{"id":"e3a869f7-6490-471d-8183-7fe9cd8172fb","arxiv_id":"1908.05388","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"FEM simulations show toroidal, UU, and EE transformer structures span a wide range of leakage inductance and parasitic capacitance, with interleaved EE windings drastically cutting leakage at the cost of higher capacitance.","lead":"The paper uses finite element simulations to compute leakage inductance and parasitic capacitance for six high-frequency transformer designs with different core shapes and winding arrangements. The results map the trade-off between these parasitic parameters to help designers choose a structure for fast-switching converters.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reported primary-secondary capacitance is likely inflated by primary-to-core capacitance in iron-powder designs, undermining the parasitic trade-off map.","rationale":"The reader's weakest assumption was that the FEM simulations faithfully model the real transformers and that the parasitic values are not experimentally validated. My concern is narrower and more specific: the capacitance extraction method itself appears to conflate winding-to-winding capacitance with capacitance to a grounded conductive core. This is a load-bearing issue for the central claim because the headline numerical range (30.603 to 6444.137 pF) and the proposed trade-off between leakage inductance and series parasitic capacitance depend on the Table VI values being inter-winding capacitances. The evidence is internal to the paper: the iron-powder versus ferrite comparison shows a factor of 8-40 reduction for types a, b, and e but almost no change for c and d, which is exactly the signature of a geometry-dependent core-to-ground contribution added in parallel. I credit the paper for clearly stating its simulation assumptions, for showing field plots, and for comparing multiple winding arrangements; the leakage-inductance extraction using short-circuited secondary energy is a standard and reasonable method. The issue is not an internal inconsistency in the simulation outputs but an ambiguous interpretation of the capacitance quantity. This can be resolved by a targeted recomputation or an impedance measurement, so the appropriate disposition remains conditional rather than outright rejection. I agree with the reader that validation is needed, but the more decisive check is isolating the core-ground contribution, which the reader's broader mesh-convergence/experimental-validation concern does not directly capture.","tokens_in":7828,"tokens_out":6052,"duration_ms":64014,"concrete_test":"Re-run the electrostatic FEM for Table VI types a, b, and e (and the interleaved EE cases in Table VII) with the iron-powder core set to floating, and compute the primary-secondary capacitance by integrating the normal D-field over the secondary winding surface only, rather than by total-volume energy with a grounded or fixed-potential core. If the extracted values drop to the order of the ferrite entries (1.3-7.3 pF), the published 30.6-76.3 pF values include primary-to-core capacitance and the trade-off map must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The capacitance values in Table VI are extracted from total electrostatic stored energy with the secondary held at 0 V and, for the conductive iron-powder core, the core included in the electrostatic model (Section IV-B, Eqs. 7-9, Fig. 9). If the core is grounded or otherwise at a fixed potential, the computed energy includes the primary-to-core capacitance in parallel with the primary-to-secondary capacitance. The reported quantity is then not the winding-to-winding capacitance claimed in the abstract, and the low-capacitance entries for iron-powder types a, b, and e (30.603, 76.258, and 60.577 pF) are not comparable with the ferrite entries (1.346, 1.869, and 7.312 pF), which the paper attributes only to core conductivity. The overlaid types c and d are nearly unchanged with core material because winding-to-winding coupling dominates, which is consistent with the core contribution being geometry-dependent rather than a pure winding-to-winding measurement. Since the central claim is a trade-off between leakage inductance and series parasitic capacitance, using these values without separating the core-ground contribution can misrepresent the design space, especially for structures where the primary is close to the core.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript uses 3D finite-element simulations to compute magnetizing inductance, leakage inductance, AC resistance, and primary-to-secondary parasitic capacitance for six high-frequency transformer configurations (two toroidal, one UU, one EE, plus two toroidal core-size variants) designed for 400 V/400 V, 8 kVA, 10 kHz operation. It also studies three interleaved winding arrangements for the EE core. The central claim is that the investigated structures span a wide range of leakage inductance (5.12 to 736.25 µH) and parasitic capacitance (1.346 to 6444.137 pF) so that designers can choose a structure that balances the two parasitics for a given application.","tokens_in":8058,"tokens_out":3749,"duration_ms":40605,"significance":"If the extracted parasitic values are correct, the paper provides a useful design-oriented comparison of core geometries and winding arrangements, and the interleaving study quantifies an important trade-off. The use of FEM energy-based extraction is standard, and the systematic comparison across six structures, including two core materials, is a strength. However, the capacitance extraction appears to include core-to-winding contributions when a conductive iron-powder core is present, and the absolute values are not validated by measurements, mesh-convergence studies, or analytic checks. The claimed quantitative design map therefore needs correction and additional support before the central conclusion can be accepted.","major_comments":[{"comment":"The reported 'capacitance between primary and secondary windings' is extracted from total electrostatic stored energy with the secondary held at 0 V. In the companion insulation test of Fig. 8 the cores are explicitly grounded, and the text does not state the core boundary condition for the capacitance extraction. If the conductive iron-powder core is also grounded (or otherwise at a fixed potential), the stored energy includes the primary-to-core capacitance in parallel with the primary-to-secondary capacitance. The low iron-powder entries for types a, b, and e (76.258, 30.603, and 60.577 pF) would then not be comparable with the ferrite entries (1.869, 1.346, and 7.312 pF) as winding-to-winding values, and the abstract's trade-off claim is based on a potentially inflated quantity. Please state the core potential used in the capacitance simulation; if the core is grounded, rerun with the core excluded or with partial capacitances separated, and revise the design-space conclusions accordingly.","section":"Section IV-B, Eqs. (7)-(9), Fig. 9, Table VI"},{"comment":"The leakage inductance of case a is 736.25 µH against a magnetizing inductance of 2061.5 µH, implying a coupling coefficient of only about 0.74 for a toroidal transformer. This is a surprising result that likely depends on the assumed homogeneous multi-layer winding model and on mesh resolution. No mesh-convergence study, solver settings, or analytic estimate is reported. Without these, the quantitative range of leakage values in Table V, and the design conclusions drawn from it, are not established.","section":"Section IV-A, Table V"},{"comment":"The simulations rely on several unvalidated modeling assumptions: the B-H curve of Iron Powder-Mix-08 (Fig. 1), homogeneous winding layers with fixed 0.15 mm insulation, 2 mm bobbin thickness, and relative permeability of 1 for all non-core materials. Since the paper's central claim is a quantitative trade-off map, at least one experimental measurement or a comparison with a published benchmark transformer is needed to confirm that the FEM model reproduces parasitic values with acceptable accuracy. As written, the absolute values are not verifiable.","section":"Section II and Section IV"},{"comment":"The interleaved-winding capacitance values (6704.1, 2875.0, and 3832.3 pF) are extracted with the same method as Table VI, so they inherit the same core-ground issue. In addition, the conclusion that leakage inductance is reduced while capacitance increases is qualitatively expected, but the quantitative values and the comparison between types a and b should be revisited once the capacitance extraction is corrected.","section":"Section IV-C, Table VII"}],"minor_comments":[{"comment":"There are multiple typos and grammatical errors that should be corrected, including 'capacitacne' (Table VI), 'transforemrs' and 'distribuiton' (Fig. 9 caption), and 'silicon vanish' (Section IV-B).","section":"Throughout"},{"comment":"Equation (2) is not typeset correctly: the denominator contains an extra '(c)' and mismatched parentheses, so the core-loss formula is not usable as printed. Please rewrite it with clear notation and define all constants consistently with Table II.","section":"Equation (2)"},{"comment":"The text says the maximum electric field of 67 kV/mm in Fig. 8(a) is 'much lower' than the dielectric strength of silicon varnish (120 kV/mm); while it is lower, the margin is not large, and the wording should be adjusted to reflect the actual ratio.","section":"Section IV-B"},{"comment":"The citation to 'IEEE Std. C57.12.01' is incomplete; the full standard number, year, and title should be provided, and reference [8] should be cited with page numbers or a DOI where available.","section":"References"},{"comment":"The terms 'case 1' and 'case 2' for the toroidal cores are used in the text and figures but are not explicitly defined in Table III; please add labels so that the reader can map the cases to the dimensions.","section":"Section III"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nQuick read: this is a useful comparative FEM dataset, not a methodological breakthrough. They simulate six transformer structures (toroidal, UU, EE, plus interleaved EE variants) at 10 kHz/8 kVA and report leakage inductance, magnetizing inductance, AC resistance, and parasitic capacitance. The range they find is large (leakage from ~5 µH to ~736 µH; capacitance from ~1 pF to ~6.4 nF), so the broad claim that core shape and winding arrangement dominate the parasitic trade-off is credible.\n\nWhat's genuinely new is the systematic side-by-side comparison, including the interleaved-winding variations, with a consistent 80-turn, 1600-mm² core setup. The energy-based FEM procedure is standard, and the paper is straightforward about the assumptions (B-H curve, 0.15 mm insulation, 2 mm bobbin). Credit for checking core conductivity's effect by swapping iron powder for ferrite.\n\nNow the soft spots, in order of how much they matter.\n\nFirst, the capacitance numbers are probably not pure primary-to-secondary values. The electrostatic simulation applies 1 V to the primary and 0 V to everything else, and Fig. 8 explicitly grounds the core for the insulation voltage test. If the core is grounded in the capacitance run as well, the stored energy includes primary-to-core capacitance in parallel with primary-to-secondary. For the iron-powder cases (a, b, e) the 30–76 pF readings could be largely core-to-winding capacitance, not winding-to-winding. That would make Table VI a mixed map, not the clean trade-off claimed. The authors need to separate or clearly define the reference conditions (core floating, core grounded, or core tied to secondary).\n\nSecond, no experimental validation or mesh-convergence study. The 736 µH leakage for type a (toroidal, 3-layer) is more than a third of its own magnetizing inductance; that ratio is suspicious and suggests the winding model might be capturing something other than leakage. At minimum, the paper needs a sensitivity analysis and a cross-check against measured prototypes.\n\nThird, missing FEM details: solver, element order, boundary conditions, mesh size, frequency-dependent skin-effect handling. Without these, the numbers are not reproducible.\n\nThis isn't a takedown. The comparison is plausible and the dataset could be a useful reference for designers choosing between toroidal, UU, and EE with different winding strategies. But right now it's a preprint-level design study, not a validated result.\n\nRecommendation: send to peer review if the authors are willing to add the missing validation and fix the capacitance extraction. A serious referee can push those changes. I wouldn't desk reject it, but I also wouldn't let it through as-is.","headline":"Useful comparative FEM dataset for transformer parasitics, but the capacitance extraction likely mixes core-to-ground effects into the primary-secondary values and needs validation.","tokens_in":8489,"tokens_out":3823,"would_cite":false,"duration_ms":35332,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A family of six transformer structures spans leakage inductance from 5.12 to 736.25 µH and primary-to-secondary parasitic capacitance from 30.6 to 6444 pF, so winding and core layout can be chosen to balance the two parasitics.","keywords":["high-frequency transformer","leakage inductance","parasitic capacitance","finite element method","winding arrangement","interleaved windings","toroidal core","EE core"],"falsifier":"Build one of the simulated transformers, say the EE core with P-S-P-S-P-S-P-S interleaving predicted at 5.12 µH leakage and 6704 pF capacitance, measure its open- and short-circuit impedance over 10 kHz to 1 MHz with an impedance analyzer, and check whether the extracted values match the simulation; a large mismatch, or a mesh-refinement study that shifts the FEM results substantially, would overturn the claim.","tokens_in":7693,"feed_emoji":"⚡","tokens_out":7128,"duration_ms":66582,"temperature":0.7,"pith_summary":"The paper sets out to show that the parasitic elements that dominate high-frequency transformer behavior are design variables, not fixed drawbacks of the magnetic core. Through finite element simulation of six 400 V/400 V, 8 kVA, 10 kHz transformer structures—two toroidal geometries with overlaid and non-overlaid windings, one UU core, and one EE core, plus three interleaved EE winding layouts—it reports leakage inductance spanning 5.12 to 736.25 µH and primary-to-secondary capacitance spanning 30.6 to 6444 pF (and up to 6704 pF with interleaving). The point of the spread is practical: it gives designers a catalog for choosing a structure whose parasitic profile matches the application, such as using leakage inductance as resonant tank inductance or keeping capacitance low for fast SiC and GaN switches. The paper concludes that no structure dominates and that balancing leakage against capacitance is a necessary trade-off.","feed_headline":"Transformer shape swings leakage from 5 to 736 µH","feed_subtitle":"Simulation of six 10 kHz isolation transformers shows how winding layout trades leakage inductance against parasitic capacitance.","key_machinery":"The carrying mechanism is the finite-element energy method for parasitic extraction. Leakage inductance is obtained from the magnetic energy stored in the whole space with the secondary short-circuited, and parasitic capacitance from the electric energy stored with 1 V applied across the windings, so the computed quantities are direct integrals of field solutions rather than closed-form estimates. The simulations use 3D models of toroidal, UU, and EE cores with an Iron Powder-Mix-08 B-H curve, AWG 8 windings, 0.15 mm inter-conductor insulation, and a 2 mm bobbin, with magnetizing inductance held nearly equal across designs so that the parasitic comparison is clean.","core_discovery":"For a 1:1, 400 V, 8 kVA transformer at 10 kHz with 80 turns, 1600 mm² core area, and iron-powder cores, the FEM calculations show that leakage inductance is set mainly by window-area volume and winding layout: the two toroidal non-overlaid cases store the most leakage energy (736.25 and 324.89 µH), while the overlaid toroids and interleaved EE cores store the least (15.21, 9.23, and 5.12 µH). Parasitic capacitance behaves oppositely: overlaid toroids and interleaved EE windings develop the largest primary-secondary capacitances (6444 pF and up to 6704 pF), while non-overlaid toroids and the UU core stay below 76 pF with the iron-powder core. The paper's central claim is that this wide, quantifiable spread is a usable design handle: each structure occupies a different point on a leakage-versus-capacitance map, and interleaving is a deliberate lever that trades inductance down while capacitance goes up.","pith_inferences":["One consequence the paper leaves implicit is that the six structures trace a leakage-versus-capacitance frontier: the lowest-capacitance designs (non-overlaid toroids, 30–76 pF) sit at the highest leakage (325–736 µH), and the lowest-leakage designs (interleaved EE, below 20 µH) sit above 2.9 nF. A designer could plot these points and pick the nearest structure to a required (L,C) target.","The strong dependence of capacitance on core conductivity—76.3 pF versus 1.87 pF for one toroid when iron powder is replaced by ferrite-NiZn—suggests core material choice is as powerful a lever as winding geometry. Combining a non-conductive core with interleaving to reach low leakage and low capacitance simultaneously is a natural next simulation.","Since all results are for one operating point (10 kHz, 1:1, 8 kVA), a testable extension is to sweep frequency and power and see whether the ordering of structures by parasitic severity holds, especially as skin effect changes leakage at higher frequencies."],"forward_implications":["A designer can intentionally choose a transformer structure whose leakage inductance serves as the resonant tank inductance, removing a separate inductor; the paper cites a prior result where this shrank system volume by 15 percent.","For fast-switching SiC and GaN converters, low-leakage interleaved EE windings come with high parasitic capacitance, so the injected high-frequency current and resulting EMI must be checked before adoption.","Because magnetizing inductance is held nearly constant across structures, the wide spread in parasitic values is attributable to geometry and winding layout rather than to different core flux levels.","Near-identical leakage values can accompany very different capacitances: interleaved EE types a and b give 5.12 versus 19.90 µH and 6704 versus 2875 pF, so selecting a structure should use both parameters together.","Magnetic saturation also enters the selection: the toroidal cores show no saturation at full load while the UU and EE cores have some saturated points, adding a constraint beyond parasitics."],"supporting_citations":[{"why":"Supplies the finite-element energy method used to extract primary-secondary parasitic capacitance from electric field energy.","marker":"[9]"},{"why":"Supplies the leakage-inductance calculation approach from magnetic stored energy with the secondary short-circuited.","marker":"[3]"},{"why":"Provides the design equations that set 80 turns and 1600 mm² cross-section for all structures.","marker":"[6]"},{"why":"Sets the 10 kV insulation test standard used to evaluate electric field stress and insulation adequacy.","marker":"[8]"},{"why":"Provides the core-shape comparison and winding trade-offs that motivate the selection of toroidal, UU, and EE structures.","marker":"[7]"},{"why":"Shows the motivating application where transformer leakage inductance is used as the resonant tank inductance, reducing system volume.","marker":"[4]"}],"fun_headline_variants":["Transformer leakage spans 5 to 736 µH by structure","Leakage inductance vs capacitance: a 100x trade-off","Interleaving cuts leakage to 5 µH, boosts capacitance to 6.7 nF","Six transformer structures mapped for leakage-capacitance trade-offs","FEM study: winding layout sets transformer leakage from 5 to 736 µH"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The computed parasitic values are taken from FEM simulations that assume the Iron Powder-Mix-08 B-H curve, homogeneous winding layers with 0.15 mm insulation, a 2 mm bobbin, and unit relative permeability for non-core materials, without experimental validation or a mesh-convergence check.","fun_headline_variants_meta":{"raw":{"variants":["Transformer leakage spans 5 to 736 µH by structure","Leakage inductance vs capacitance: a 100x trade-off","Interleaving cuts leakage to 5 µH, boosts capacitance to 6.7 nF","Six transformer structures mapped for leakage-capacitance trade-offs","FEM study: winding layout sets transformer leakage from 5 to 736 µH"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000993,"raw_usage":{"total_tokens":4206,"prompt_tokens":940,"completion_tokens":3266,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":556,"completion_tokens_details":{"reasoning_tokens":3167}},"tokens_in":556,"tokens_out":3266,"duration_ms":23170,"temperature":1.0,"reasoning_tokens":3167,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:14:34.612560+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build one of the simulated transformers, say the EE core with P-S-P-S-P-S-P-S interleaving predicted at 5.12 µH leakage and 6704 pF capacitance, measure its open- and short-circuit impedance over 10 kHz to 1 MHz with an impedance analyzer, and check whether the extracted values match the simulation; a large mismatch, or a mesh-refinement study that shifts the FEM results substantially, would overturn the claim.","supporting_citations":[{"cited_title":"A finite -element analysis approach to determine the parasitic capacitances of high -frequency multiwinding transformers for photovoltaic inverters,","cited_arxiv_id":null,"evidence_quote":"Supplies the finite-element energy method used to extract primary-secondary parasitic capacitance from electric field energy."},{"cited_title":"Calculation of Leakage Inductance for High-Frequency Transformers,","cited_arxiv_id":null,"evidence_quote":"Supplies the leakage-inductance calculation approach from magnetic stored energy with the secondary short-circuited."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the design equations that set 80 turns and 1600 mm² cross-section for all structures."},{"cited_title":"High-Frequency Transformer Design for Modular Power Conversion From Medium-Voltage AC to 400 VDC,","cited_arxiv_id":null,"evidence_quote":"Sets the 10 kV insulation test standard used to evaluate electric field stress and insulation adequacy."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the core-shape comparison and winding trade-offs that motivate the selection of toroidal, UU, and EE structures."},{"cited_title":"Design of Leakage Inductance in Resonant DC -DC Converter for Electric Vehicle Charger,","cited_arxiv_id":null,"evidence_quote":"Shows the motivating application where transformer leakage inductance is used as the resonant tank inductance, reducing system volume."}],"review_version":1}