SMPS output-network and measured-loop tuner review

The previous tool is replaced because its Bode drawing was a shaped illustration, not the response of a supplied circuit or a measured loop. Damping extracted from an arbitrary load transient does not establish a general SMPS phase margin. No earlier automatic component tuning or stability verdict is retained.

Supported models

Panel Basis Output scope
Capacitor resonance Per-part series RLC, Z = ESR + jωESL + 1/(jωC); identical parts in parallel; banks combine through complex admittance SRF, bank impedance versus frequency and sampled local peaks. Positive ESR/ESL and integer quantities required. No board/plane/common mounting inductance.
Output-capacitor sizing Ideal buck CCM ΔIL = Vout(1−Vout/Vin)/(Lfs); triangular ripple budget ΔV ≈ ΔIL/(8fsC)+ΔIL·ESR; IC,rms = ΔIL/√12 Effective capacitance requirement after ESR budget, not a part recommendation. Ripple sizing is suppressed outside CCM or above any bank SRF. Even below SRF, the lumped-capacitor estimate excludes resonance dynamics.
Load increase budget Full current deficit for user-supplied duration: ΔV ≈ ΔI·t/C + ΔI·ESR + ESL·ΔI/tedge A first-order budget, not exact response. No inferred loop recovery, negative load step, saturation or ringing. Mixed-bank ESR is the low-frequency effective value Σ(nC²ESR)/(ΣnC)²; ESL is the high-frequency parallel limit. These separate asymptotic contributions are not a full multi-branch transient solution.
Loaded LC filter Source → (Rseries + jωL) → (Rload ∥ capacitor bank); H = Zshunt/(Rseries + jωL + Zshunt) Absolute voltage gain and phase, sampled peak, gain at switching frequency, DC divider gain. Resistive load, constant lumped parameters, no magnetic/geometry model or converter-loop stability claim.
Measured-loop adjustment Supplied unwrapped open-loop data multiplied by K(1+jf/fz)/(1+jf/fp) Hypothetical lead/lag gain and phase; interpolated 0 dB crossings and PM = 180° + phase. No extrapolation, automatic pass/fail, general stability conclusion or compensator component synthesis.

All sweeps use 501 logarithmic points. Sampled extrema can miss a narrow resonance; the UI states that limitation and exposes sweep limits. Loop data must use strictly increasing frequencies and the negative-feedback return-ratio phase convention. Multiple crossings and phase branch selection require a complete loop/Nyquist review; phase margin alone is not a stability certificate.

Sources

Complex impedance and lead/lag calculations are explicit algebra, not a claimed validated regulator-control synthesis method. The UI keeps effective C, temperature/DC-bias effects, actual ESR/ESL, capacitor ratings and the controller’s approved output-network region as required design inputs/review items.

Executed checks

node --test tests/smps-design-tools.test.cjs: 8 tests passed.

node --check assets/js/smps-design-tools.js passed. DOM tests are lightweight fixtures, not a browser visual test or hardware correlation.

Integration

Keep the catalog ID tool-smps-loop-tuner, filename smps-loop-tuner.html, label “SMPS Output Network & Loop Tuner”. The fragment loads /assets/js/smps-design-tools.js?v=1 and calls SMPSDesignTools.mount(...). It needs no retired shared tools engine or lifecycle initializer.