Common mesh or isolated island? What ITU-T K.27 says about earthing layout inside the building
7 min read
In the equipment room of a central office being expanded, the fitter is about to clamp the new row of cabinets to the cable tray running under the suspended ceiling. The site foreman stops him: does that row belong inside the zone the original design — years earlier — isolated from the rest of the room, or inside the common earthing mesh that runs everywhere along the other corridors? The file only says earthing system built to standard practice. No drawing shows where the island ends and the mesh begins. Work stops for two days, not because of a fault, but because of a question nobody had written down at the time.
Two networks, not one
ITU-T K.27 (03/2015), Bonding configurations and earthing inside a telecommunication building, still in force — checked on the official ITU page, edition 201503, superseding the 03/91 and 05/96 editions — describes two alternative configurations for earthing equipment inside a telecommunication building, not the same question as K.47 on lightning, which concerns the cable outside, in a trench or overhead.
The first network is the CBN, common bonding network: the one everything connects to, without restriction. The Introduction states it plainly: “The CBN is the principal bonding and earthing network inside the building”. Every frame, every cable tray, every protective conductor can join it. The second is the IBN, isolated bonding network, defined in clause 3.2.4: “A bonding network that has a single point connection (SPC) to either the common bonding network or another isolated bonding network. All IBNs considered here will have a connection to earth via the SPC” — an island with a single connection point to the CBN, or to another island, still earthed through that one point.
The roughly two-metre window
Between the two networks there is no wall, only an interface of defined size: the SPC window (SPCW), clause 3.2.9. “The interface or transition region between an isolated bonding network (IBN) and the common bonding network (CBN). Its maximum dimension is typically 2 metres. The SPC bus-bar (SPCB), or frame, lies within this region and provides the interface between IBN and CBN” — typically two metres, not a fixed limit but an order of magnitude for designing the room around. The sentence that follows is the one a distracted specification skips: “Conductors (e.g., cable shields or DC return conductors) that enter a system block and connect to its IBN must enter via the SPCW and connect to the SPCB or frame” — cable shields, DC returns (a matter of routing here, not the voltage range already covered elsewhere): every conductor entering the isolated block passes through there, not through whichever point is convenient.
The examples in Annex B make it operational: “All conductors and cables connecting to the system block shall pass near to the SPC (i.e., through the SPC window)”, and cables foreign to that block but crossing the area must be kept apart — “Alien cables crossing the area of the IBN must be spaced sufficiently apart from cables connecting to the SPC and the system block”. A fibre cable with a metallic element crossing the island’s area does not pass through by habit: it does so if the drawing says so.
What you gain, what you give up
Clause 8 weighs the two choices without declaring a winner. The island’s advantage is shielding: “a high level of shielding is attainable from DC through tens of kHz or perhaps hundreds of kHz depending on the size of the IBN”, because “the single point connection between the IBN and CBN results in negligible current flowing between CBN and IBN” — one single contact, almost no current crossing it. The price is restriction: “To limit the risk of electric shock between an IBN and the surrounding CBN, it is necessary to limit the size of the IBN (both horizontal and vertical extent)”; “Passageways that form the boundary between IBN and CBN, should have a minimum width imposed” — even passageway width becomes a design figure. The text sums it up bluntly: “Disadvantages of IBN installation are cable routing restrictions and the additional expense (compared to mesh-BN) of maintaining the isolation”.
Whoever chooses the common mesh gives up that shielding but gains freedom of routing: “equipment frames may be connected to the surrounding CBN without restriction”. Neither is the right choice outright: it is right for that building, that traffic, that equipment density.
The row of cabinets that arrives two years later
This is where the decision becomes a matter of layout, not electrical engineering. A specification that settles for earthing system compliant with standards does not tell whoever arrives later whether the new row belongs in the mesh or must pass through the SPC window. A drawing has to say so, not the memory of whoever built the first phase: the chosen configuration — mesh, or a bonding-mat island, or a star island with an isolated DC return, the three examples K.27 itself describes in clause 10 and Annex B — the position and dimensions of the SPC window on the plant drawing, the minimum width of the boundary passageways, and the list of conductors bound to pass through there. Without those lines, whoever works the room afterwards guesses, or redoes it.
Maintenance gives no discounts
An earthing network degrades in silence, not with an alarm: clause 9 covers exactly that, what an initial acceptance test does not see. On the common mesh, small changes stay small: “small changes that occur in the CBN generally have only a small effect on its shielding capability”, and reinforcing it just means adding conductors when needed. On the island it is the opposite: “IBN systems are more difficult to maintain, because craft-person activity is liable to result in inadvertent interconnections between IBN and CBN, violating the desired single point connection, and introducing surge currents into the IBN” — one cable clamped at the wrong point is enough to undo years of isolation. That is why the standard also gives a way to check it: “Verification of single point connection in a DC-I system is facilitated if this connection is made with a conductor, around which, a DC clamp-on ammeter can be clamped. Zero current confirms single point connection” — a clamp meter, zero current, the only acceptable result. And it closes with a recommendation that applies to both networks: “It is recommended that systematic verification be performed on all bonding configurations and earthing connections inside a telecommunication building”.
What we haven’t verified
Compliance with K.27, like any ITU-T Recommendation, is voluntary: “Compliance with this Recommendation is voluntary”. We have not read Annex A in full, on the frequency-domain theory of shielding, nor the standards IEC 60364-5-54 and IEC 60364-4-41 cited for low-voltage installations, paywalled texts we did not check. We neither claim nor rule out a specific Italian transposition of K.27: not verified against a national primary source.
Two pillars, on a building’s earthing
First pillar: the configuration — mesh or island, and which variant — written into the specification we check, together with the SPC window drawing, the boundary passageway widths and the list of conductors bound to pass through it. Acceptance testing checks that the drawing and the room match, not that someone remembers where the island is.
Second pillar: that zero-current clamp reading, repeated over the years at the same SPC point, does not stay a single isolated measurement. With CSIDIA, the group’s other company, continuity checks, layout drawings and maintenance records come together into a single map of the earthing network, on which an AI flags drift — a current appearing where it was meant to stay at zero — and the crew steps in: on-premises, with no deep integration into the client’s network, or a dedicated cloud with a data centre in Italy, always with shared management. It applies to a data centre as much as to a central office with one long-standing isolated area.
From the site visit, at no cost, we bring the list relevant to your room: configuration stated in the design or missing, SPC window drawn or only passed down by word of mouth, last continuity check dated or never carried out — blanks included. Talk to an engineer: the extra row of cabinets arrives sooner or later, and the question of where it should connect should not go unanswered.