OEM engineering guide
AFPM Core Design Guide: Topologies, Materials & Manufacturing
Use this AFPM core design guide to compare topologies, SMC and laminated steel, cooling paths, loaded air gaps and manufacturing routes before preparing a supplier brief.
By AxialFluxCore · Published · Evidence reviewed
Start with the four decisionsFour critical decisions for AFPM design
Topology comparison: keep the rotor/stator stack explicit
AFIR is a double-stator/single-rotor arrangement. TORUS and YASA place a stator between two rotors; YASA uses a yokeless segmented armature. Confirm the actual flux-return and winding arrangement on the drawing. Topology definitions and comparison.
Axial stack schematics, not winding or magnetic-circuit drawings. Amber = stator; slate = rotor. Arrows show axial order, not flux direction. Gaps and supports are not to scale.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Topology | Axial order | Structural distinction | Decision to resolve |
|---|---|---|---|
| SSSR | Stator → rotor | One active gap; evaluate the net axial bearing load. | Screen when the simple stack fits the duty; verify bearing life and cooling. |
| AFIR / DSSR | Stator → rotor → stator | Internal rotor, two stators; symmetry can balance net rotor pull. | Evaluate two stator assemblies and unequal-gap loading. |
| TORUS / SSDR | Rotor → stator → rotor | Internal toroidal stator between external rotors; yoke/flux path depends on variant. | Define winding arrangement and heat path before quoting the core. |
| YASA / SSDR | Rotor → segmented stator → rotor | Yokeless segmented armature with individually supported teeth. | Compare tooth retention, winding access and cooling alongside active mass. |
Compare torque in N·m at a stated speed and duty. For N·m/kg, state whether the denominator is active material or the complete motor, and whether the figure is continuous or peak. A topology name alone does not set these conditions.
Core materials: compare processed SMC and laminated steel
SMC processing guidance supports 3D flux shaping. Laminations remain a candidate, including for segmented YASA stators. The table below is a procurement and simulation checklist; it is not a measured head-to-head benchmark.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Dimension | SMC powder core | Electrical-steel core |
|---|---|---|
| Flux geometry | 3D flux and pressed tooth shapes are options; check tooling access. | Orient sheets along the intended flux path; detail strip winding or segment stacking. |
| Loss comparison | Use a named powder grade, density, heat treatment and specimen. | Use a named steel grade, sheet thickness, cutting process and stack factor. |
| Magnetizing demand | Import the actual B–H curve into the candidate model. | Do the same; a single permeability number cannot represent saturation. |
| Mechanical integration | Specify support and allowable loads for the processed part. | Specify stack retention, burr control and insulation integrity. |
| Commercial decision | Quote tool cost, cavities, scrap and inspection at the requested volume. | Quote strip/segment tooling, assembly labor and inspection at the same volume. |
Evidence boundary: the manufacturer distinguishes SMC ring tests from electrical-steel sheet tests. Read the comparison-method note, p. 2. Amorphous materials require a separate grade and process review; no equivalent processed-core dataset is established in this guide.
A reproducible material-data example
These are typical manufacturer values for Somaloy 700HR 1P, compacted at 800 MPa, from the 2025 datasheet, with 0.4% Kenolube, a room-temperature die and heat treatment at 530°C in air. The 1 T ring measurements use an OD55/ID45/H5 mm specimen and the cited IEC 60404-6:2003 method. They are not measurements of an AxialFluxCore stator. Source: pp. 4–5.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Frequency | Specific core loss | Interpretation |
|---|---|---|
| 100 Hz | 10 W/kg | Reference for this material specimen. |
| 400 Hz | 43 W/kg | Re-evaluate loss as electrical frequency rises. |
| 1,000 Hz | 125 W/kg | Do not assume all SMC stays below 50 W/kg at 1 T. |
| Matching steel component | Not established in this guide | Request matched test conditions before ranking routes. |
Thermal design: start at the winding and end at the coolant
The cited in-core cooling research evaluates channels inside laminated YASA segments and verifies its modeling with a motorette. It demonstrates an alternative heat path; it does not establish one resin conductivity, power cutoff or temperature reduction for every design.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Candidate route | Boundary to define | Validation needed |
|---|---|---|
| Air / housing conduction | Ambient, airflow, mounting contact and duty cycle. | Measure steady and transient winding temperatures; do not select by rated kW alone. |
| Potting to a cooled support | Bond-line thickness, conductivity, contact area and coolant temperature. | Qualify voids, cure, expansion and insulation; characterize the actual thermal path. |
| In-core or direct cooling | Channel geometry, flow, pressure drop, coolant and seals. | Use a representative segment test before extending the model to the full stator. |
First-order engineering model: R = t / (k × A) for a uniform conduction layer, where t is thickness (m), k conductivity (W/m·K), and A contact area (m²). This gives K/W and excludes contact resistance, voids, spreading and convection. Use it to expose missing inputs, then validate the full network.
Mechanical integrity: budget each active air gap
The rotor structural study connects magnetic loading with axial disc displacement. Translate that principle into a drawing-specific clearance budget; neither a universal ±0.1 mm tolerance nor a fixed deflection percentage is established here.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Term | Assumed allowance (mm) | Owner / evidence to obtain |
|---|---|---|
| Nominal cold gap | 0.80 | Electromagnetic and mechanical design agreement. |
| Closing assembly stack error | −0.10 | Drawing tolerances and measured runout. |
| Bearing axial motion | −0.05 | Bearing and load-case data. |
| Loaded relative deflection | −0.12 | Coupled structural model of both opposing surfaces. |
| Thermal closing displacement | −0.08 | Hot geometry / thermal-expansion model. |
| Remaining modeled clearance | 0.45 | Compare with an agreed minimum and unmodeled allowances. |
Calculation: 0.80 − 0.10 − 0.05 − 0.12 − 0.08 = 0.45 mm. These are assumed maximum closing allowances for one illustrative condition. The arithmetic is not a release decision: add relevant vibration, overspeed, fault and life effects, avoid double-counting, and evaluate both gaps for a dual-sided machine.
Manufacturing: estimate force, then qualify the route
Pressure is not press capacity
For a preliminary axial pressing estimate, F = p × A, with pressure p in MPa (= N/mm²) and projected powder area A in mm². For an illustrative unslotted ring with OD 200 mm and ID 120 mm, A = π/4 × (200² − 120²) ≈ 20,106 mm². At an assumed 600 MPa, F ≈ 12.06 MN, or 1,230 metric tonnes-force (1 tf = 9,806.65 N).
The selected pressure is illustrative within published grade-dependent processing ranges. Slots, cavities, multi-level tooling, press derating, density gradients and ejection must be reviewed by the supplier. Outer diameter alone cannot decide whether segmentation is required.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Stage | Evidence to deliver | Hold point |
|---|---|---|
| Material and process | Grade, density target, heat treatment, B–H/loss data and representative sample. | Material/process changes require a new comparison. |
| Tooth and winding integration | Drawing datum scheme, coil insertion path, support and insulation stack. | Resolve assembly access before releasing hard tooling. |
| Rotor/stator assembly | Calculated forces, fixture layout, controlled separation and retention plan. | Review the actual sequence and fixture load capacity. |
| Validation and RFQ | Inspection plan, magnetic/thermal checks, quantities and acceptance criteria. | Quote unknowns explicitly; agree qualification scope before production. |
Two worked selection scenarios
These are hypothetical OEM briefs and calculated screening inputs, not customer projects or measured performance claims. Electrical frequency f = pole pairs × rpm / 60; inverter switching frequency is a separate input.
Risks, triggers and mitigation
The following is an engineering review register. Severity and acceptance limits must be assigned from the actual duty and failure consequences, not assumed from the topology name.
On narrow screens, swipe the table or focus it and use the arrow keys.
| Risk | Trigger to investigate | Recommended mitigation |
|---|---|---|
| Magnet heating | Slotting and inverter harmonics may produce unacceptable rotor loss. | Model the actual spectrum and magnet segmentation; validate temperature. No universal speed trigger is established here. |
| Tooth or insulation damage | Pressing, ejection, winding or machining changes the part. | Approve representative processed samples; do not assume a powder datasheet describes the finished component. |
| Torque ripple | Pole/slot choice or skew is selected on cogging alone. | Check loaded ripple, average torque, winding factor and manufacturability together. |
| Rotor contact | Worst-case gap closes under runout, thermal growth or load. | Measure assembly runout and validate hot, loaded clearance against a project-specific limit. |
Evidence anchors: harmonic and topology modeling, component-test comparability and structural loading. The mitigation steps are this guide’s engineering recommendations.
Method, evidence strength and limits
Scope: early architecture, core procurement and prototype planning for OEM AFPM motors. This is a targeted evidence review, not an exhaustive literature survey or a production release specification. Compare the same envelope, voltage/current limits, duty, cooling, mass definition and manufacturing volume. Historical sources support design principles; reconfirm current material availability and processing with the supplier.
FAQ: resolve the next design decision
Architecture and comparison
Is YASA always the highest torque-density choice?
No universal ranking is established. Compare continuous or peak torque at the same cooling conditions and with the same mass definition, including housing and supports when evaluating a complete motor.
Are AFIR and TORUS the same topology?
No. AFIR places the rotor between two stators. TORUS places an internal stator between two rotors. YASA also uses two rotors but removes the stator yoke and uses a segmented armature.
Can a YASA stator use laminations?
Yes. The cited in-core cooling research uses laminated stator segments. “YASA” identifies an architecture, not a mandatory powder material.
Should a low-cogging design use a coreless stator?
It is a candidate to evaluate, not an automatic recommendation. Removing stator teeth removes that slotting source of cogging, but winding support, effective gap, copper loss and loaded torque ripple still need analysis. Slotless and coreless are not interchangeable specifications.
Materials, cooling and mechanical limits
Does SMC necessarily beat steel above 1 kHz?
No crossover applicable to every grade is supported here. Match flux density, excitation waveform, temperature and specimen processing; include copper and magnet losses in the motor decision.
Can I use the measured loss values directly in a quotation?
Use them as named reference inputs only. Ask the supplier to confirm the selected grade, density, treatment and test method, and to identify any difference between ring data and the finished geometry.
What potting conductivity should I specify?
Derive it from the loss and thermal-resistance budget, bond-line thickness, contact area, coolant boundary and allowable temperature. Also qualify cure, voids, insulation and thermal expansion. This guide supplies no universal minimum.
What nominal air gap should I release?
Choose it jointly with electromagnetic performance and minimum running clearance. Account for tolerances, bearing motion, hot growth and loaded deflection for each active gap; the worked budget is illustrative only.
Manufacturing and supplier handoff
Is there a fixed maximum diameter for a pressed SMC core?
No diameter cutoff is established here. Estimate force from pressure times the actual projected powder area, then confirm press, tooling, ejection and density constraints. Segmenting changes both the pressing area and assembly work.
How should machining be approved?
Ask the supplier to qualify the proposed machining route and remeasure dimensions, insulation and magnetic performance on representative parts. Do not specify an unverified surface treatment as a universal remedy.
What should I require for rotor assembly?
Request a reviewed fixture and retention plan sized from calculated magnetic forces and the actual assembly sequence. Validate load capacity and separation control before assembly; motor power alone is not a fixture specification.
What belongs in the first RFQ?
Provide the duty map, candidate topology, drawing revision, geometry, material assumptions, loss/thermal limits, inspection criteria, prototype quantity and annual volume. Mark unknowns explicitly so suppliers can price the same scope.
Next steps for engineering and sourcing
Architecture team
Prepare two candidates with a common duty map, envelope, cooling boundary and mass definition. Record why each could fail.
Core / manufacturing team
Send a revision-controlled drawing, candidate grades, winding interfaces, gap budget and proposed acceptance tests for DFM review.
Sourcing team
Request comparable prototype and production quotes, tooling assumptions, sample evidence, lead times and a list of unconfirmed requirements.
Source register and review dates
Evidence checked on . Links open the specific paper or document in a new tab. Publication dates describe source versions; they do not imply a new test by AxialFluxCore.
- Wang et al. — AFPM Motor Topologies Magnetic Performance Comparison
2024-01-13 · Energies 17(2), 401 · DOI: 10.3390/en17020401
Primary analytical/3D FEA comparison of YASA, AFIR and offset AFIR. Conclusions depend on the modeled geometry, material and loading; they are not a universal topology ranking.
- Höganäs — Somaloy product and processing guidance
Live manufacturer guidance · accessed 2026-09-20
Supports 3D flux capability and grade-specific processing routes. Supplier guidance is not independent evidence that SMC outperforms every steel grade.
- Höganäs — Somaloy 1P Material Data, pp. 4–5 (PDF)
2025 edition · document 2272HOG-3 · accessed 2026-09-20
Typical 700HR 1P data at 800 MPa compaction; ring specimen OD55/ID45/H5 mm. Values are material-test inputs, not finished-stator guarantees.
- Höganäs — Somaloy 3P Material Data, comparison method, p. 2 (PDF)
Manufacturer technical note · accessed 2026-09-20
Explains why SMC ring data and electrical-steel Epstein sheet data are not directly interchangeable. Match specimen and processing conditions before comparison.
- In-Core Cooling for YASA Axial Flux Machines: Analysis and Experimental Validation
Online 2024-12-26 · IEEE TIA 61(2), March–April 2025 · DOI: 10.1109/TIA.2024.3522874
Public abstract reports cooling channels in laminated core segments, modeling and motorette validation. This is evidence of a cooling route, not a guarantee for another complete motor; full-text details were not used.
- Pranjić & Virtič — Rotor mechanical stress analysis of a double-sided AFPM machine (PDF)
2017-12 · Journal of Energy Technology 10(4), pp. 57–70
Primary structural study of a double-rotor prototype; supports coupling magnetic attraction and disc deflection. Its dimensions are not production tolerances for other machines.