Why Transformers Use Silicon Steel Sheets as Cores: Loss, Grades and Lamination Design

Why a Transformer Core Needs a Dedicated Magnetic Steel

A power or distribution transformer couples energy between two windings through a closed magnetic circuit - the core. That core has to carry an alternating flux with the least possible loss, hour after hour, for decades. In nearly every oil-immersed and dry-type unit the core is stacked from grain-oriented silicon steel strip, also called electrical steel, GOES or silicon steel sheet. The choice is not tradition; it follows from three measurable properties of the alloy and of the way the strip is processed.

Silicon raises the electrical resistivity of the iron and narrows its hysteresis loop.

Grain orientation aligns the easy magnetization direction of the crystals with the flux path.

Thin, insulated laminations confine eddy currents to a very small cross-section.

Thickness tolerances, grade designations and guaranteed magnetic properties for these products are set out in GB/T 2521.2, IEC 60404-8-7 and ASTM A876, and every figure quoted below can be checked against those documents.

What Grain-Oriented Silicon Steel Is

Silicon steel is an iron-silicon alloy produced as thin cold-rolled strip for magnetic circuits. In grain-oriented grades the crystal grains are aligned by a controlled rolling and annealing sequence, producing the Goss texture, so that the easy magnetization direction of the iron lattice lies along the rolling direction of the strip.

Because flux in a transformer core follows defined paths along the limbs and yokes, a steel whose best magnetic direction coincides with those paths gives the highest permeability and the lowest specific loss in service. Non-oriented grades, whose grains are randomly oriented, are preferred where the flux direction rotates, as in motor and generator stators.

Cold-rolled grain-oriented strip is supplied in nominal thicknesses of 0.23 mm, 0.27 mm, 0.30 mm and 0.35 mm, with thinner 0.20 mm and 0.18 mm grades used in high-efficiency designs. GB/T 2521.2 defines the Chinese grade system and its tolerances, IEC 60404-8-7 is the international equivalent, and ASTM A876 is the North American specification for the same class of material.

The Three Jobs Silicon Does in the Steel

Iron is already ferromagnetic, so silicon is not added to create magnetism. It is added to improve the behaviour of the steel under alternating excitation, and it does so in three ways that can be verified on a mill test certificate or in a laboratory.

Higher electrical resistivity, lower eddy-current loss. Induced circulating currents meet more electrical resistance, and eddy-current loss is inversely proportional to resistivity, so that component of core loss shrinks directly.

Narrower hysteresis loop, lower hysteresis loss. The area enclosed by the B-H loop becomes smaller, and hysteresis loss per cycle is proportional to that area, so each magnetization cycle wastes less energy.

Lower magnetostriction, quieter cores. The small dimensional change of the steel under magnetization is reduced, which lowers the magnetostrictive hum of an energized core.

The trade-off is that silicon slightly reduces the saturation induction of the steel and makes the strip harder and more brittle. Grain-oriented products therefore use a compromise level of about 2.9-3.5% silicon. Standards classify these grades by guaranteed magnetic properties, meaning core loss and magnetic induction, rather than by silicon content, so a silicon percentage alone does not identify a grade.

Where Core Loss Comes From: Hysteresis and Eddy Currents

As soon as a transformer is energized, the alternating flux consumes power in the core before any load is connected. This iron loss, or core loss, has two physical origins.

Hysteresis loss is the energy dissipated in each cycle as magnetic domains are reoriented against internal friction. It rises with frequency, roughly with the square of flux density, and with the area of the hysteresis loop.

Eddy-current loss comes from currents induced inside the core, circulating in planes perpendicular to the flux direction. The loss scales with the square of frequency, roughly with the square of flux density, with the square of lamination thickness, and inversely with resistivity.

Because core loss appears whenever the unit is connected to the supply, it is identical to the no-load loss of the transformer: it runs 24 hours a day whatever the load, and it is a major driver of lifetime operating cost. This is why the loss figures of the core steel, and not only its price per tonne, decide the efficiency class of the finished transformer.

Why Cores Are Laminated and Coated

A solid block of steel would let large eddy currents circulate through a huge conducting cross-section, and the core would overheat within minutes. The cure is to divide the core into thin sheets stacked with the sheet plane parallel to the flux and perpendicular to the plane in which eddy currents circulate.

Thin sheets shorten the eddy-current path. Since the loss varies with the square of thickness, splitting a core into 0.23-0.35 mm sheets instead of one solid block reduces that component by orders of magnitude.

Insulating coating blocks interlaminar currents. Each sheet carries a surface insulation whose classes are defined in IEC 60404-1-1, so eddy currents cannot leak from one lamination to the next.

Stacking factor shapes the design. Coating thickness and surface roughness mean the sheets fill about 95-97% of the theoretical core volume. Thinner sheets lower eddy loss but also lower the stacking factor and raise manufacturing cost, so thickness is selected against the target loss class.

Production follows a standard sequence: the grain-oriented coil is slit to width, punched or cut into lamination shapes such as E-I pairs, three-limb E shapes or rectangular strips for wound cores, then stress-relief annealed to restore the magnetic properties damaged by cutting, and finally stacked with interleaved joints and clamped before the windings are added.

Typical Performance Values for Grain-Oriented Grades

Parameter	Typical value or grade example	Basis
Nominal thickness	0.23 / 0.27 / 0.30 / 0.35 mm	GB/T 2521.2; IEC 60404-8-7
Core loss P1.7/50 (1.7 T, 50 Hz)	Grade 30Q120: 1.20 W/kg maximum	GB/T 2521.2, where 30 = 0.30 mm, Q = grain-oriented, 120 = 1.20 W/kg
Magnetic induction B8 (800 A/m)	1.78 T minimum for the 30Q120 class	GB/T 2521.2
Silicon content	about 2.9-3.5%	Typical value; grades are classified by magnetic properties, not by silicon level
Stacking factor	about 0.95-0.97	Typical value, stated on the mill test certificate
Surface insulation	coating class per IEC 60404-1-1	IEC 60404-1-1

These figures are indicative of standard commercial grades. Guaranteed values should always be read from the latest edition of the relevant standard together with the mill test certificate issued for the specific lot.

Questions Buyers Ask About Silicon Steel Cores

Q: What do the characters in grade 30Q120 mean?
The number 30 is the nominal thickness in hundredths of a millimetre, the letter Q denotes grain-oriented material, and 120 is the maximum specific total loss P1.7/50 in hundredths of a watt per kilogram, so the grade is guaranteed at or below 1.20 W/kg.

Q: Why is core loss often called no-load loss?
Because it is present whenever the transformer is energized and does not depend on the connected load. It is measured on the low-voltage side with the secondary open, which is exactly the no-load test condition.

Q: Why not machine a core from one solid block of steel?
A solid core would offer a large conducting area to induced currents, producing severe eddy-current heating. Laminating the core with insulated sheets breaks those current paths and is the reason stacked laminations are used.

Q: Can non-oriented silicon steel be used in a transformer core?
It can be used in small, low-cost units, but its higher and more variable loss makes it uneconomic for distribution and power transformers, where the flux follows a fixed direction and grain-oriented strip is the better match.

Q: Does a higher silicon content always mean a lower core loss?
No. Increasing silicon helps resistivity and the hysteresis loop, but it also lowers saturation induction and makes the strip brittle. Mills optimise the level, and standards judge the finished grade by guaranteed loss and induction rather than by chemistry alone.

Q: What role does the surface coating play?
The coating insulates adjacent laminations, prevents interlaminar short circuits and contributes to corrosion resistance during storage. Coating class selection follows IEC 60404-1-1 and affects the stacking factor of the finished core.