CRGO lamination for reactors and inductors: design considerations

1. Start from the waveform, not the grade

A lot of CRGO-laminering content quietly assumes a near-sinusoidal voltage and a clean magnetization loop. Reactors and many inductors don’t live there.

• Line reactors / shunt reactors – almost sinusoidal, but with non-negligible harmonics and sometimes strong DC bias from converter imbalance.
• DC chokes / PWM inductors – current is ripple on a DC level; flux is a mix of offset plus triangular or trapezoidal swing.
• Medium-frequency magnetics – square or quasi-square excitation, sometimes at kHz range.

Grain-oriented steel behaves differently under these conditions than in the 50/60 Hz, sine-wave test used in standard loss ratings. A recent study on GOES wound cores at ~2 kHz even shows specific losses lower for square voltages than for quasi-sine at the same peak flux, because the harmonic content shifts where eddy currents concentrate in the strip.

So before picking “M3, 0.27 mm” out of habit, lock down:

• Real waveform at the core (not the ideal drawing)
• Peak flux density including transient over-shoot
• DC bias level over life
• Frequency range, including any interharmonics

Everything else—stacking factor, joint style, gap scheme—hangs off those four.

2. Flux density ranges that actually work in CRGO reactors and inductors

Datasheets will happily quote saturation around 1.9–2.0 T for grain-oriented electrical steel, with a reasonably linear region up to roughly 1.2 T.

In practice for power reactors and iron-core inductors you rarely want to be that brave.

Typical working bands

These are indicative, not a substitute for your own B-H curves and lifetime model:

A classic cut-core design guide for grain-oriented steel shows useful “linear enough” behavior up to ~1.2 T even under DC bias if the gap is chosen correctly.

Voor line and shunt reactors, you usually run closer to transformer practice, but:

• Include DC bias from system imbalance and control offsets.
• Overweeg short-time overloads from fault conditions and tap changes.

Voor inductors in switching supplies, you’ll normally accept lower B_peak because:

• You’re pushing into higher frequency where core loss climbs.
• The winding window is often the real bottleneck.

Rule of thumb that keeps projects out of trouble: Design first against B_{max,hot,biased}, not room-temperature B_max. Then check whether the grade you wanted still makes sense.

3. Lamination stack: stacking factor, burrs, and real cross-section

Everyone writes “stacking factor 0.96” on the slide. Reality is messy.

What stacking factor really changes

Stacking factor directly hits the effective iron cross-section. Lower factor → less steel → higher flux density than you thought → early saturation and extra loss. A standard magnetic-core handbook points out that misaligned burrs and poor insulation between laminations can easily erode stacking factor enough to matter at power levels where CRGO is used.

Belangrijkste punten:

• Punching burr orientation – If burrs all face one way in the stack, the solid “bridge” region is localized. If they are random, interlaminar contact spreads everywhere and both stacking factor and eddy-current loss degrade.
• Coating thickness – Better coating = better interlaminar resistance, but slightly worse stacking factor. Steel mills and standards encode this tradeoff through coating classes.
• Pressing pressure and flatness – Non-flat laminations create micro-gaps. A GOES technical data sheet explicitly stresses the need for lamination flatness during annealing and stacking to avoid residual stress and unplanned gaps.

Voor reactor cores, stacking factor is slightly more forgiving than in high-efficiency transformers, because many designs are already gap-dominated. But once you move into high-flux, low-loss HV shunt reactors, small errors in effective area show up as extra watts and unexpected hotspot locations.

Numbers to put in your internal checklist

You don’t need to put all of this in the RFQ, but design around:

• Assumed stacking factor for calculation: 0.94–0.96 for high-quality thin CRGO with good coating; 0.90–0.93 if you know stamping is rougher or thickness is higher.
• Maximum burr height at stamping: usually a few percent of sheet thickness at most; confirm with the lamination vendor, because this drives how aggressive you can be.
• Pressing / clamping scheme: single yoke press for small cores vs distributed clamping pads to avoid bending the limbs.

If you are reusing a transformer lamination tool for a reactor, double-check that the Echt stack height after coating and pressing still matches the magnetic design. It often doesn’t.

4. Joint style and step-lap behavior in reactor cores

CRGO lamination blogs spend a lot of time on step-lap for transformers. The physics carries over to reactors and inductors, just with different priorities.

• Step-lap joints distribute flux more smoothly across overlapping steps, reducing local flux peaks, core losses, and audible noise.
• Butt or non-mitred joints are simpler, can be cheaper, but concentrate flux and magnetostriction at the joint.

In reactors:

• Voor HV shunt reactors en large line reactors, step-lap is usually justified: less local saturation at peak flux, less sensitivity to tolerances in joint machining, easier time with noise specs.
• Voor small inductors and chokes, a simpler joint may be fine, because the gap dominates the reluctance and the joint region is not the main bottleneck.

Whatever joint you use, make sure your drawing and RFQ talk about:

• Overlap length and tolerance (for step-lap)
• Joint machining flatness if a cut core is used
• Whether the joint is treated as part of the deliberate air-gap or intended to be as close to zero as possible

Leaving the joint strategy “implicit” often ends with the supplier using their transformer default, which may not fit your reactor’s DC bias and waveform.

5. Gaps and discretely distributed gaps in CRGO reactors

Gaps are where reactor cores quietly generate extra loss.

Concentrated vs distributed gaps

Academic work on iron-core shunt reactors with discretely distributed air-gaps compares:

• a single global gap per limb, and
• multiple smaller gaps distributed along the laminated limb.

It shows how distributing the gap can adjust inductance, leakage inductance, and loss separately, and how fringing around each gap adds local eddy-current loss.

For power reactors, this leads to a few design levers:

• Single large gap – simple to build, but strong fringing; high local loss and heating around the gap if the winding is too close.
• Multiple smaller gaps – lets you smooth flux, shape leakage, and sometimes reduce local hotspot severity, at the cost of more complicated stacking and machining.

For inductors, a classic iron-core design guide for C-cores emphasizes:

• Gap length dominates inductance once the core is highly permeable.
• Fringing effectively shortens the gap; the simple L ≈ N²μA/lg equation inflates inductance if you ignore it.

So, don’t leave gap geometry vague.

Some practical notes for CRGO lamination stacks with gaps

• Non-magnetic spacers (e.g. fiberglass, stainless) should be called out by material and thickness, not just “insulating shim”.
• Edge chamfers near the gap reduce sharp fringing peaks. Small detail, but helps for long-life HV gear.
• Minimum distance from winding to gap: specify an electrical + thermal clearance. Fringing-induced hotspot on the innermost turns is a common failure root cause.

And no, a “typical transformer gap practice” sentence in the spec is not enough when your reactor is expected to operate close to saturation under DC bias.

6. Magnetostriction, vibration, and noise in reactor laminations

Most noise articles target transformers, but the same magnetostriction phenomena show up in largish reactors and inductors: laminations strain slightly as flux reverses, and the stack vibrates.

Recent engineering-oriented notes on CRGO magnetostriction make a few points that carry straight into reactor and inductor stacks:

• Magnetostriction varies noticeably across CRGO grades and processing routes.
• Noise is not just material; lamination geometry, stack design, and clamping all turn that strain into real sound.
• Flux density, harmonic content, and DC bias are the main knobs.

For reactors:

• Line and shunt reactors near populated areas may have acoustic limits similar to transformers, especially in substation buildings.
• Reactors in industrial plants still care, but the surrounding equipment can mask a lot; thermal and loss limits dominate instead.

Design checklist for the stack:

• Avoid very sharp local flux peaks at joints; step-lap helps here.
• Gebruik uniform clamping pressure so laminations don’t rattle against each other.
• If noise is a hard constraint, consider specifying a lower-magnetostriction CRGO tier and document test conditions (frequency, induction, mounting) so the supplier’s and your measurements match.

7. Thermal behavior: steel, stack, and cooling path

CRGO has reasonably high thermal conductivity and a high Curie temperature (often around 730 °C for standard grades).

Two consequences that matter in reactors/inductors:

1. The core can safely run hotter than the windings, thermally speaking. Work on GOES wound cores shows lower core losses at elevated temperatures, thanks to increased resistivity.
2. Your hot-spot model needs to acknowledge that oil, air, and structural steel all influence the temperature gradient across the lamination stack.

For lamination stack design:

• Don’t block every axial cooling path with solid clamps; leave some thermal “chimneys” through the stack.
• When using epoxy or bonding, check its thermal conductivity and temperature rating, not just mechanical strength.
• In oil-immersed reactors, lamination stack geometry can guide oil flow. Rounded edges and reasonable clearances help avoid stagnant pockets.

Thermally, CRGO will usually forgive you. The winding insulation system won’t.

8. What to actually specify in the RFQ for CRGO lamination stacks (reactors & inductors)

Most RFQs specify grade, thickness, and coating, maybe “step-lap”. Standards guides point out that grade codes and loss tables only tell half the story; the rest sits in how the laminations are turned into a core.

For reactors and inductors, add some precision.

8.1 Steel and basic geometry

Specify:

• Material class – e.g. grain-oriented electrical steel with a named grade or loss band at a reference induction and frequency.
• Dikte – 0.23 / 0.27 / 0.30 mm etc.
• Coating type – high-resistance vs mechanical-robust, and whether it is compatible with your annealing and oil or varnish system.
• Core geometry – EI, UI, C, toroidal, or custom stacked block, with all critical dimensions and tolerances.

8.2 Stack and joints

Include:

• Target stacking factor and how it will be verified (mass vs theoretical volume, or dimensional checks).
• Maximum burr height after punching/laser.
• Joint method – step-lap or not; overlap length and sequence if step-lap is required.
• Whether stress-relief annealing after cutting/stacking is required; some manufacturing processes include a final anneal that recovers much of the steel’s magnetic performance.

8.3 Gaps and machining

For gapped CRGO cores:

• Total gap length and distribution (single vs multiple gaps).
• Machining tolerance on each gap.
• Spacer material and their tolerances.
• Any edge treatment near the gap to control fringing.

8.4 Tests and acceptance

You don’t need a million tests. But define a small, clear set:

• Core loss and magnetizing current at a named induction, frequency, temperature, and waveform.
• Dimensional checks on limb lengths, stack height, and joint alignment.
• If noise matters: a simple acoustic test condition (mounting, distance, frequency, induction).

This way, if a reactor later runs hot or saturates early, you can tie it back to either design assumptions or stack execution, without guessing.

9. Quick internal checklist before you sign off a CRGO lamination stack for a reactor/inductor

Not exhaustive, but it catches many of the issues that show up late:

1. Did we size B_max for the real waveform and DC bias, at operating temperature?
2. Is the assumed stacking factor backed by a realistic manufacturing route?
3. Is the joint style (step-lap or not) aligned with our flux density and noise goals?
4. Is the gap scheme consistent with inductance, leakage inductance, and loss targets? 
5. Have we written down gap and joint tolerances that a shop can realistically hold?
6. Is the RFQ explicit enough that two different lamination suppliers would build essentially the same stack?

If any answer is “not sure”, that’s usually where future failure analyses come from.

FAQ: CRGO laminations in reactors and inductors