Grain-Oriented Electrical Steel: Core Loss; Coating Integrity and the Cutting Burr That Costs a Transformer

A transformer core built from the specified grade measures higher loss than the design predicted, and the first suspicion is the steel. More often the loss came in at the shear line, where the cutting burr closed the interlaminar insulation.

This note is for buyers and design engineers specifying grain-oriented electrical steel. It covers what the core loss grade does not describe, how coating and cutting change the result, and the incoming checks that separate a supplier problem from a process problem.

Grain-oriented steel is bought on core loss, yet coating and cutting decide what a wound core actually delivers

The grade is defined by a maximum core loss at a stated induction and frequency, measured on an Epstein sample prepared to a standard.

Your core is not an Epstein sample. It has been slit, sheared, stacked, clamped and possibly annealed, and each step moves the number.

Buy against the grade and design against the process. Treating the datasheet figure as a property of the finished core is the most common source of disappointment.

Grain orientation comes from rolling and annealing, and the grade is a position on that curve

The grains are aligned so that the easy magnetisation direction follows the rolling direction. Better alignment means lower loss for the same thickness.

Higher grades cost more because the processing to reach them is slower and the yield is lower. The price step between adjacent grades is rarely proportional to the loss step.

Model the trade before choosing a grade. Sometimes a thicker core of a cheaper grade costs less overall than a thin high-grade core with tighter assembly tolerances.

Core loss has two parts, and which one dominates depends on frequency

Hysteresis loss scales with frequency, while eddy current loss scales with the square of frequency and the square of thickness.

At mains frequency both matter. At higher frequencies the thickness becomes the dominant lever and thin grades earn their price.

Ask for loss separation data if the application runs above mains frequency. A single loss figure at fifty hertz says little about behaviour at four hundred.

Domain refinement is a surface treatment that is easily lost in handling

Mechanical or laser scribing divides the domains and reduces loss by a measurable margin. The effect lives in a surface layer a few micrometres thick.

Inappropriate annealing, aggressive handling or a coating repair that removes the treated layer will erase part of the benefit that was paid for.

Confirm whether the grade includes domain refinement and what the supplier's handling guidance is. The gain is real and it is fragile.

The insulating coating is thin, and its integrity decides both loss and corrosion

The coating provides interlaminar resistance. Once it is breached, eddy currents circulate between laminations and the core behaves like a solid block.

Coating types differ in temperature resistance and in weldability, and the choice constrains the annealing and assembly route.

Inspect the coating at the coil edges, where handling damage concentrates. A visual check plus a surface resistance test on a sample catches most of it before the core is built.

Cutting and burr introduce loss the datasheet never mentions

Shearing creates a burr and a plastically deformed zone along the cut. Both raise local loss and, when stacked, the burr can bridge adjacent laminations.

Sharp tooling, correct clearance and regular tool maintenance are the controls. A worn shear produces a burr that no amount of good steel can compensate for.

Measure burr height on production cuts rather than on the first article only. Tool wear is progressive and the loss curve follows it.

Stress relief annealing can restore loss or destroy the coating

Cutting introduces stress that raises loss, and a stress relief anneal recovers most of it. The temperature and atmosphere must suit the coating.

An anneal that is right for one coating type will damage another, and the damage is not always visible. Loss measurements before and after tell you which happened.

Ask the steel supplier for the annealing window of the specific coating, and record the actual furnace profile used. The two together are the evidence if a batch underperforms.

Stacking factor decides how much core fits in the space you have

The lamination thickness plus the coating thickness determine how much of the core window is actually magnetic material. Insulation occupies volume that does not carry flux.

Thinner steel raises loss per unit volume at a given induction but allows tighter winding. The interaction with the bobbin and the window area is a design decision.

Calculate with the real stacking factor rather than the nominal thickness. A core that does not fit is a redesign, and the steel is rarely the cause.

Magnetostriction and noise lead back to the steel more often than designers expect

The material changes dimension slightly as it magnetises, and the resulting vibration radiates as sound at twice the supply frequency and its harmonics.

Grade, coating stress and clamping pressure all affect the level. A noise complaint on a transformer often ends in a material or assembly decision made years earlier.

The core is one component in a power conversion assembly whose other parts carry their own constraints, described in this overview of power electronics devices and modules, which is useful context when the acoustic requirement belongs to the whole unit rather than the core.

What to specify and what to test on arrival

Specify the grade by its standard designation, the coating type, the nominal thickness and the permitted burr height after cutting.

Add the annealing window, the packing and edge protection, and the certificate requirement per coil with the coil number traceable to the delivery.

On arrival, check thickness, coating surface resistance on a sample, and flatness. A coil that has taken a set in transit will fight the stacker and the loss will follow.

Supply is concentrated upstream, and the chain matters commercially

Grain-oriented steel is produced by a small number of mills, and capacity additions take years because the rolling and annealing lines are specialised. Availability therefore moves independently of transformer demand, and allocation is often decided before price is discussed.

The electricity transmission and distribution build-out has tightened the balance, which is why lead times for the steel and for the finished transformer have moved together.

The supply picture and the demand drivers behind it are set out in this analysis of the grain-oriented electrical steel market, which is a reasonable starting point when a project schedule depends on mill allocation rather than on price.

FAQ: questions buyers ask

Why is my core loss higher than the grade allows?

Because the grade is measured on a standard sample. Cutting burr, coating damage, clamping stress and the annealing profile all add loss, and in most investigations the process contributes more than the steel.

Does a higher grade always pay back?

Only when the loss saving is valued over the asset life and the assembly can preserve it. A high grade cut with worn tooling and stacked with damaged coating delivers less than its datasheet and costs more than the grade below it.

Can I anneal to recover cutting stress?

Yes, within the window the coating allows. The temperature and atmosphere must match the coating type, and the result should be verified by measurement before and after rather than assumed from the furnace setting.

How do I check coating integrity in receiving?

A surface resistance test on a sample plus a visual inspection at the coil edges, where transit damage appears first. Record the result against the coil number so a later core problem can be traced to a delivery.

What causes transformer noise at twice line frequency?

Magnetostriction in the core, amplified by clamping and by the mechanical resonance of the assembly. Grade, coating stress and the stacking pressure all influence the level, and changing one of them alone rarely solves it.