Core Design
How to Specify a Custom Nanocrystalline Core: RFQ Template, AL, Permeability & Acceptance Tests
Who this guide is for
- • Engineers writing a purchase-ready custom core specification
- • CMC / EMI, metering and protection CT, RCD, and 20–100 kHz HF transformer designs
- • Large toroidal power and CT cores where catalog sizes fail
What you will leave with
- • Enforceable μ / AL test language manufacturers can quote against
- • Application kits, R / Z / F loop choices, and realistic drawing rules
- • A copy-paste RFQ specification that drives FAI and acceptance tests
A custom nanocrystalline transformer core or nanocrystalline toroidal core is specified when stock geometry, core AL value, initial or effective permeability, loop shape, or loss acceptance cannot be met from a catalog part. This guide shows how to write a custom core RFQ for a common-mode choke core, current transformer core, RCD sensor, or mid-frequency power transformer so the manufacturer can quote, build, and test against enforceable conditions.
Designing around a nanocrystalline core is more than picking a material grade. The core must fit the mechanical envelope, support the winding scheme, and deliver magnetic performance the circuit can verify. This is the process we use at CenturaCores when turning operating conditions into a quoteable, buildable specification.
Choose your path
Leads nanocrystalline core material selection, custom geometry qualification, magnetic acceptance testing, and manufacturing-aware design workflows (including CoreMagna AI). IEEE member; focuses on EV, EMI, metering, and industrial power magnetics. See the full author profile.
Published August 19, 2026 · Technically reviewed August 2026 by Manoj Kumar, Chief Technologist · Standards references checked against current applicable editions of IEC 60404 / IEC 61869-2 / IEEE C57.13 · ~18 min read
1. What "custom" actually changes (and what it does not)
Nanocrystalline cores start as melt-spun Fe-based ribbon (typical FINEMET-class chemistry: Fe–Si–B–Nb–Cu)1, rapid-solidified in the roughly 18–35 μm ribbon-thickness class used across commercial grades2, wound to shape, then magnetic-field annealed to form an ultrafine grain structure on the order of ~10 nm. That anneal, not the winding alone, sets permeability, loss, and hysteresis loop shape.
Design-validation note: At 50–100 kHz and above, classical eddy-current loss rises with ribbon thickness, flux density, waveform, and temperature. Many designs still use 20–25 μm ribbon successfully; when measured loss or temperature rise approaches the budget versus ferrite, specify a thinner class (often ≤ 18 μm) and lock core-loss acceptance at the real (B, f, T) point rather than treating thickness as a categorical rule.
| Parameter | Decided at custom stage | Fixed by material / base process |
|---|---|---|
| Dimensions | OD, ID, HT, shape (toroid, oval, C-cut, etc.) | Base ribbon alloy family |
| Magnetic spec | Target μi / AL at defined (f, T, H) | Bs ≈ 1.2–1.25 T @ 25 °C for common Fe-based grades2 |
| Hysteresis | Loop shape (R, Z, F) via field anneal | Density ≈ 7.2 g/cm³ (typical published Fe-based grades) |
| Mechanical | Case type, epoxy, impregnation, gap | Post-anneal material brittleness |
Centura note: Stock sizes cover many CMC/CT envelopes. Custom pays off when you need a non-catalog window, a locked μi window at temperature, a defined cut gap, or a large power/CT geometry catalogs do not carry.
2. Start from the magnetic job, not the catalog part number
Write one sentence that locks the core's job. Everything else follows from it.
| Function | Magnetic priority | Spec must include |
|---|---|---|
| CMC / EMI | High Zcm; withstand imbalance / surge | Iline, Zcm or IL vs f, ΔI / fault I, Tmax, case class |
| Metering CT | Low Im, low phase error vs burden | Class (0.1 / 0.2S / 0.5), burden VA, Is, f, temp |
| Protection CT | Predictable excitation to ALF / Vk | Class (5P/10P/PX), ALF or Vk, Rf, DC offset margin |
| RCD / leakage sensor | High μ at mA-level excitation | IΔn, frequency content, asymmetry, noise floor |
| HF power transformer | Loss at B, f; thermal Ae·Wa | P, fsw, topology, Bpk target, ΔT, winding scheme |
If function is ambiguous ("high μ core for filter"), manufacturers cannot choose loop shape, anneal, or test method correctly.
3. The quantities manufacturers actually quote against
Do not specify "μ = 80,000" alone. Specify what is measured. Nominal μ without test conditions is not enforceable.
3.1 Geometry → magnetic path → AL
Notation used below:
- Kstack: ribbon stacking / fill factor (geometry)
- kw: waveform form factor (volts-to-flux), e.g. 4.44 sine, 4.0 square-ish
For a toroid (approximate):
le ≈ π · (OD + ID) / 2
Ae = Kstack · ((OD − ID) / 2) · HT
AL = (μ0 · μe · Ae) / le L = N² · AL
CenturaCores design starting point: use Kstack = 0.75–0.85 for typical epoxy-impregnated nanocrystalline toroids in preliminary sizing unless measured on the finished part. Do not assume 1.0.
Spec practice: quote either a minimum AL at stated conditions, or a minimum μi / μe at stated f, H (or B), and T, plus dimensional tolerances.
3.2 Application μ bands (CenturaCores design starting points)
CenturaCores engineering guidance: for preliminary sizing only. These bands reflect grades and anneals we commonly recommend; they are not guaranteed material specifications. Final μ / AL must be verified on the specified part under the agreed test conditions.
| Application intent | Typical μi / μe band | Notes |
|---|---|---|
| Gapped power / PFC-ish | ~5k–20k effective | Gap dominates μe |
| Medium μ / broadband | ~30k–80k | EMI and many HF designs |
| High μ CMC / CT / RCD | ~80k–120k+ | Sensitive to anneal and stress |
| Ultra-high μ specialty | 100k–several×100k | Tight process; define H and f carefully |
Custom RFQs should pick a window (e.g. μi ≥ 80,000 @ 10 kHz, 0.1 A/m, 25 °C), not a single marketing number.
3.3 Saturation and design flux
For common Fe-based nanocrystalline grades used in CT/EMI/power magnetics, published saturation induction is typically about Bs ≈ 1.2–1.25 T @ 25 °C2. Do not write "1.7 T" unless you are on a different alloy family with datasheet proof. The design Bpk table below is CenturaCores RFQ-review guidance, not a standards limit.
| Use case | Typical design Bpk | Why |
|---|---|---|
| Metering CT | ≤ ~0.2–0.5 T at In + burden | Keep Im and phase error low |
| Protection CT | Toward ~1.0 T with margin for fault / ALF | Saturation under offset |
| CMC | Check differential / imbalance AT | Ferrite's lower Bs (~0.3–0.5 T class) often forces larger area or more turns under the same imbalance ampere-turns; always check both materials against the actual ΔI and thermal budget |
| HF transformer | Often 0.2–0.6 T | Loss ∝ ~fα Bβ |
Peak flux (waveform form factor kw)
Here kw is the waveform form factor (~4.44 sine, ~4.0 square-ish), not the stacking factor Kstack.
3.4 Core loss: give the operating point
Nanocrystalline loss is low, but RFQs fail when they omit (B, f, T). Classical Steinmetz-type models of the form Pv ≈ k · fα · Bβ are useful for relative sizing only. Fitted k, α, and β depend strongly on grade, excitation range, temperature, waveform, and unit system, so we do not publish fixed public coefficients here. Require measured W/kg or W/set at the real operating point.
Acceptance language example: "Core loss ≤ X W at 0.3 T, 50 kHz, 100 °C, sinusoidal excitation, per agreed method aligned with IEC 60404 measurement practice / agreed fixture."3
4. Application-specific specification kits
4.1 Common-mode choke (CMC)
Must-have fields
- • Line current (Arms) and peaks
- • Required Zcm or insertion loss vs frequency (min curve or table)
- • Max differential / imbalance / surge current the core must not saturate under
- • Turns, wire, bifilar / multi-line notes
- • Ambient + hotspot; case (e.g. PBT Class B) if PCB-mounted
- • Target μi or AL with conditions
Nanocrystalline CMC value is high μ (often ~80k–90k class on our CMC series, as a CenturaCores starting band) and Bs ~ 1.2 T versus typical Mn-Zn ferrite ~0.3–0.5 T. Spec the imbalance, or the choke can pass a μ test and still fail EMC under load. For unbuffered CMCs or magnetics with DC, prefer a flat / transverse-field-annealed characteristic (our RFQ shorthand: F-loop; see Section 6).
Lcm ≈ Ztarget / (2π · ftarget)
Then back-solve N and AL.
See CMC cores and EMI filter cores.
4.2 Current transformer (metering / protection)
Must-have fields
- • Ip / Is, accuracy class
- • Burden VA (meter + wiring + auxiliaries), not nameplate only
- • f (50/60 Hz) and harmonics expectation
- • Protection: ALF or knee-point Vk, Rf, DC offset assumption
- • Dimensional envelope around busbar / cable
Minimum Ae (sinusoidal, first pass)
CenturaCores preliminary sizing rule: often start with Bmax ≈ 1.0 T as an area check, then derate for metering accuracy. Adding ~10% area for metering and ~20% for protection (DC offset) is a useful first pass only. Final margins follow IEC 61869-2 / IEEE C57.13 requirements, burden, ALF or Vk, and the system fault study4, not this heuristic alone.
Full worked method: CT selection guide.
4.3 Mid-frequency power transformer (20–100 kHz)
- • Topology, Vin/Vout, power, duty / waveform
- • fsw and expected harmonics
- • Isolation / creepage driven window loss
- • Target efficiency or allowable core loss + ΔT
- • Toroid vs cut C/E for winding and leakage
- • Ribbon thickness class (20–25 μm typical; ≤ 18 μm when loss at 50–100 kHz+ must stay competitive)
Cross-check with the EV charger core guide for OBC / DCFC.
4.4 RCD / residual current sensors
Specify excitation at mA-level, expected frequency content of fault/leakage, and whether the sensor is voltage or current output. High μ and low magnetizing current dominate; mechanical stress after anneal can destroy the μ you paid for.
5. Shape, cut cores, and drawing rules that prevent scrap
| Shape | Prefer when | Spec traps |
|---|---|---|
| Toroid | Lowest leakage, CMC/CT, HF | ID must clear windings + insulation; state ID after case |
| Oval / rectangular | Flat packs, busbars | Define reference axes and corner radii |
| C / cut cores | Easy bobbin winding, gaps | Gap length, mating flatness, impregnation |
| E / bar | Legacy tooling, special | Joint reluctance dominates μe |
Wound ribbon cores are not machined steel. On the drawing, show OD/ID/HT (or A/B/C) before and after coating/case if both matter, max OD / min ID after finish, cut gap g, mating finish, and clamp force limits.
Effective permeability with a gap
DIY / adjustable C-pairs with 0–2 mm gaps can swing AL across roughly ~25–200 nH/turn² class ranges. Production parts should use a fixed, process-controlled gap.
6. Hysteresis loop / anneal: R, Z, and F (specify, don't imply)
Field anneal sets anisotropy and loop shape. Labels such as R / Z / F are common supplier and manufacturing shorthand (including CenturaCores RFQs). They are not a universal IEC/IEEE naming standard. Always pair the letter with physical requirements: remanence ratio, anneal field orientation, and DC-bias or pulse behavior.
| Loop | Character | Typical use |
|---|---|---|
| R-loop (round) | Lower remanence, softer knee, low loss | Power, many filters, general magnetics |
| Z-loop (square) | High remanence / squareness | Switching, pulse, some specialized sensors |
| F-loop (flat) | Transverse-field annealed; low remanence (Br/Bs < 10%); near-constant μ under DC bias | Unbuffered CMCs, power inductors with DC, bias-tolerant EMI |
RFQ line: "Loop / anneal characteristic: R (round, longitudinal) / Z (square) / F (flat, transverse-field, low Br/Bs, DC-bias tolerant); meet μ and Br/Bs limits below under stated test conditions."
Annealing furnace capacity is often the lead-time bottleneck. Unrealistically tight μ windows without test H/f force re-anneals and delays.
7. Mechanical, coating, and post-anneal fragility
After crystallization anneal, cores are high-performance and brittle. Micro-cracks and clamp stress shift μ and loss even when dimensions look fine. Specify finish (bare / epoxy / varnish / PBT case), impregnation, clamp limits, packing, and handling. For 50–100 kHz+ designs where measured loss approaches the budget, call out a thinner ribbon class (often ≤ 18 μm) as a validation option, not an automatic requirement. See also our packaging guide.
8. Quality, tests, and acceptance
Ask for tests that match the design, not a generic certificate of conformity. CenturaCores quality procedures cover material, dimensional, and electrical checks used for custom cores.
| Parameter | Typical method | Spec as |
|---|---|---|
| Dimensions | CMM / gauges | Drawing tolerances |
| AL or μ | LCR / mutual inductance at agreed f, H, T | Min (and max if needed) |
| Core loss | Wattmeter / agreed HF fixture | Max at (B, f, T) |
| Excitation / Vk (CT) | Secondary excitation curve | Vk, Im @ Vk |
| Insulation (cased) | Megger / hipot as agreed | Min MΩ / V |
| Visual | Cracks, epoxy voids, cut mating | AQL |
Compare CoC numbers to the same conditions written on the PO. μ @ 50 Hz, 0.4 A/m, 25 °C is not μ @ 100 kHz, elevated T.
9. Copy-paste custom core specification (technical contract)
Use this as the body of the RFQ / drawing notes, or download the full template as PDF, DOCX, RTF, or TXT from the RFQ template resource page. Attach a dimensioned PDF; for oval/rect include datum corners. Prefer a filled template back from us? Request a quote with your drawing.
CUSTOM NANOCRYSTALLINE CORE SPECIFICATION Buyer: ____________________ Date: __________ Rev: ____ 1. FUNCTION [ ] CMC/EMI [ ] Metering CT [ ] Protection CT [ ] RCD [ ] HF transformer [ ] Other: ____ One-line duty: _______________________________________________ 2. ELECTRICAL / MAGNETIC OPERATING POINT Frequency: ____ Hz / kHz (range: ____ to ____) Waveform: [ ] sine [ ] square [ ] PWM Duty: ____ Form factor kw (info): ____ (e.g. 4.44 sine / 4.0 square) Bpk or V·s: ____ T or ____ V·µs Margin for transient: ____% Temp: ambient ____°C hotspot ____°C Max core loss: ____ W (or ____ W/kg) at B=____ T, f=____, T=____°C Ribbon thickness class: [ ] 20–25 µm [ ] ≤18 µm [ ] other: ____ 3. PERMEABILITY / AL (enforceable) Target: µi or µe ≥ ____ OR AL ≥ ____ nH/N² Test: f=____, H=____ A/m (or B=____ mT), T=____°C, fixture: ____ Loop type: [ ] R-round [ ] Z-square [ ] F-flat (low Br, DC bias) Br/Bs limits if any: ____ 4. APPLICATION EXTRAS CMC: Iline=____ A; Zcm/IL table attached; max ΔI/surge=____ A CT: class=____; burden=____ VA; Is=____ A; ALF/Vk=____; Rf=____ kA RCD: IΔn=____ mA; bandwidth=____ 5. GEOMETRY Shape: [ ] toroid [ ] oval [ ] rect [ ] C-cut [ ] E [ ] other Drawing No: ________ OD/ID/HT or A/B/C: ________ Gap (if cut): g=____ mm; faces: ____ Finish: [ ] bare [ ] epoxy [ ] case ______ Color/marking: ____ Stacking factor assumption for design: Kstack=____ (info only) 6. ENVIRONMENT / MECHANICAL Mounting: ____ Clamp limits: ____ Potting/PCB notes: ____ Packaging: export anti-vib / moisture barrier required: [ ] Y [ ] N 7. QUALITY Reports required: dimensions, µ/AL, loss, (CT excitation), FAI Sampling: ____ 8. COMMERCIAL Qty: proto ____ / prod ____ Target lead time: ____ Ship to: ____
10. Failure modes we see weekly (spec mistakes)
- Nominal OD only, no tolerances / no after-case ID → windings do not fit.
- μ without (f, H, T) → CoC passes, choke fails at temperature.
- CMC specified without imbalance current → saturates under real load.
- CT burden = meter VA only → wiring eats the accuracy class.
- Wrong Bs myth (1.7 T) in thermal/size calc → undersized for common ~1.2 T Fe-based grades.
- Loop type omitted (or R/Z/F used without stating remanence / anneal / DC-bias needs).
- Ignoring post-anneal brittleness → μ drop after shipping/clamping.
- Over-tight μ window + vague test → anneal queue delay and rejects.
- Assuming 20–25 μm ribbon is automatically inadequate at 100 kHz+ without measuring loss at the real (B, f, T) point.
11. Worked mini-examples (RFQ-grade)
Example A: CMC core sketch
- • Iline 16 A, Zcm ≥ 1 kΩ @ 100 kHz, Tmax 105 °C, PCB case
- • High-μ toroid, μi ≥ 80,000 @ agreed LCR; check Bs for ΔI = 2 A imbalance
- • If DC or strong bias is present, specify F-loop (flat / transverse, low remanence) with Br/Bs limits
- • Output: OD/ID/HT + AL min + case drawing + IL vs f table on PO
Example B: 400/1 A, 5P10, 15 VA
- • Vs @ ALF ≈ √15 × 10 ≈ 38.7 V
- • N = 400, f = 50 Hz, Bmax = 1.0 T → Ae ≈ 44 mm²; apply CenturaCores preliminary ~20% protection area check → ≥ 53 mm² before standards review
- • Spec: nanocrystalline toroid, Ae ≥ 53 mm², excitation curve, class 5P10
Example C: 50 kHz, 0.3 T transformer core
- • Core loss ≤ X W @ 0.3 T / 50 kHz / 100 °C
- • Prefer round / low-loss anneal (R-loop shorthand); define Ae from V/N, Wa from copper + isolation
- • Validate ribbon thickness against measured loss; consider ≤ 18 μm if the loss budget is tight near 100 kHz
12. Frequently asked questions
What does Centura need to quote a custom nanocrystalline core?
Function, operating (B or V·s, f, T), enforceable μ/AL test conditions, loop type (R / Z / F shorthand plus remanence or anneal requirements), dimensioned drawing, finish/gap, ribbon thickness class when HF loss matters, acceptance tests, qty/lead time. Uncertain grade: we recommend from the application kit above.
Which shapes are available for custom nanocrystalline cores?
Toroid, oval, rectangular, C/cut, E, and bar. Many are made directly from customer drawings with datums and tolerances.
What frequency range is realistic?
EMI cores often from tens of kHz into hundreds of kHz / ~1 MHz class depending on design; HF power magnetics commonly 20–100 kHz. Above ~50–100 kHz, validate measured loss at the real (B, f, T) point and consider a thinner ribbon class when eddy current approaches the loss budget.
How is a nanocrystalline core manufactured?
Melt-spun thin ribbon → slit/wind → magnetic-field anneal (crystallization + anisotropy) → finish/case/gap → electrical test. Brittleness after anneal drives packaging and handling rules.
What quality documentation should I request?
Request lot test data matched to your PO conditions (dimensions, μ/AL, loss or CT excitation, and FAI as applicable). See the quality page for testing and acceptance context.
References and standards
Numerical material claims above should be read with the cited literature and the edition of each standard applicable to your project. Always confirm against the supplier datasheet for the exact ribbon grade and anneal.
- Yoshizawa, Y., Oguma, S., Yamauchi, K. "New Fe-based soft magnetic alloys composed of ultrafine grain structure." Journal of Applied Physics, 1988. Foundational FINEMET-class Fe–Si–B–Nb–Cu nanocrystalline alloy description.
- Commercial Fe-based nanocrystalline ribbon datasheets (e.g. FINEMET / equivalent Fe–Si–B–Nb–Cu families): typical Bs near 1.2–1.25 T @ room temperature and ribbon thicknesses in the tens of micrometers; confirm the exact grade used on the PO.
- IEC 60404 series, Magnetic materials – Methods of measurement of magnetic properties. Use the parts applicable to your specimen geometry and frequency when writing loss / permeability acceptance methods.
- IEC 61869-2, Instrument transformers – Additional requirements for current transformers; IEEE C57.13, Standard Requirements for Instrument Transformers. Accuracy class, burden, ALF, and related CT requirements belong in the RFQ when specifying CT cores.
For CenturaCores manufacturing test and acceptance context, see the quality page. Match lot test reports to your PO conditions.
Notation quick reference
| Symbol | Meaning |
|---|---|
| Kstack | Ribbon stacking / fill factor (~0.75–0.85 typical) |
| kw | Waveform form factor (~4.44 sine, ~4.0 square) |
| le, Ae | Mean magnetic path length, effective core area |
| AL | Inductance factor (H/turn² or nH/N²) |
| μi, μe | Initial / effective relative permeability |
| Bs, Bpk | Saturation induction / design peak flux density |
| R / Z / F | Supplier shorthand for round / square / flat (transverse) anneal characteristics; always define physical remanence and bias requirements |
Next step inside CenturaCores
Run geometry and turns in the designer, attach the completed spec block and drawing to a quote request, and ask for FAI plus μ/AL and loss (or CT excitation) on the stated operating point.
Request a Custom Quote
Send the RFQ block and drawing for a buildable custom core quote.
Request Quote