A discrete power device is a standalone semiconductor that performs one dominant power function — switching current, converting AC to DC, blocking reverse current, or clamping abnormal voltage. Rectifier diodes, MOSFETs, IGBTs, thyristors, TVS and ESD diodes, Zener diodes, and silicon-carbide devices all belong to this class. They matter because the chosen device often decides a system’s efficiency, heat, voltage margin, EMI behavior, fault robustness, and cost: in a phone charger, solar inverter, EV on-board charger, motor drive, or UPS, the power semiconductor is frequently the component that determines how much input energy becomes heat instead of useful output. This article maps the device menu, explains why the choice shows up as measurable system results, and gives designers and buyers a four-question review that catches most selection errors before a BOM locks.
One function per package, many functions per system
The dividing line against integrated circuits is functional, not physical. An IC combines many transistors, diodes, and logic on one die and handles control, sensing, or regulation; a discrete part does one job at the point of power. A PWM controller IC decides when a converter should switch; the discrete MOSFET performs the switching; the discrete rectifier handles the AC-to-DC conversion; a TVS protects the rail. The three work as a system — controller, switch, and passives — and the reliability of the whole depends on how they are selected and integrated, not on any single component. The same pattern repeats across product categories: a DC-DC converter, a motor drive, and a UPS each consist of a control brain, discrete power stages, protection devices, and passives.
| Component class | Typical role | Example |
|---|---|---|
| Discrete power semiconductor | One power or protection function | MOSFET, rectifier diode, IGBT, TVS, SiC Schottky diode |
| Integrated circuit | Multiple functions on one chip | PWM controller, gate driver, MCU, battery-management IC |
| Passive component | Stores, filters, or senses energy | Capacitor, inductor, resistor, transformer, fuse |
The device menu: who does what
The right category depends on what the circuit must do. Rectifier diodes and bridge rectifiers convert AC to DC and block reverse current — the input stage of adapters, chargers, and industrial supplies. Schottky and ultrafast diodes cut forward drop or recovery time where output rectification and freewheeling paths create loss. MOSFETs are voltage-controlled switches for DC-DC converters, synchronous rectification, battery systems, and motor drives. IGBTs take over in medium-to-high-voltage, high-current stages such as industrial drives, welding, traction, and UPS. Thyristors (SCRs, TRIACs) provide controlled AC power for phase control and soft starters. Protection comes from TVS diodes for surge clamping, ESD diodes at I/O and data lines, and Zeners for reference and low-power clamping. SiC Schottky diodes and SiC MOSFETs extend the same menu to high-voltage, high-efficiency, and thermally demanding stages in solar inverters, EV charging, and energy storage.
Each category lives on the Good-Ark site as a family page rather than a single part — the rectifier families, MOSFET families, and IGBT families — which makes the catalog itself a useful first map of the technology space.
Why the choice shows up as heat, margin, and field failures
Device selection shows up in measurable system results. Conduction and switching losses decide how much input energy becomes output rather than heat: conduction loss tracks forward voltage or RDS(on) at the operating current, and switching loss tracks the energy per event (Eon, Eoff, recovery charge) times frequency. Thermal performance follows from the same loss, divided by package and board thermal resistance, to a junction temperature that must stay inside the derated curve for the expected life. Voltage and current margin determine survival: the device must tolerate switching overshoot, surge currents, and fault events within its SOA or surge rating. Switching behavior affects EMI through rise/fall times, ringing, and recovery; protection devices keep transients off gate drivers and interfaces. Sourcing attributes — package availability, traceability, qualification status, PCN policy, MOQ, lead time — determine whether the design can be manufactured and maintained over the product’s life, and MOSFET data itself is only comparable when measured per the JEDEC JESD24 test methods. None of these six in isolation makes the choice; selection is a constraint problem across all of them.
Two decisions that most designs meet
The MOSFET-versus-IGBT decision is the most common crossover. Both are controllable switches with voltage-driven gates, but their loss mechanisms differ. A power MOSFET’s conduction loss scales with RDS(on) and switching is fast, which suits low-to-medium voltage and higher-frequency stages. An IGBT’s conduction loss scales with VCE(sat) and carries a bipolar tail at turn-off, which suits medium-to-high-voltage, high-current stages at lower frequency. There is no clean voltage boundary; the crossover depends on the topology, bus voltage, current, frequency, cooling, and gate-drive plan, and the honest method is to compare losses at the actual operating point. Bourns’ IGBT-versus-MOSFET analysis walks through where the crossover actually falls, and All About Circuits’ SCR primer covers the thyristor side of the switching menu.
The second recurring choice is among diode types at the same function. Standard rectifiers block reverse current at line frequency, where recovery speed is irrelevant. Fast-recovery diodes exist for converter output rectification and freewheeling, where trr converts directly into loss. Schottky diodes trade higher leakage and lower voltage class for a forward drop that is often half that of a PN diode — right for low-voltage output stages, wrong for the mains side of the circuit. TVS diodes clamp transients with a defined waveform and standoff voltage, where the energy handling matters as much as clamping. Zeners use controlled reverse breakdown for reference and low-power clamping. For every diode role, the comparison must happen at the intended current and temperature: rectifiers are compared by VRRM, average current, forward drop, surge current, and recovery; protection devices by standoff, breakdown, clamping, capacitance, and the relevant test waveform. A companion explainer on semiconductor diode basics is available from Electronics Tutorials; the practical details of verifying these numbers against the ordering code, datasheet revision, and supply documents before a part reaches the BOM are covered in Good-Ark’s practical guide to discrete device sourcing.
What a real power stage contains
A working example ties the menu together. A solar-inverter power stage still needs: SiC or IGBT switches for the DC-AC conversion; a PV bypass diode path around shaded cell groups to prevent hot-spot heating — the mechanism behind bypass-diode selection in shading events is analyzed in SolarEdge’s technical note on bypass diode effects; rectifier and bridge diodes in auxiliary and bias supplies; TVS protection on gate drives and control inputs; and ESD diodes on the communication and sensing interfaces. Each discrete function is small, but system efficiency and field reliability are the sum of their choices. Good-Ark documents its photovoltaic application picture — from bypass diode modules protecting strings to switching devices in the inverter — on the PV inverter application page.
The same pattern repeats in an EV on-board charger, an industrial motor drive, or a UPS: a controller decides, discretes switch and rectify, protection devices keep abnormal events off the silicon, and passives shape the waveforms. The engineering task is to choose each discrete part from its real function rather than from a habit of “MOSFET for everything” or “SiC for everything”, then validate the final device in the actual circuit — a rule that applies as much to the wide-bandgap sourcing guide as to the oldest rectifier on the menu.
Why they matter, stated as a rule
Discrete power devices matter because the power path is discrete. The controller IC is replaceable and cheap relative to the consequence; the MOSFET, IGBT, or diode sits in the path that carries the energy, sets the efficiency, and absorbs the faults. Degrees of freedom are real: the same converter can be built with a fast silicon rectifier, a Schottky, or a SiC diode, and each changes the loss, thermal, EMI, and cost profile of the whole. That is why the device choice is an engineering decision with system-level consequences rather than a catalog lookup.
The distinction between a discrete and an IC is also a supply-chain distinction. A discrete power part is often qualified for a specific package, die, and site, and its change control is tracked through PCN and lot traceability; substituting a catalog-equivalent part without re-checking those documents is how an apparently identical BOM ends up with a different thermal, surge, or reliability behavior in production. Sourcing disciplines belong to the same engineering conversation as the circuit design, which is why the sourcing guide referenced above spends as much text on ordering codes and documents as on electrical parameters.
The four-question design review
There is a practical way to keep the menu straight in a design review: assign each silicon decision to one of four questions — what is the function (rectify, switch, protect, reference), what is the stress (voltage, current, frequency, temperature), what is the thermal path (package, board, cooling), and what evidence exists (datasheet revision, qualification, traceability). A review that can answer all four for each device has covered most of the way an error enters a BOM; a review that stops at the function and the part number has not. Temperature deserves its own line because it is where most discrete-power choices fail late. The junction temperature is the sum of the ambient at the mounting point and the loss times the thermal resistance, and a device chosen at 25 °C ambient can be the wrong device at 70 °C in the same enclosure. The datasheet curves for derating, leakage, and on-resistance exist precisely to move the comparison to the working temperature, and the device with the best headline number is often not the device with the best number at the operating point.
For a BOM that has passed the four questions, the next step is validation at the real operating point: confirm the datasheet revision against the ordering code, review the thermal curves at the enclosure temperature, and test a sample in the actual circuit before volume commitment, using Good-Ark’s sample and quote process as the checkpoint for datasheets, cross-references, and sample requests against the final part list.