Current Transformer Interface Design Guide

A current transformer measures alternating current by producing a smaller current in proportion to the one flowing in a primary conductor. It provides isolation for free, needs no supply of its own and works from the very low currents of a standby load up to the large currents of a fault, which is why it is still the standard sensor in electricity metering and industrial protection.

How a Current Transformer Works

The primary is usually a single conductor passing through a magnetic core, and the secondary is a winding with many turns on the same core. The alternating current in the primary produces a magnetic flux in the core, and the secondary current is the primary current divided by the turns ratio, provided the secondary circuit is closed through a low impedance.

The transformer therefore behaves as a current source rather than a voltage source. Its output current follows the primary current closely, and the voltage across the secondary is whatever the secondary circuit allows it to be. That is why the output must always have a defined load, and why an open secondary is a dangerous condition.

Because the proportion holds over a wide range, a single device can measure a few milliamperes of primary current and still represent a fault current a thousand times larger, within the limits imposed by the core. Those limits, rather than the ratio, are what determine the useful range.

Turns Ratio and Burden Resistor

turns ratio is quoted as the primary to secondary ratio, for example 1000 to 1, and it sets the scale factor of the measurement. A 100 ampere primary with a 1000 to 1 ratio produces 100 milliamperes in the secondary, which is a convenient level to feed into a burden resistor and an amplifier.

The burden resistor converts the secondary current into a voltage, and its value determines how much of the available compliance the device uses. A larger burden gives a larger signal and better noise immunity, but it also requires the core to develop more voltage, which brings the saturation limit closer. The datasheet usually specifies a maximum burden for a given accuracy class.

The shunt should be a low temperature coefficient resistor with a tight tolerance, because its value appears directly in the measurement. Where the burden is a long thin trace instead of a resistor, its temperature coefficient and its tolerance both enter the result, which is a common cause of a measurement that drifts with load.

Core Saturation and Rating Factor

core saturation occurs when the flux in the core reaches the limit the material can carry, at which point the secondary current stops following the primary and the measurement collapses. It happens at a primary current that depends on the burden, the core size and the material, and it is specified as an accuracy limit factor.

A protection transformer is designed to saturate late, so that it can represent a fault current accurately, while a metering transformer is designed for accuracy at normal load and may saturate much earlier. Using a metering device for protection produces a measurement that is wrong precisely when it matters most, which is a design error worth avoiding.

Direct current in the primary also affects the core. A transformer cannot transfer direct current, and a large offset in the primary current drives the core into partial saturation once per cycle, which produces a distorted secondary waveform and an error in any measurement derived from it. Where direct current is present, a different sensing technology is usually required.

Phase Shift and Accuracy Class

phase shift is the displacement between the primary current and the secondary current, and it is caused by the magnetising current and the core losses. It matters because power measurement depends on the angle between voltage and current, so a phase error appears directly as a power error even when the magnitude is correct.

Accuracy class defines both the magnitude error and the phase error over a specified range of current, with the tightest classes reserved for revenue metering. A class 0.5 device holds half a percent of magnitude error over its rated range, which is adequate for monitoring and not adequate for billing without calibration.

Compensation is possible in firmware once the characteristic of the device is known. Measuring the phase error at several currents and storing the correction removes most of it, which is why a calibrated class 1 device can outperform a class 0.5 device that has not been characterised in the actual circuit.

Protection and Open Secondary

An open secondary is the fault that destroys current transformers and the equipment connected to them. With no load, the secondary current has nowhere to go, and the core develops whatever voltage is needed to try to maintain the flux, which can reach hundreds or thousands of volts and break down the insulation.

Protection is therefore two fold: a shorting terminal on the secondary that is closed before the measuring circuit is disconnected, and a voltage limiting device across the secondary winding that clamps the voltage if the circuit opens unexpectedly. The clamp must be a device that tolerates the continuous secondary current without conducting during normal operation.

Terminal blocks for current transformer circuits are usually designed so that shorting the secondary requires no additional links, and the drawing should state clearly that the secondary must never be left open. In an installation where a meter is replaced, this is the single procedure that prevents an expensive failure.

Measurement and Sampling

The burden voltage is an alternating signal at the supply frequency, and it has to be converted in a way that preserves its magnitude and its phase. A converter with a simultaneous sampling input and a good antialiasing filter is the usual choice, and the sample rate should be an integer multiple of the line frequency so that the measurement window contains a whole number of cycles.

Sampling a whole number of cycles removes the error caused by a partial cycle at the end of the window, which appears as ripple in the calculated root mean square value. Where the line frequency can vary, the sample rate should follow it, or the window should be long enough that a partial cycle is a small fraction of the total.

Where several phases are measured, the channels have to be sampled at the same instant. A multiplexed converter introduces a delay between channels that appears as a phase error between phases, and the power calculated from those channels is then wrong. Simultaneous sampling, or a correction based on the known sampling delay, removes the problem.

Layout and the Isolation Barrier

The secondary circuit is isolated from the primary conductor, and the barrier provided by the transformer is only as good as the layout around it. The secondary side should be treated as a safety extra low voltage circuit, with its own reference and its own ground area, and no copper should cross the barrier except through the transformer winding.

The burden resistor and the protection device belong close to the transformer terminals so that the loop carrying the secondary current is small. A long loop adds inductance, which affects the phase response, and it also picks up magnetic interference from nearby conductors, which appears directly in the measurement because the signal is a current rather than a voltage.

Where the transformer is a split core type that is clamped around a cable, it should be positioned so that the cable passes centrally through it. An off centre conductor, or one that returns outside the core, produces a measurement that is lower than the true current and that varies with the cable position. The same reasoning applies to any magnetic sensor and the methods for keeping interference out of the resulting signal are described in our articles on EMI immunity in mixed signal design and mixed signal board design.

Verification and Common Faults

Verify the channel with a known current and a reference meter, at several points across the range and at the actual operating frequency. Comparing the phase as well as the magnitude, using a reference that provides both, shows whether the compensation is correct rather than merely whether the scale factor is right.

A reading that is consistently low by a small percentage is usually a burden or calibration error, while one that falls sharply at higher currents indicates that the core is saturating. Measuring the secondary waveform with a scope distinguishes the two: a sinusoidal waveform that has grown smaller indicates a scaling problem, while a waveform with flattened peaks indicates saturation.

A measurement that drifts with temperature usually has a copper burden rather than a resistor, or a burden resistor with a poor temperature coefficient. In a humid installation, surface leakage across the high impedance nodes of the amplifier adds a further error that appears only in the field, and the techniques in conformal coating are the usual remedy.

FAQ

What happens if the secondary is left open? The core develops a very high voltage that can destroy the insulation and endanger anyone touching the terminals. Always short the secondary before disconnecting the measuring circuit.

How do I choose the burden resistor? Start from the maximum secondary current and the input range of the measuring circuit, then check that the resulting burden voltage is below the maximum specified for the accuracy class you need.

Can a current transformer measure DC? No. A transformer transfers only alternating current, and a direct component in the primary drives the core towards saturation. A shunt or a Hall effect sensor is used where direct current is present.