The portfolio
Projects in converter design, engineering automation and machine learning. Open a project to see how it works and what its results can tell you.
New to AI? Try the first lesson →Engineering work you can read at its original source.
My EDN guide to choosing a boost PFC operating mode, starting with what your power supply needs.
DecisionWhich conduction mode fits the design priorities?
What existsPublished by EDN under Rafael Collado's byline.
LimitApply the guide to your own specifications and component limits.
Read projectA practical map of the eight dependent decisions in an LLC resonant converter design.
DecisionCheck whether specifications, definitions, selected components, operating range, and stress calculations form one traceable design chain.
What existsA primary EEPower article under Rafael Collado's byline documents the eight-step method; this site adds a review map that preserves its dependencies.
LimitA worked design method is not project-specific design approval or hardware validation.
Read projectWhat I am building, what has been tested, and what remains to be done.
A research workflow that turns simulated flyback evidence into a traceable next simulation check.
DecisionDecide which additional simulation check can separate competing explanations, or whether the simulated evidence is too weak to continue.
What existsDefined a simulation-only evidence-to-next-check contract; no hardware diagnostic tool or validated model is claimed.
LimitNo hardware measurements, bench validation, safe-measurement guidance, or transfer from simulated cases to physical converters.
Read projectAn evaluation contract for deciding when a fast magnetic estimate may be used, warned, escalated, or refused.
DecisionChoose use, warn, request higher fidelity, or abstain according to the available evidence.
What existsDefined a public evaluation contract for baselines, grouped validation, residuals, unfamiliar inputs, and abstention.
LimitNo trained model, benchmark, public dataset, project-specific threshold, or hardware validation is presented.
Read projectA prototype specification for deciding whether approved documentation supports an answer, needs more context, or requires expert review.
DecisionChoose answer with citations, ask for missing context, or escalate to expert review.
What existsSpecified a three-route answerability workflow and its evaluation plan; no implemented system or accuracy result is claimed.
LimitNo public approved-source corpus, benchmark, freshness guarantee, production system, or replacement for expert review is presented.
Read project