{"id":"1814199c-34c7-4371-83ec-be6b3c55bbe3","arxiv_id":"1908.04594","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Standardized state-space connection operations allow DC-DC converter systems, including filters, loads, controllers, and cascaded stages, to be assembled from reusable building blocks while preserving access to impedance and control transfer functions.","lead":"This paper proposes a modular way to build small-signal state-space models of DC-DC converter systems by connecting smaller submodels, such as filters, converters, and controllers, through standardized operations. The approach lets engineers extract impedances and control transfer functions from one model at any stage, including closed-loop and cascaded converter setups.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Closed-loop operation in §3 is incomplete: with nonzero D13 (or D21/D22) and controller feedthrough DC≠0, C and D must be corrected; the paper only assumes D23≈0.","rationale":"The reader's verdict is CONDITIONAL with medium correctness risk. My stress-test pass confirms that the central claim—that standardized connection operations produce a correct full model at any stage—has a real soft spot in the closed-loop construction. The series-connection algebra in Section 4 appears internally correct, and the time-domain validations support the examples. The problem is that the general controller connection procedure in Section 3 is not exact as written; it silently assumes that the third column of DOL is zero (or that its product with K is negligible for all outputs of interest), whereas the text only flags D23≈0. In common buck-type converters D13 is nonzero, so the assumption is not vacuous. Because this affects transfer functions that the paper explicitly promises (input/output impedance, control-to-output), the claim should be conditional on either restricting the method to controllers with D_C=0 and converters with D13=0, or updating C and D with the correction terms. This is a scoping/correctness issue, not a rejection of the method; the examples appear valid and the approach is useful.","tokens_in":16212,"tokens_out":11438,"duration_ms":118481,"concrete_test":"Use a buck converter averaged model with D13=I_L (e.g., x=[iL; vC], iin = D·iL + I_L·d) and a proportional controller with D_C=1. Construct the closed loop with error e=r-vout, vout≈C2·x. Compare two models: (1) exact substitution e=r-C2·x in both state and output equations, giving ACL=AOL-BOL3·K, CCL=COL-DOL3·K, DCL3=DOL3; (2) the paper's rule ACL=AOL-BOL·K with C,D unchanged. Compute Yin(s)=iin/vin with r=0 and the control-to-output Gco(s)=vout/r from both. Any nonzero difference in the Bode plots confirms the missing DOL3·K term. If D_C=0 for all allowed controllers, the test will pass and the scope should be stated.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The main load-bearing assumption is in Section 3. Equations (5) and (6) construct the open-loop model by augmenting states and then set ACL = AOL - BOL·K while declaring BCL = BOL, CCL = COL, DCL = DOL. This is exact only when the third column of DOL is zero, because closing the loop substitutes e(t) = r(t) - K·[x;xC] into both the state equation and the output equation. From (5), the third column of DOL is [D13·D_C, D23·D_C]^T. The paper explicitly assumes D23≈0 (Section 3) but does not assume or justify D13=0 or D_C=0. For a buck-type converter, iin = d·iL, so the linearized model has D13 = I_L ≠ 0. Any controller with D_C≠0 then makes DOL(1,3)≠0 and the closed-loop output for iin becomes C1·x + D11·vin + D12·iout + D13·D_C·(r - Kξ) = (C1 - D13·D_C·K_sel)·ξ + ... + D13·D_C·r. The paper's unchanged C matrix misses the C1 - D13·D_C·K_sel correction, so extracted input admittance and other iin-related transfer functions are wrong. The same reasoning applies to D21 and D22 feedthrough in vout if the controlled variable is fed back through outputs rather than pure states. The paper's own examples avoid the defect because all controllers in Appendix C have D_C=0 and the boost model has D13=0, but the operation is stated for the general controller model (3), so the method is broader than its assumptions.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes a modular state-space modeling framework for DC-DC converter systems. It defines a generic two-port state-space model with an additional control input, a controller model, and two standardized connection operations: adding controllers and closing control loops (Section 3) and series-connecting two two-port subsystems (Section 4). In Section 6, three examples (a multiloop buck converter, a boost converter with input filter, and a cascaded boost-buck system) are built from these operations and validated against time-domain circuit simulations. The paper claims that control-oriented and impedance transfer functions can be extracted from the single resulting model at any modeling stage, open or closed loop, single converter or series connection of converters.","tokens_in":16529,"tokens_out":10593,"duration_ms":101931,"significance":"If correct, the framework would be a practically useful way to assemble and maintain converter system models without re-deriving state-space equations. The series-connection derivation leading to Equations (13)–(15) is algebraically sound, and the examples are validated against independent circuit simulations with no fitted parameters, which is a real strength. However, the control-loop operation in Section 3 is mathematically incomplete for the general controller and converter models stated, which undermines the claim that impedance transfer functions can be extracted after loop closure in all cases. Since the main contribution is the generality of the two connection operations, this gap is load-bearing and needs to be fixed before the paper can be accepted.","major_comments":[{"comment":"The open-loop model is incomplete. Substituting the controller output u(t)=CC xC(t)+DC e(t) into the two-port model (2) gives iin = C1 x + D13(CC xC + DC e) + D11 vin + D12 iout and vout = C2 x + D23(CC xC + DC e) + D21 vin + D22 iout. Therefore COL in (5) should be [C1, D13 CC; C2, D23 CC], not [C1, 0; C2, 0] as printed. As written, the open-loop model drops the contribution of the controller states to both outputs whenever D13 or D23 is nonzero, which occurs for buck-type converters where iin depends on the duty cycle (D13 = IL ≠ 0). This affects any input-current or input-admittance transfer function extracted from the open-loop model.","section":"Section 3, Eq. (5)"},{"comment":"The closed-loop construction is also incomplete. If the feedback variable is the output voltage, the error is e = r − vout with vout = C2 x + D21 vin + D22 iout + D23 ctl, not e = r − C2 x unless D21 = D22 = D23 = 0. The K-matrix choice in (8) only feeds back C2 x. Consequently, even with the paper's assumption D23 ≈ 0, the closed-loop state and output equations miss the terms −BOL,3 D21 vin, −BOL,3 D22 iout, and the corresponding DOL corrections. The sentence 'The B, C, D matrices of the open-loop and closed-loop case are identical' is thus correct only under additional unstated conditions (e.g., D13 = D21 = D22 = 0 and DC = 0). Since Table 2 promises extraction of Zout, Yin, and other transfer functions at any modeling stage, this is a load-bearing gap rather than a cosmetic one.","section":"Section 3, Eqs. (6) and (8)"},{"comment":"The numerical examples do not exercise the problematic terms in the general loop-closing operation. All controller models in Appendix C have DC = 0, and the boost model in (18) has D13 = 0; the buck converter is cited from Reference [27] rather than given, so the reader cannot check whether D13 is nonzero. The time-domain validations compare inductor current and output voltage (state variables or outputs with small feedthrough), but never the input current iin after loop closure. Thus the examples are consistent with a restricted case and do not demonstrate the claimed generality. I recommend adding a benchmark with D13 ≠ 0 and/or DC ≠ 0, or explicitly narrowing the scope of Section 3.","section":"Section 6 and Appendix C"}],"minor_comments":[{"comment":"Many DOIs in the reference list appear garbled (e.g., 'doi:1A.11AJ/...' in [1], [2], and others); these should be corrected to the standard DOI format.","section":"References"},{"comment":"In the line defining the abbreviations, the term 'D1 11L' appears to be a typo for DL11; the notation should be fixed.","section":"Section 4, Eq. (15)"},{"comment":"The superscript notation for 'I2 control' is inconsistent (sometimes rendered as 'I2' and sometimes as 'I²'); please standardize.","section":"Section 6.1"},{"comment":"The caption states that solid and dashed lines correspond to 'with input filter' and 'without input filter', but the figure legend is not explicit; please make the line styles unambiguous in the figure itself.","section":"Figure 8"},{"comment":"The abstract uses 'cascaded converters' while the main text predominantly uses 'series connections of converters'; consider unifying the terminology.","section":"General terminology"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope and the series-connection part of the contribution is solid. The main issue is that Section 3 overstates the generality of the loop-closing operation; the examples sidestep the problem, so the paper as written could mislead readers who apply the operation outside the restricted cases. I would send it back for revision rather than reject, because the fix appears straightforward: either derive the correct C, D, and B updates for general feedthrough, or clearly restrict the claimed scope and justify the restriction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper gives a practical building-block method for assembling small-signal state-space models of DC-DC converter systems from two-port g-parameter blocks. The explicit series-connection formulas (13)-(15) are algebraically correct and genuinely convenient, and the controller-attachment step (5)-(8) is a clean way to keep a single state-space model through loop closures. The examples—multiloop buck, boost with input filter, series boost+buck—match circuit simulations, which is real evidence the method works for the cases shown. I'd say the core idea is sound and worth having in the literature.\n\nThe soft spot is exactly what the stress-test note flags. In Section 3, closing the loop updates only A via ACL = AOL - BOL K and leaves B, C, D untouched. That is exact only when the third column of DOL is zero, i.e., both D13*DC and D23*DC vanish. The paper explicitly assumes D23≈0 but never mentions D13. For a buck converter, the linearized input current has D13 = IL ≠ 0, so any controller with DC≠0 makes the closed-loop input-admittance transfer function wrong. The paper's own controllers all have DC=0 (Appendix C), so the examples are safe, but the method is stated for general controllers (3). This is a scoping error, not a crack in the whole enterprise. The fix is either to state the assumption D13=0 (or DC=0) up front, or to include the C and D update terms for nonzero feedthrough.\n\nOther observations: no code or data accompany the preprint, which makes reproducing the figures slower but not impossible. The reference list looks appropriate and no self-citation inflation. The series-connection derivation relies on inverting 1 + D11^L D22^S; that's a genericity condition you'd expect.\n\nVerdict: deserves a serious referee—the series-connection half is solid and the closed-loop gap is fixable with a scope statement. As it stands I would not accept without revision. For a reader, it's a useful toolbox paper: if you build converter models often, the formulas save time and avoid algebra errors. I'd bring it to a reading group on converter modeling methods.\n\nRecommendation: peer review it, with a request to correct or explicitly scope the closed-loop operation.","headline":"A genuinely useful and mostly correct modular state-space composition method for DC-DC converters, but the closed-loop step misses D13 feedthrough corrections and needs a scope statement before it can be accepted.","tokens_in":17063,"tokens_out":2724,"would_cite":true,"duration_ms":26633,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A complete DC-DC converter system model can be assembled from reusable building blocks by two standardized state-space connection operations, with no need to re-derive frequency-domain transfer functions when filters, loads, or control…","keywords":["state-space averaging","small-signal modeling","DC-DC converters","two-port networks","g-parameters","modular modeling","control loops","cascaded converters"],"falsifier":"Take a converter model whose $D$ matrix has a nonzero $D_{13}$ or $D_{23}$ term, apply the closed-loop update $A_{\\mathrm{CL}} = A_{\\mathrm{OL}} - B_{\\mathrm{OL}} K$ with $B$, $C$, $D$ unchanged, and compare the resulting output impedance or control-to-output gain with the exact model obtained by substituting the feedback law directly; any measurable discrepancy refutes the generality of the operation.","tokens_in":15979,"feed_emoji":"⚡","tokens_out":13893,"duration_ms":117220,"temperature":0.7,"pith_summary":"The paper argues that the tedious, error-prone workflow of re-deriving a DC-DC converter's small-signal model whenever filters, loads, or control loops are added can be replaced by a modular assembly process. It defines one generic two-port state-space model whose inputs are input voltage, output current, and control signals and whose outputs are input current and output voltage, and it provides two standardized connection operations: one attaches a controller and closes a loop, and one connects two subsystems in series. Every operation preserves the same model structure, so control-to-output, impedance, admittance, and gain transfer functions can be read off from a single model at any stage. If this works as claimed, engineers could build and maintain models of multiloop or cascaded converter systems by composing small, separately verified building blocks.","feed_headline":"Two operations assemble DC-DC converter systems from building blocks","feed_subtitle":"One two-port state-space model covers filters, control loops, and cascaded stages with no re-derivation.","key_machinery":"The central object is the generic two-port state-space model with inputs $v_{\\mathrm{in}}(t)$, $i_{\\mathrm{out}}(t)$, and $\\mathrm{ctl}(t)$ and outputs $i_{\\mathrm{in}}(t)$ and $v_{\\mathrm{out}}(t)$: an inverse-hybrid (g-parameter) representation that treats input voltage and output current as independent variables. Two operations carry the argument: controller insertion, which augments the state vector and updates the $A$ matrix via $A_{\\mathrm{CL}} = A_{\\mathrm{OL}} - B_{\\mathrm{OL}} K$ with $K$ choosing the controlled variable; and series connection, whose matrix formulas (13)-(15) eliminate the shared internal voltage and current between source and load. Their defining property is that each operation returns another model of the same two-port structure, which is what makes the frequency-domain extraction formulas collected in Table 2 valid at every modeling stage.","core_discovery":"The central claim is that a two-port inverse-hybrid (g-parameter) state-space description, augmented with control inputs, is closed under the two operations needed to build practical converter systems: adding a controller and series-connecting a source with a load. Closing a loop is done by augmenting the state vector with the controller states, connecting the controller output to the converter control input, and updating the system matrix through $A_{\\mathrm{CL}} = A_{\\mathrm{OL}} - B_{\\mathrm{OL}} K$, where $K$ selects the variable to feed back; $B$, $C$, and $D$ are left unchanged under the stated assumption that direct feedthrough from the control input to the controlled variable is negligible. Series-connecting two subsystems is done by eliminating the shared internal voltage and current with the coupling formulas (13)-(15), which produce a combined model of the same two-port form. Because the operations preserve the structure, one state-space model represents the whole system at every stage, and the transfer functions listed in Table 2 can be extracted from it at any point.","pith_inferences":["Going beyond the paper, the series-connection formulas are pure matrix arithmetic, so the entire assembly procedure could be automated: a user would supply building-block matrices and a connection list, and a simple computer-algebra program would generate the full model and all Table 2 transfer functions.","The paper's closed-loop update is exact only when the feedthrough terms $D_{13}$ and $D_{23}$ vanish; a natural generalization would be to substitute the feedback law algebraically into the output equations as well, covering converters with nonzero direct control feedthrough.","Because the operations do not depend on the physical nature of the building blocks, the same two-port assembly rules could be tried on averaged models of other power-electronic subsystems, such as bidirectional converters or inverter input stages, as long as those models fit the same g-parameter state-space form."],"forward_implications":["After adding an input filter, a non-trivial load, or another converter stage, the existing model is updated by a connection operation instead of a full re-derivation of the state-space equations.","Control loops can be closed one at a time, innermost first, with control-to-output and impedance transfer functions available after each step.","The same connection rules apply to passive elements, controlled converters, and entire series-connected systems, so a model built once can be reused as a building block in a larger system.","Because the operations preserve the two-port structure and linearity, all small-signal frequency-domain analyses remain valid at any modeling stage.","Controller tuning can take source and load interactions into account directly, since the impedance seen at any port is extractable from the same model at any point in the assembly."],"supporting_citations":[{"why":"Establishes state-space averaging, the modeling technique whose averaged small-signal models are the starting point for the paper's building blocks.","marker":"[22]"},{"why":"Supplies the inverse-hybrid two-port (g-parameter) characterization that defines the paper's generic model structure.","marker":"[20]"},{"why":"Provides the unified state-space model of PCM-controlled converters used as the buck converter building block in the multiloop example.","marker":"[27]"},{"why":"Offers ready-to-use two-port state-space models for buck, boost, and buck-boost converters that the method treats as reusable building blocks.","marker":"[30]"},{"why":"Gives the comprehensive two-port state-space converter modeling reference recommended as a source of building blocks.","marker":"[34]"},{"why":"Introduces unterminated small-signal behavioral converter models, supporting the strategy of modeling loads separately and connecting them later.","marker":"[9]"},{"why":"Defines I2 average current mode control, the scheme used in the multi-loop example.","marker":"[41]"},{"why":"Defines the Type 1-3 controller structures used as controller building blocks in the examples.","marker":"[37]"}],"fun_headline_variants":["Two operations assemble DC-DC systems from state-space blocks","Modular two-port models build converter systems without re-derivation","Building-block state-space models simplify DC-DC modeling","Plug-in state-space blocks compose DC-DC converters via two operations","Two-port state-space blocks connect to form whole converter systems"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the control input has no direct feedthrough to the measured controlled variable or to the input current, because the closed-loop construction updates only the system matrix $A$ while leaving $B$, $C$, and $D$ fixed; if either direct path is nonzero, the closed-loop model is only approximate.","fun_headline_variants_meta":{"raw":{"variants":["Two operations assemble DC-DC systems from state-space blocks","Modular two-port models build converter systems without re-derivation","Building-block state-space models simplify DC-DC modeling","Plug-in state-space blocks compose DC-DC converters via two operations","Two-port state-space blocks connect to form whole converter systems"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00078,"raw_usage":{"total_tokens":3441,"prompt_tokens":934,"completion_tokens":2507,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":550,"completion_tokens_details":{"reasoning_tokens":2424}},"tokens_in":550,"tokens_out":2507,"duration_ms":19822,"temperature":1.0,"reasoning_tokens":2424,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:38:44.693068+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a converter model whose $D$ matrix has a nonzero $D_{13}$ or $D_{23}$ term, apply the closed-loop update $A_{\\mathrm{CL}} = A_{\\mathrm{OL}} - B_{\\mathrm{OL}} K$ with $B$, $C$, $D$ unchanged, and compare the resulting output impedance or control-to-output gain with the exact model obtained by substituting the feedback law directly; any measurable discrepancy refutes the generality of the operation.","supporting_citations":[{"cited_title":"A general unified approach to modelling switching- converter power stages","cited_arxiv_id":null,"evidence_quote":"Establishes state-space averaging, the modeling technique whose averaged small-signal models are the starting point for the paper's building blocks."},{"cited_title":"Two-port characterization of PWM voltage regulators at low frequencies.IEEE Transactions on Industrial Electronics, 35(3):444–450, 1988","cited_arxiv_id":null,"evidence_quote":"Supplies the inverse-hybrid two-port (g-parameter) characterization that defines the paper's generic model structure."},{"cited_title":"Smithson and Sheldon S","cited_arxiv_id":null,"evidence_quote":"Provides the unified state-space model of PCM-controlled converters used as the buck converter building block in the multiloop example."},{"cited_title":"Dynamic Profile of Switched-Mode Converter: Modeling, Analysis and Control","cited_arxiv_id":null,"evidence_quote":"Offers ready-to-use two-port state-space models for buck, boost, and buck-boost converters that the method treats as reusable building blocks."},{"cited_title":"Wiley-VCH, Weinheim, Germany, 2017","cited_arxiv_id":null,"evidence_quote":"Gives the comprehensive two-port state-space converter modeling reference recommended as a source of building blocks."},{"cited_title":"Lee, and Dong Dong","cited_arxiv_id":null,"evidence_quote":"Introduces unterminated small-signal behavioral converter models, supporting the strategy of modeling loads separately and connecting them later."},{"cited_title":"Lee, Paolo Mattavelli, and Pei-Hsin Liu","cited_arxiv_id":null,"evidence_quote":"Defines I2 average current mode control, the scheme used in the multi-loop example."},{"cited_title":"Optimum feedback amplifier design for control systems","cited_arxiv_id":null,"evidence_quote":"Defines the Type 1-3 controller structures used as controller building blocks in the examples."}],"review_version":1}