Discuss the fundamental theories of new energy magnetic component design in detail.

The fundamental theories, key materials, structural processes, and optimization methods for the design of new energy magnetic components are introduced systematically, but a complete knowledge framework 'from fundamentals to methods' has not yet been formed, and there is a lack of systematic design processfor beginners. Focus on the following three aspects

• Systematic process or step-by-step guide for new energy magnetic component design
  
• Differentiated design points for magnetic components in typical topologies such as LLC, DAB, OBC
  
• Engineering tools or software platforms for magnetic component modeling, loss calculation, EMI suppression, etc
  
  

I. Fundamentals: Grasp the physical essence in 4 points

1. Two laws: Ampère's circuital law (H·l = NI) + Electromagnetic induction (U = N·dΦ/dt)—All formulas are variations of these two.
   
2. Three limitations: No magnetic core saturation (Bmax < Bsat), no winding overheating (Pcu + Pfe ≤ allowable loss), no geometric exceedance (AP method yields Aw·Ae ≥ (L·I·Irms)/(KB·J)).
   
3. Four types of losses: Copper loss (DC + skin effect + proximity effect), iron loss (Steinmetz three-term: hysteresis + eddy current + residual loss), air gap fringe loss, stray magnetic field EMI.
   
4. Five materials: Power ferrite (high frequency, low loss), Fe-Si-Al (high Bs), Fe-Ni-Mo (low loss, high μ), amorphous/nanocrystalline (20–100 kHz), SiC/GaN-compatible high-frequency low-μ powder cores.
   

II. Methods: 10-step systematic design process
Step Key Actions Quick Estimation Formulas / Tools New Energy Special Notes
0 Requirement input Topology, power, voltage ratio, fs, η, ΔT, volume, cost Automotive standards: -40 ℃ to 105 ℃, 20-year lifespan, ISO 26262
1 Select topology LLC, DAB, PSFB, CLLC, bidirectional Buck/Boost LLC emphasizes resonant inductor Lr, DAB emphasizes leakage inductance Llk, different magnetic integration approaches
2 Calculate electrical quantities L, Ipeak, Irms, ΔB, ΔI L = (Vin·D)/(fs·ΔI) Wide-bandgap devices → fs 100 kHz–500 kHz, ΔB set to 0.08–0.15 T to prevent ferrite overheating
3 Select magnetic core material + shape PC95, PC96, N87, KoolMμ, gas-atomized Fe-Si and composite powder cores, high flux, etc., handbook Pfe/f curves Automotive 'low height' → EQ/ER/planar cores; bidirectional flow → no-gap wound cores to prevent saturation
4 Rough geometric calculation AP = AwAe ≥ (L·I·Irms)/(KB·J·Bmax) Different J for air/liquid cooling: natural cooling J = 3–4 A/mm², liquid cooling up to 6–8 A/mm²
5 Determine air gap/effective μ δ = μ0·N²·Ae/L (corrected 1+δ/√Ae) Distributed air gap (powder core) better than concentrated air gap (ground center leg) → reduce fringe flux radiation
6 Number of turns N N = L·ΔImax / (ΔBmax·Ae) Round to integer → verify Bmax Under premise of meeting L value, N↓ → parasitic parameter C↓ → better EMI
7 Wire gauge + winding method Skin depth δ = 66/√f (mm) Litz wire strand diameter ≤ 2δ, layers ≤ 2 Interleaved parallel → balanced multi-strand current in same slot; foil winding + insulation film to reduce height
8 Loss breakdown Pfe = K·f^α·B^β·Ve; Pcu = Irms²·Rac PSIM/Maxwell/PExprt After high frequency, Rac/Rdc > 2, FEA necessary; automotive standard requires 'iron loss ≤ 100 mW/cm³'
9 Thermal resistance + temperature rise ΔT = (Pfe + Pcu)·Rth; Rth ≈ 1/(22·√Ve) Icepak/Fluent Planar core bottom soldered with copper sheet → thermal resistance reduced by 30%; fill gaps with thermal conductive adhesive
10 Closed-loop verification Prototype → impedance analyzer + network analyzer + thermal imager Measure Saber/PSIM model parameters → update Cpar, Lleak, Rac; EMI scan 150 kHz–30 MHz

III. LLC vs. DAB magnetic component differences quick reference (6.6 kW OBC example)

Parameter LLC resonant transformer DAB phase-shift transformer
Series inductance Independent Lr or integrated leakage inductance Llk (few μH) Must use large leakage inductance Llk (20–40 μH)
Magnetic integration Easy to convert Lr → leakage inductance, saving one core Large leakage inductance → requires 'conjugate + leakage integration,' complex structure
Air gap Small air gap (0.1–0.3 mm) Large/distributed air gap to prevent saturation
Frequency 80–150 kHz fixed 20–100 kHz frequency modulation
EMI Resonant sinusoidal current, good low-frequency EMI Square wave → high di/dt, high common-mode noise
Efficiency Peak > 97% Light-load ZVS easily lost, requires Burst mode

IV. Actionable 'three-piece suite' toolchain

1. Rapid iteration: PExprt + PSIM
   – Input Vin, Vout, P, fs, automatically traverse core library, output initial version of 'N, δ, wire gauge, loss, temperature rise' in 5 min; one-click generate PSIM magnetic circuit model, directly run closed-loop efficiency plot.
   
2. Fine verification: Simulation Ansys Maxwell + Icepak
   – 2D → 3D frequency domain FEA, accurately calculate 'skin effect + proximity effect + fringe flux'; loss map automatically imported into Icepak, perform liquid cooling plate + thermal adhesive joint simulation, ΔT < 40 ℃ sign-off.
   
3. EMI closed-loop: PSIM EMI Toolkit
   – Bring Lpar, Cpar extracted from Maxwell back to PSIM, run CISPR 25 150 kHz–30 MHz conducted emission, automatically provide common-mode filter (Lcm, Ccm) parameters, reduce 'adding Y capacitors after board return.'
   

V. Design Checklist
□ Bmax @105 ℃ < 0.28 T (PC95)
□ Pfe + Pcu ≤ 1 % Pout @ full load
□ Hotspot temperature rise ≤ 40 K (ambient temperature 85 ℃)
□ Leakage inductance tolerance ±5 % (LLC) / ±10 % (DAB)
□ Winding resonant frequency > 10×fs
□ CM EMI < 60 dBμV @ 530 kHz
□ Pass IST 1000 thermal cycles (-40 ↔ 125 ℃)
□ Core fixation force ≥ 10 g acceleration (automotive vibration standard)

Summary: New energy magnetic component design = multi-objective optimization of 'electromagnetic laws × material properties × thermal-EMI-mechanical constraints'; turning the 10-step process into an Excel template + three major software interfaces transforms 'experience art' into 'engineering science,' enabling accurate calculation of Bmax, loss, temperature rise, EMI, and lifespan in one go, with prototypes passing on the first try.