A 0.66 kV current transformer is the least glamorous component in a metering cabinet and one of the easiest to get wrong. It is usually bought late, installed by an electrician rather than an engineer, and expected to hold a 0.5 class while sharing a busbar with a 1,600 A feeder. When the utility's check meter disagrees with the sub-meter by two per cent, the transformer is the first suspect — and often the wrong one.

This note covers what actually moves a low-voltage CT reading: core construction, the air gap in a split-core unit, the 5–10 VA burden window, conductor position, stray fields, and the difference between a class that is guaranteed and a class that is merely printed on the label.

  • A split-core CT has a deliberate air gap; its magnetising current — and therefore its ratio error — is higher than a solid busbar core of the same size.
  • Class 0.5 is guaranteed inside a burden window (typically 25–100 % of rated VA) and a current range (5–120 % of rated primary).
  • Conductor position matters: an off-centre primary or a return conductor through the window changes the flux and the reading.
  • For retrofit sub-metering, split-core is a mechanical advantage — install without breaking the busbar — not an accuracy one.

1. What a low-voltage CT is actually measuring

A busbar-type CT at 0.66 kV is a ring core slipped over a conductor, with the secondary wound on the core and terminated at a small terminal block. The primary is the busbar itself: one turn. The ratio is set by the number of secondary turns.

Because the primary is a single turn, the core sees only the difference between primary ampere-turns and secondary ampere-turns — the excitation current. Everything that makes the CT imperfect lives in that difference: core material, air gap, burden and frequency.

2. Accuracy class, ratio error and phase displacement

Two numbers define a metering CT's accuracy: ratio error (how far the secondary current is from the nominal ratio, in per cent) and phase displacement (the angle between primary and secondary current, in minutes).

A tariff meter uses both. An electronic meter measures voltage and current and multiplies them; if the current is 20 minutes out of phase with the voltage, the energy reading is wrong even when the ratio is perfect. This is why a CT sold as "0.5 class" must hold both limits across the specified current range, not just at rated current.

ClassRatio error at 5–120 % InPhase displacementTypical use
0.5±0.5 %±30 minSub-metering, internal billing
1.0±1.0 %±60 minIndication, load surveys
0.5S / 0.2STighter, from 1 % InTighterTariff metering with light-load periods

3. The burden window is the specification

A class 0.5 core is tested at a stated burden, usually 5 VA or 10 VA for low-voltage units. Connect a meter that draws 2 VA, plus 4 m of 2.5 mm² lead out and back (about 0.06 Ω, or 1.5 VA at 5 A), and the burden is already 3.5 VA. Add a second meter in series for the utility's check and the total can exceed the rated 5 VA.

Above the rated burden the core saturates earlier, the excitation current rises, and the ratio error grows. Below about 25 % of rated burden the error does not disappear — it moves to the other side of the curve, and the "accuracy" becomes an accident of the test point.

Practical rule: add up every load in the secondary circuit, including the wiring, and specify a CT whose rated burden is at least 1.25× that sum and no more than 4× it.

4. Busbar versus split-core: the air gap is the difference

A solid busbar-type core is a closed magnetic circuit. A split-core unit is two half-cores that clamp around the conductor, and the joint between them is a small but real air gap.

The gap raises the reluctance of the magnetic circuit, which raises the magnetising current needed to produce the same flux. The consequences are predictable:

  • Higher ratio error at low current, because the excitation current is a larger fraction of the primary current.
  • Earlier saturation, so the linear range is shorter than for a solid core of the same iron.
  • Sensitivity to the joint: a speck of paint or a rust film in the mating faces increases the effective gap and degrades the error curve. Clean faces are part of the installation, not a factory detail.

Good split-core designs compensate by using a larger core, a higher-grade strip, or active electronics. Without one of those, a split-core unit that claims class 0.5 at 5 VA is claiming the best case of a compromised magnetic circuit.

5. Installation moves the reading more than the class does

Three installation effects show up repeatedly in sub-metering disputes:

  • Off-centre primary. The conductor should pass through the centre of the window. Near the edge, the flux distributes unevenly and the ratio error shifts — small for a well-designed core, large for a cheap one.
  • Return conductor. If the return conductor passes back through the window, its current subtracts from the primary and the reading can fall to near zero. This is the classic "why is this feeder reading half?" call.
  • Adjacent conductors. A nearby heavily loaded busbar induces a stray field that adds to or subtracts from the measured flux. Keep some distance, or use a toroidal core with a screened case.

None of these is a transformer defect, and none is visible on a test certificate.

6. Where low-voltage CTs are going: self-calibration

Because a split-core CT is cheap to install and expensive to remove, research has moved toward correcting its error in place rather than tightening its manufacture. Two lines of work are worth knowing:

A 2022 study in Measurement proposed an online self-correction method for split-core CTs in low-voltage distribution networks — estimating the error from operating data and compensating for it, rather than relying on the factory calibration alone. A 2024 Energies paper described a full-range charged calibration device for low-voltage CTs, aimed at testing units across their whole current range instead of at a single point.

Both address the same reality: in low-voltage metering, the installation and the operating range dominate the error budget, so the interesting engineering is in calibration and compensation, not in the core alone.

7. A selection table for the common cases

ApplicationTypeClassBurdenNote
New panel, feeder meteringBusbar type0.55–10 VAClosed core, best accuracy per size
Retrofit sub-meteringSplit core1.0 (0.5 with care)5 VAInstall without breaking the busbar
Utility check meterBusbar type0.5S10 VAAccuracy down to light load
Load survey onlySplit core1.05 VATrend data, not billing

8. Frequently asked questions

Can a split-core CT be used for billing?

Only if it is specified and tested as a metering unit — class 0.5S or 0.5 with a stated burden and a calibration certificate. Many split-core units sold for "monitoring" are class 1.0 or unclassified, which is not the same thing.

Why does my CT read low at light load?

Because the excitation current is a larger fraction of a small primary current, and because a class 0.5 core is only specified down to 5 % of rated current. If the load is routinely below that, use a 0.5S or 0.2S core, or a smaller ratio.

Does the cable size through the window matter?

Only in that it must fit, and that the conductor should be centred. A very thick cable that fills the window is fine; a thin cable lying against one side is not.

How often should low-voltage CTs be checked?

At commissioning and whenever a billing dispute appears. A ratio test at one point plus a burden measurement catches most problems; a full calibration across the range is worth it for tariff metering.

9. References and further reading

  1. An online self-correction method to improve accuracy of split-core current transformer in low-voltage distribution networks, Measurement (Elsevier, 2022). ScienceDirect
  2. Optimization Design Study of a Full-Range Charged Calibration Device for a Low-Voltage Current Transformer, Energies 17(11): 2496 (2024). MDPI
  3. Design and Application of a Low-Ratio High-Accuracy Split-Core Core-Balance Current Transformer, IEEE — split-core construction and error control. IEEE Xplore
  4. IEC 61869-2, Instrument transformers — Part 2: Additional requirements for current transformers.

Choosing a 0.66 kV unit for a distribution board or a retrofit? Our current transformer range lists the window size, ratio, class and burden per model — the window dimension is usually the first thing that rules a model out.