Insulation monitoring device (IMD) — measuring principle per IEC 61557-8

The IT system and first-fault detection guide describes what an insulation monitoring device (IMD) does: continuously monitor the insulation resistance between the live conductors of an IT system and earth, so that a first earth fault is signaled before a second fault leads to a dangerous loop short-circuit. This article goes deeper into how an IMD technically implements that measuring principle, as specified in IEC 61557-8.

The challenge: measuring while the network stays live

An IMD must measure the insulation resistance of a network that remains live and in normal operation — the measurement must not disrupt normal operation and has to work despite the present system voltage, network capacitance (cable length, filter capacitors), and any other active equipment on the same network. This requires an active injection signal from the IMD itself, separate from the system voltage, whose resulting current is measured to derive the insulation resistance.

AC injection method versus AMP/varying-DC injection method

In practice, IEC 61557-8 distinguishes two main principles for the injection signal:

AC injection method: the IMD injects a low-frequency AC signal (typically a few Hz, offset from the system frequency to avoid interference) between the live conductors and earth, and measures the resulting signal current to compute the insulation resistance. This method is simple but can be sensitive to network capacitance on large, extensive networks.
AMP method (varying-DC injection): the IMD injects a DC signal with varying amplitude (an "amplitude modulated pulse"-like signal) and measures the response. This method is more resilient to high network capacitance and is often applied on networks with a lot of power electronics (variable-speed drives, rectifiers) where AC injection can suffer from harmonic noise or capacitive leakage currents.

The choice between the two methods is not a matter of "better" or "worse" in an absolute sense, but of suitability for the specific network: network size, capacitance, presence of power electronics, and the number of already-present IMDs (with multiple IMDs on coupled networks, interference between their injection signals must be avoided, often through frequency- or time-multiplexing between the devices).

Alarm thresholds: Ra1 and Ra2

An IMD operates with two configured alarm thresholds for the measured insulation resistance:

Ra1 (pre-alarm): a higher resistance threshold that signals a declining insulation resistance before it reaches a critical level — a warning that maintenance or investigation is needed, without immediate danger yet.
Ra2 (main alarm): a lower resistance threshold indicating that the insulation resistance has reached a level requiring action — in practice often linked to the threshold above which an IT system still functions safely under a first fault per NEN 1010/IEC 60364-4-41.

The exact threshold values are set per installation based on the nominal system voltage and the required safety margin, rather than being a single universal fixed value.

Coupling impedance: why not every IMD fits every network

The coupling impedance of an IMD (the impedance through which the device itself is connected between the live conductors and earth) must be sufficiently high relative to the network impedance and relative to any other measuring or protective equipment connected to the network, so that the IMD itself does not disrupt normal network operation and does not cause false readings through interaction with other equipment (such as variable-speed drives with their own EMC filter capacitors to earth). On networks with high capacitance to earth (long cable runs, many filter capacitors), the IMD must be specifically selected for a measuring principle and coupling impedance that can handle that — an IMD that works fine on a small, short IT network can give unreliable readings on a large network with substantial capacitive loading.

Practical relevance

When selecting an IMD for an IT system, the installer/designer should not only look at the nominal voltage class, but also at the injection principle (AC or AMP/varying-DC) in relation to network capacitance and present power electronics, the configurability of Ra1/Ra2, and — with multiple IMDs on coupled networks — the ability to synchronize with one another to prevent interference between the injection signals.

Common mistakes

Applying a standard AC-injection IMD to a large, high-capacitance network (long cables, many filter capacitors) without checking whether the measuring principle is suitable.
Placing multiple IMDs on mutually coupled IT networks without synchronization, allowing the injection signals to interfere with one another and cause faulty readings.
Not tuning Ra1/Ra2 to the actual system voltage and required safety margin, leaving the factory setting unchanged instead.
Ignoring the IMD's coupling impedance on networks with substantial power electronics, which can lead to interaction with EMC filters and unreliable readings.