What voltage must a central-office device withstand? What EN 300 132 says about ICT power supply

During a blackout, a central-office backup battery drops below its normal range while the standby generator starts. A network device reboots on its own, no visible fault, no alarm logged beforehand. The specification it had been bought against said only power supply −48 V: no range, no tolerance, no reference to a standard. Was that device compliant with something, or only with the nominal voltage printed on its label?

Three interfaces, one family of standards

ETSI — Europe’s telecommunications standards body — publishes the EN 300 132 series through its Environmental Engineering technical committee, in three parts: alternating current, EN 300 132-1 V2.2.1 (2022-11); direct current at −48 V, EN 300 132-2 V2.8.1 (2024-10); direct current up to 400 V, EN 300 132-3 V2.3.1 (2023-01). Not three unrelated standards, but the same physical interface — the point between the power supply system and the ICT equipment — described for three different technologies, with the same approach.

The purpose stated in part 2 is to make equipment and power supply systems of different origin interchangeable within the same network. The text puts it this way: “to facilitate the installation, operation and maintenance in the same network of ICT equipment and systems from different origins”. A specification that simply cites the value −48 V without referencing the standard is asking for a nominal figure, not a verifiable interface.

−48 V is never exactly −48 V

Part 2 fixes the nominal voltage at one precise point, interface “A”: “The nominal voltage at interface “A” shall be -48 VDC with positive conductor connected to earth” — the reference is to the companion standard ETSI EN 300 253 on earthing. But the nominal value is never the real value in service: on float charge the battery sits closer to −54/−57 V, on discharge it drops. The standard therefore defines a normal service voltage range: “The normal service voltage range for the -48 VDC nominal supply at interface “A” shall be from -40,5 VDC to -57,0 VDC.”

Outside that range, but not yet out of standard, there is an abnormal range within which the equipment must not be damaged, even if it may stop operating: from 0,0 V to −40,5 V, and from −57,0 V to −60,0 V. The sentence introducing those values is blunt: “ICT equipment designed to work at 48 VDC nominal voltage at the interface “A” shall not suffer any damage when subjected to the following voltage ranges defined in table 2.” When the voltage returns within the normal range, service must resume on its own, with no manual intervention. Why those figures, rather than round numbers, is explained in a note: “In most cases the nominal voltage of interface “A” is based on a 24 cells lead-acid battery.” Twenty-four lead-acid cells, not an arbitrary number.

The jump to 400 V DC

Part 3 covers a different interface, “A3”, for a case that is becoming ever more common in data centres: feeding power directly in DC up to 400 V, skipping the double AC/DC conversion inside the rack. The scope states it up front: “the nominal voltage at power interface “A3” of ICT equipment defined in the present document is DC voltage up to 400 V”. The normal service voltage range is set by two thresholds: “minimum voltage: 260 VDC” and “maximum voltage: 400 VDC”.

Here too the value is not arbitrary: a note explains that “336 V is a nominal voltage defined with 168 lead-acid battery cells multiplied by the nominal cell voltage 2 V” — 168 cells, not 24: seven times the −48 V battery, so the same battery racks can be reused. The standard also sets a reference test voltage: 365 V ± 15 V.

The measurement that proves the range at acceptance

What a specification often skips is that the voltage range is not a claim to take off a datasheet: part 2 describes a step test, referencing the test standard EN 61000-4-29 for measurement technique. The equipment is driven from the minimum to the maximum of the normal range, and back, in a 0.1-second step; the edge of that step must fall between 1 and 50 microseconds, measured on a 100 Ω resistive load per the test standard. The pass criterion is blunt: no degradation in service performance during or after the test. For equipment with redundant dual inputs, the same test is repeated in three configurations: both inputs powered together, one powered with the other live, one powered with the other disconnected.

It is not a test performed by eye with a multimeter during a routine acceptance check: it requires a test generator compliant with the referenced standard. It is the kind of test asked of the equipment supplier at type-approval stage, not one redone on site — but it is the document a specification can demand to see before installation.

Where each piece of proof sits today

The EN 300 132 conformity certificate, where it exists, sits with the manufacturer or the laboratory that ran the step test — it does not travel with the supply unless the specification explicitly asks for it. The rectifier’s datasheet, with the cell count and the declared float voltage, sits with the electrical installer. The battery maintenance log, where it exists for the electrical plant as well as for the fibre, sits with whoever runs the room. The calibration of the measuring instruments used to check voltage in the field sits with whoever last ran the check. Four places, and none of them shows up by default in a routine network test record.

See the service · Talk to an engineer

What we haven’t checked

This is not technical advice, nor a compliance judgement. We read the scope and the voltage-range clauses of parts 1, 2 and 3 of EN 300 132 in the versions downloaded from the ETSI website on 28 August 2026; we did not read the technical annexes in full, referenced for measurement detail and calculation examples, nor the companion standard ETSI EN 300 253 on earthing. We did not verify whether or how EN 300 132 is adopted as a harmonised standard under any specific Italian regulation for public network procurement: the foreword only lists the ETSI system’s adoption and national-transposition dates. Anyone drafting a specification on this basis should have the full text, annexes included, checked by a qualified power-systems engineer.

What to write in the specification, what to check at acceptance

In the specification: compliance with the part of EN 300 132 relevant to the power technology required (AC, −48 V DC, or up to 400 V DC); the normal service voltage range the equipment must withstand without degradation, not just the nominal value; where the equipment has redundant inputs, an explicit requirement for the step test across every configuration the standard defines; and a requirement for the certificate or test report issued by the manufacturer or a laboratory.

At acceptance, reading −48 V off the label is not enough: check that the test report cites the exact version of the standard and the reference table; that the installed battery’s cell count matches the declared range; and that the document is actually available, not just promised in a clause.

Two pillars, applied here

The manufacturer’s test report, the rectifier’s datasheet, the battery maintenance log and the instrument calibrations become, with CSIDIA, the group’s other company, a single map on which an AI checks, before acceptance, whether the declared voltage range is backed by a real test report or only by a nominal value on a datasheet. It is the same approach for the data centres we work with. Within the client’s own perimeter: on-premises on self-contained machines with no deep integration, or a dedicated cloud with a data centre in Italy, always with shared management.

The site visit, at no cost, produces the list of your equipment powered in the central office or the data hall — for each item, whether the voltage range required in the specification matches the standard or is only a nominal value, and whether the test report proving it is genuinely retrievable today or only claimed. Including the boxes that stay blank: it’s yours to keep even if we don’t go on to work together.