certificate.
For extensive grids, heavy-current injection remains, with a remote electrode far enough away to separate the zones of influence — “e.g., 5 km for extended earthing systems” — and a test current above the interference: “This is generally ensured for test currents above 50 A”. One prescription has to be written down beforehand, because on site the temptation runs the other way: “Earth wires and cable sheaths with earth electrode effect, which are operationally connected to the earthing system, shall not be disconnected for the measurement”.
The measuring point is a choice, not a detail
The report gives a value: of which point? For the telecommunication circuit entering a substation, K.107 is blunt: “the potential should be measured between this point (earthing at the telecom room) and the remote earth”. The point is where the circuit terminates, not the most convenient pit to reach. In an equipment room that means writing it down — the equipotential bonding bar at the entry, not the installation — and finding it again five years later, with the care demanded by the earthing layout choice between common mesh and isolated island. Meanwhile the soil shifts under the electrode: “the earthing impedance can vary with the season as the temperature, moisture content, and density of the soil change”. A number with no date and no season cannot be compared with the next one.
The Italian duty exists; the content of the report does not
In Italy the verification is not optional. Presidential Decree 462 of 22 October 2001 (text in force on Normattiva, last updated 29 February 2020), in Article 4(1), requires the employer to carry out “regolari manutenzioni dell’impianto” and a “verifica periodica ogni cinque anni” — ours: regular maintenance of the installation and a periodic verification every five years — cut to every two years for construction sites, premises used for medical purposes and environments at higher fire risk. Paragraph 3 adds that the report must be kept and produced to the enforcement authorities; Article 7(2) calls for an extraordinary verification after an “esito negativo della verifica periodica” or a “modifica sostanziale dell’impianto” — a failed periodic verification, or a substantial modification: one more row of cabinets, a new entry route. What the decree does not do is state what the report must contain: Article 4(2) defers to European UNI CEI technical standards. That field is filled in by the specification, or left blank.
What we do not know
Compliance with K.107 is voluntary: “Compliance with this Recommendation is voluntary.” We have not read EN 50522 (2010) or IEEE Std 81 (2012), both paid documents: the values above are the ones K.107 itself publishes, attributing them to those sources. We have not read CEI 64-8 or the IEC 62305 series, and we quote no threshold from them. Table 1 and Figures 8-10 of K.107 are images inside the PDF: we extract no values from them. We have not read ITU-T K.104 or K.26, which the text cites. Finally, Article 2 of Presidential Decree 462/2001 still names ISPESL while Article 7-bis, added in 2019, speaks of INAIL: we have not verified on a primary source the instrument that transferred those functions.
The two axes, applied to an earthing value
First axis: in the specification we review, the measuring point, the method, the probe geometry, the test frequency, the state of external connections and the date with the soil conditions become written lines. At acceptance we check that the report carries them: not earthing system verified, but a value that can be reproduced at the same point.
Second axis: those values, repeated every two or five years, are only useful if they are comparable. With CSIDIA, the other company in the group, reports, instrument calibration certificates and installation drawings converge into a single map of the network, on which an AI flags the drift — the same point creeping upward between two verifications, or a measurement taken by a different method from the previous one — and the crew acts on it. It runs on-premises, with no deep integration into the client’s network, or in a dedicated cloud with data centres in Italy, always under shared management. It holds for a central office as much as for a data centre with the delivery substation twenty metres from the hall.
From the site visit, at no cost, we bring the list that applies to your site: measuring point stated or generic, method written down or absent, probe geometry recorded or not, date of the last periodic verification and of the next — empty boxes included. Talk to one of our engineers: the upstream fault arrives without notice, and the question of how that number was measured should not be left unanswered.