🔌 MOSFET vs BJT Transistors: Which to Choose and Why

Two Different Physical Mechanisms, One Job

Both the bipolar junction transistor (BJT) and the metal-oxide-semiconductor field-effect transistor (MOSFET) do the same fundamental job — using a small control signal to govern a much larger current flow between two other terminals — but they get there through physically different mechanisms, and that difference cascades into nearly every practical design decision an engineer makes when choosing between them.

A BJT is a current-controlled device. A small current injected into the base terminal, on the order of microamps to milliamps, controls a much larger collector-to-emitter current, with the ratio between them defined by the current gain, β (beta or h_FE), typically 20–300 depending on the part. Because the base-emitter junction is a forward-biased PN junction, it behaves like a diode: driving it requires continuous current flow, not just a static voltage, and that base current is a real, non-zero load on whatever circuit is driving the transistor.

A MOSFET is a voltage-controlled device. The gate is electrically isolated from the channel by a thin layer of silicon dioxide (the "oxide" in MOSFET), so in steady state essentially no DC current flows into the gate at all — only the transient current needed to charge and discharge the gate's parasitic capacitance during a switching transition. This single structural difference — a diode-like current-driven junction versus a capacitor-like voltage-driven insulated gate — is the root cause of almost every other contrast between the two device families.

Drive Requirements: Base Current vs Gate Charge

Driving a BJT into saturation requires supplying continuous base current for as long as the device needs to stay on, sized as I_B ≥ I_C / β_forced, where designers typically use a forced beta well below the datasheet's typical β (often 10:1 rather than the full rated gain) to guarantee hard saturation across temperature and part-to-part variation. That base drive current is a permanent power draw on the driving stage for the entire on-time, and a BJT held in deep saturation stores minority carriers in the base region that must be actively removed before it can turn off — this is storage time, and it's a primary reason BJTs are slower to switch off than MOSFETs.

Driving a MOSFET means charging its input capacitance (specified on datasheets as total gate charge, Q_g, typically 10–150 nC for power MOSFETs) up to the gate threshold and beyond, via a transient current pulse rather than a sustained DC current. Once the gate reaches the target voltage, holding it there costs essentially zero steady-state current — the tradeoff is that the transient turn-on/turn-off current can be substantial (often 1–2 A peaks for fast power switching), which is exactly why dedicated gate driver ICs exist: a microcontroller GPIO pin can rarely source or sink enough peak current to switch a power MOSFET quickly on its own. The gate-to-drain (Miller) capacitance in particular creates a voltage plateau during switching transitions that gate drive circuits must be sized to push through quickly, or switching losses balloon.

Switching Speed and Losses

MOSFETs switch faster than BJTs in essentially every comparable application, often by an order of magnitude or more. A power MOSFET can commonly switch in tens of nanoseconds, while a comparable BJT, burdened by base charge storage and the minority-carrier recombination needed to exit saturation, is typically limited to the hundreds-of-nanoseconds-to-microsecond range. This is why virtually all modern switch-mode power supplies, motor drive inverters, and DC-DC converters operating above roughly 50–100 kHz use MOSFETs (or, at higher voltage/power, IGBTs) rather than BJTs — switching losses scale with both switching frequency and transition time, and a slow-switching BJT at a high switching frequency simply dissipates too much power in the transition region between fully off and fully on.

The tradeoff shows up in conduction losses instead. A MOSFET's on-state loss is governed by R_DS(on), its drain-source on-resistance, which produces I²R conduction loss and, critically, has a positive temperature coefficient — R_DS(on) rises as the device heats up. A BJT's on-state loss is governed by V_CE(sat), typically 0.1–0.3 V for a saturated device, which is roughly current-independent and can actually give a BJT lower conduction loss than a MOSFET at very high current densities where R_DS(on) × I would otherwise dominate. This is one of the few regimes (very high current, lower switching frequency) where a BJT — or more commonly today, an IGBT, which combines a MOSFET gate with bipolar-like conduction — still has a real efficiency edge.

Thermal Behavior: A Safety-Critical Difference

This is one of the most consequential practical differences and a frequent source of field failures when misunderstood. A BJT has a negative temperature coefficient of V_BE: as junction temperature rises, less base-emitter voltage is needed to sustain the same collector current, so for a fixed base drive, collector current tends to increase with temperature. In a design with multiple BJTs sharing current — paralleled devices, or multi-emitter power transistors — this creates a positive feedback loop: the hottest device conducts more current, which heats it further, which makes it conduct still more, a runaway condition that can destroy the device (or the whole parallel bank) without any external fault. Paralleling BJTs safely requires deliberate current-sharing measures — emitter ballast resistors being the classic fix.

A MOSFET's R_DS(on) has a positive temperature coefficient — resistance rises with temperature — which is inherently self-limiting: a MOSFET that starts conducting more than its paralleled neighbors heats up, its resistance rises, and current naturally redistributes toward the cooler devices. This makes MOSFETs dramatically easier to parallel for higher current handling without the ballasting tricks BJTs require, and it's one of the underappreciated reasons MOSFETs dominate high-current power stages even in applications where switching speed isn't the primary concern.

Linear Amplification vs Switching Applications

Both device families can operate in either the linear (active) region or as a switch, but their practical strengths diverge. BJTs have historically been preferred for small-signal and audio linear amplification because of high, relatively predictable transconductance at low currents, well-behaved and thoroughly characterized SPICE models, and low input-referred noise in many topologies — a large fraction of classic and modern discrete audio amplifier and RF small-signal designs are still BJT-based for exactly this reason. MOSFETs, particularly JFETs and small-signal MOSFETs, are used in linear applications requiring very high input impedance (op-amp input stages, electrometer-grade instrumentation) where a BJT's base current draw would load down a high-impedance source unacceptably.

For switching applications — anything turning fully on and fully off rather than sitting in a linear operating point — MOSFETs are the default choice in virtually every new design below roughly 600–900 V, covering DC-DC converters, motor drivers, class-D audio amplifiers, LED drivers, and digital logic-level load switching. Above that voltage range, or at very high current with lower switching frequency, IGBTs and, in legacy or cost-constrained designs, power BJTs still appear, particularly in industrial motor drives and grid-tied power conversion.

Enhancement Mode, Depletion Mode, and the Body Diode

Nearly all MOSFETs used in modern digital and power design are enhancement-mode devices — normally off, requiring gate voltage above a threshold (V_GS(th), commonly 1–4 V for logic-level parts) to create a conductive channel at all. Depletion-mode MOSFETs, which are normally on and require gate voltage to turn them off, are comparatively rare and appear mostly in niche applications like certain current-source or fail-safe circuits where "on by default, off on command" is the desired power-up behavior. BJTs have no equivalent normally-on/normally-off distinction in the same sense — they are inherently normally-off, requiring base drive to conduct at all, though the concept of "fail-safe default state" still matters in system design and is usually handled by external circuitry rather than device physics.

A structural feature specific to power MOSFETs with no BJT equivalent is the intrinsic body diode — a parasitic PN junction, inherent to the device's physical construction, connected antiparallel across drain and source. This body diode conducts whenever the source-to-drain voltage goes sufficiently negative, which is both a free convenience (it provides a natural freewheeling path in motor drive and synchronous rectification circuits, sometimes eliminating the need for a discrete flyback diode) and a real design hazard: the body diode is typically slow and lossy compared to a dedicated fast-recovery diode, and its reverse-recovery charge can cause significant switching loss and even shoot-through current spikes in half-bridge topologies if not accounted for.

Selection Criteria: A Practical Decision Framework

• Choose a MOSFET when: switching frequency exceeds tens of kHz, the driving source has limited current sourcing capability (microcontroller GPIO, logic-level control), multiple devices must be paralleled for higher current, or the application is a digital/power switching load (DC-DC converters, motor drives, LED drivers, load switches).
• Choose a BJT when: the design is a classic linear/analog amplifier stage where well-characterized transconductance and low-frequency behavior matter more than switching speed, cost is extremely constrained in a low-frequency application, or the design is inheriting or matching an existing BJT-based topology where redesign cost outweighs the efficiency gain of switching to MOSFETs.
• Choose an IGBT (a hybrid worth mentioning): for high-voltage (600 V+), high-current, moderate-switching-frequency applications like industrial motor drives, welding equipment, and grid-tied inverters, where a MOSFET's gate drive simplicity is combined with a bipolar output stage's low conduction loss at high voltage — a compromise that dominates that specific power/voltage/frequency niche.

In practice, the vast majority of new power electronics and digital-interfacing designs default to MOSFETs unless a specific reason — usually a linear-amplification requirement or an existing legacy topology — points back to a BJT. Understanding why that default exists (drive simplicity, switching speed, and self-limiting thermal behavior under parallel operation) matters more for engineering judgment than simply memorizing the rule of thumb, because every one of those advantages has a real exception where the opposite choice is correct.