How to Reduce Transformer No-Load Losses with Better CRGO Lamination Choices

This article is about the decisions between your spec sheet and the lamination stack on the shop floor – where a quiet 5–25% of no-load loss is usually hiding.

1. Don’t start with “M4 vs M5”. Start with W/kg at a realistic flux level

Most specs still say something like:

CRGO, 0.23–0.27 mm, max loss X W/kg @ 1.5 T, 50 Hz.

On paper that sounds fine. In practice:

• Sheet loss is guaranteed on Epstein or single-sheet test samples.
• Your assembled core loss = sheet loss × building factor (BF), usually > 1.

If you only control the grade name, you’re not really controlling no-load loss.

A few practical anchors:

• Modern CRGO for dry-type and oil transformers typically runs 0.9–1.5 W/kg @ 1.5 T, 50 Hz, 0.23–0.30 mm.
• Some domain-refined, high-induction CRGO grades quote 0.80 W/kg in the same test conditions.

Instead of “M3 equivalent”, write your spec around:

1. Target core loss of the assembled core (W @ rated voltage), aligned with your efficiency class.
2. Max allowed building factor (e.g. BF ≤ 1.25 at 1.5 T).
3. Test method (Epstein vs single-sheet) and correlation rules between them.

That’s the only way to stop low-cost laminations with pretty mill certificates but ugly assembled losses from slipping through.

2. Material levers that actually move no-load loss

You don’t need a full materials lecture here. Just the levers that change your loss and cost curves in a noticeable way.

2.1 Grade and domain refinement

A 2024 study compared different GOES / CRGO types and lamination thicknesses and found a modern high-permeability grade with optimized domains gave about 66% lower core loss than a reference M6 material at the same flux density.

So, for a given kVA:

• “Regular” conventional CRGO might give you a baseline of 100% loss.
• High-induction CRGO trims that down.
• Domain-refined CRGO trims it further again – often the only realistic path if your no-load loss target is very aggressive but you don’t want amorphous.

Key point: Don’t treat “domain refined” as a marketing tag. Ask for:

• Loss guarantees by grade (e.g. C23QH080 vs C23QG085).
• The test standard and flux density used.
• Whether the domain treatment survives any post-processing heat cycles you or your supplier apply.

2.2 Lamination thickness

Thinner laminations cut eddy current paths. You know that already. The trick is to tie it to frequency and economics, not fashion.

Common ranges:

• 0.30 mm – still widely used, cost-friendly.
• 0.27 mm – typical mid-loss compromise.
• 0.23 mm – for low-loss or higher-frequency work; material cost and processing demands go up.

In many utility specs, the thinner gauges are effectively required just to meet modern ecodesign and DOE/IS 1180 style limits.

So instead of arguing forever about “0.23 vs 0.27”, map:

• Loss target at your operating flux.
• Price / kg delta for thinner steel.
• Impact on copper loss if you change cross-section.

Then pick the cheapest combination that still hits the assembled no-load loss target, not just the sheet W/kg.

2.3 CRGO vs amorphous (briefly, because the title says CRGO)

Amorphous cores can cut no-load loss by about 60–70% compared with CRGO.

For high-load power transformers, CRGO + disciplined lamination stacks is still the main workhorse. For lightly loaded distribution units, amorphous is often the right answer and this whole article becomes a bit academic. Just worth admitting that upfront.

3. Stack geometry: where “nice drawings” quietly add 5–25% loss

Now to the lamination stacks themselves – where a lot of vendors quietly decide your BF.

3.1 Packets per stack and building factor

There’s a temptation to build cores with multi-layer packets (2–3 laminations per stack) to speed up assembly.

A well-known 1000 kVA study looked at CRGO cores with 1, 2, and 3 laminations per stack (same geometry, 0.3 mm M5 grade). Result at 1.5 T, 50 Hz:

• 2-layer stacks: ~6.6% higher core loss vs 1-layer.
• 3-layer stacks: ~8.3% higher core loss vs 1-layer.

Flux leakage at corner joints and building factor both worsened with more laminations per packet.

So when a manufacturer tells you “multi-layer stacking doesn’t affect loss much”, ask them:

• For building factor curves vs flux density for different packet counts.
• For a sample no-load test comparison on otherwise identical cores.

Then decide if the assembly time saved is worth 5–8% more iron loss for 30 years.

3.2 Joint type: straight, staggered, step-lap, mitered

Here’s where the geometry of your lamination stacks really starts to pay – or cost – you.

A recent technical write-up on no-load losses listed measured results for different joint forms:

• Stepped joints vs simple staggered joints → Around 6% lower no-load loss for stepped joints in tested configurations.
• Semi-mitered joints (mix of straight and miter) → About 10–15% lower no-load loss vs fully straight joints.
• Full mitered joints with correct grain orientation → 15–25% reduction in no-load loss vs straight joints, plus lower excitation current.

Step-lap construction then spreads the flux transition over several small steps, reducing rotational flux, gaps, and joint hot spots. CRGO step-lap laminations are routinely advertised as “low no-load loss” for exactly this reason.

Design note that sometimes gets ignored:

• Joint form is not just a drawing detail.
• It’s part of your loss budget.

If your target is tight, you basically cannot afford straight butt joints with casual stacking and hope to win.

3.3 Lap width and joint area

Another subtle lever: lap width at the corner.

Evidence from recent transformer core studies says:

• Too wide a lap enlarges the discontinuous flux region and raises no-load loss.
• You need a compromise between mechanical strength and magnetic cleanliness.

So instead of “lap width: as per manufacturer standard”, specify a numeric range and require loss verification at that geometry.

3.4 Cross-section and window usage

Brief but important:

• Poor matching between core limb and yoke cross-sections pushes flux out of the preferred rolling direction and across sheet thickness, raising eddy losses.
• Rectangular cross-sections generally need ~10% more area than optimized multi-step elliptical ones for similar flux distribution.

You don’t need to redesign classical shapes, but you do want your lamination supplier and your mechanical designer talking the same language about:

• Effective stacking factor.
• Cross-section steps.
• How many “non-functional” laminations they sneak in for pack-out.

4. Process discipline: how good CRGO becomes mediocre cores

Even perfect drawings and materials lose the fight if the lamination process is sloppy.

4.1 Burr control and insulation damage

Measured data from an OEM knowledge article is pretty blunt: when burr height exceeds about 0.03 mm, you get:

• Inter-lamination shorting and a short eddy current path
• Higher local flux density and hot spots
• Scratched insulation coating, extra circulating currents

None of this appears in the datasheet. It all appears in your no-load test.

So your RFQ needs:

• Max burr height (e.g. < 0.02–0.03 mm on all edges).
• Defined inspection process (profilometer or equivalent).
• Rejection rules for scratched or flaking coating.

4.2 Mechanical stress and cutting method

CRGO is stress-sensitive. Bending, bad clamping, rough shearing – all expand the hysteresis loop and push loss up.

Lamination manufacturers now advertise:

• Computerized cut-to-length and step-lap machines for consistent overlap and minimal distortion.
• Domain-refined grades that are more sensitive to mechanical abuse, but give lower loss when handled correctly.

If you buy domain-refined steel and then punch a lot of holes or bend corners aggressively, you pay for low loss and then stress it away.

4.3 Coating and insulation system

Manufacturers like JFE Steel or thyssenkrupp Electrical Steel supply CRGO with specific coatings optimised for:

• High inter-laminar resistivity
• Stress relief
• Bonding / stacking behaviour

On your side, the only thing that matters is:

Does the coating system, in your actual stack and clamping arrangement, maintain the promised resistivity and loss?

So:

• Require coating type declaration (C-5, etc.).
• Limit the number of times laminations are re-stacked or re-worked.
• Avoid mixed coatings in one core unless the supplier can show loss data.

4.4 Building factor as a contract item, not an afterthought

Industry practice is slowly moving from “sheet W/kg only” to explicit BF targets.

For example:

• Sheet loss: ≤ 0.90 W/kg @ 1.5 T, 50 Hz.
• Assembled core loss: ≤ 1.10 W/kg equivalent → BF ≤ 1.22.

If your vendor can only hit the sheet spec by over-stressing or dirty stacking, you’ll see it in BF.

That one number quietly ties together:

• Grade
• Thickness
• Cutting and handling
• Stacking quality
• Joint design

And that’s exactly what you want.

5. Quick comparison: lamination stack decisions vs no-load loss

Table is intentionally simple. You already know the equations.

Values are indicative, based on manufacturer data and published technical studies rather than a single test run.

6. A practical CRGO lamination stack checklist for your next RFQ

You can plug this straight into a B2B spec or vendor questionnaire.

A. Material & data sheet

• Target no-load loss of assembled transformer at rated voltage and frequency.
• Max allowed building factor at 1.5 T (and at your actual flux).
• Required CRGO thickness band and grade family.
• Whether domain-refined grades are acceptable, preferred, or mandatory.
• Guaranteed sheet loss values (W/kg) + test standard (IEC/ASTM, Epstein vs SST).

B. Lamination design

• Joint type: full miter / step-lap; no “straight only” by default.
• Allowed lap width range and overlap geometry.
• Number of laminations per stack / packet (1 vs multi-layer).
• Minimum stacking factor and how it’s measured.
• Cross-section strategy (e.g. multi-step elliptical yoke vs simple rectangular).

C. Manufacturing controls

• Max burr height and edge quality limits; measurement method.
• Coating type, curing and any post-processing heat treatment.
• Rules for rejecting bent or stressed laminations.
• Cut-to-length machine capabilities – especially for step-lap and miter accuracy.

D. Verification and testing

• No-load loss and excitation current tests as per IEC 60076 / IEEE C57 at FAT.
• Agreed correction from test voltage to rated voltage (V² relationship).
• Reporting of both sheet loss and assembled core loss (with BF).
• Option for periodic witness tests on a bare core, before windings.

E. Commercial guardrails

• Price adjustments or re-work obligations if BF exceeds an agreed threshold.
• Clear rule for what happens if the mill switches to another CRGO grade mid-contract (with or without domain refinement).

This is how you stop “cheapest CRGO lamination” from silently turning into “highest lifetime iron loss”.

7. FAQ: CRGO lamination stacks and no-load losses