BJT vs MOSFET Comparison

In short: a BJT is a current-controlled device that gives high transconductance per microamp and excellent precision analog performance, while a MOSFET is a voltage-controlled, near-infinite-DC-impedance device that dominates digital logic, switching power and low-power design.

Two ways to steer a current

Both transistors let a small control signal modulate a larger current, but the physics differs, and nearly every trade-off in this article follows from it. In a bipolar junction transistor, a forward-biased base-emitter junction injects minority carriers into the base, and most of them are swept across the thin base into the collector. In a MOSFET, a gate voltage across a thin oxide builds an inversion channel, and drain current flows through that channel by drift of majority carriers.

Interviewers love this question because the clean answer is one line: the BJT is controlled by base current, the MOSFET by gate voltage. A strong answer then explains what that means for gain, impedance, speed and heat. If you want the device-level basics first, our MOSFET vs IGBT article covers the high-power end of the same family.

Transfer characteristics and transconductance

The BJT collector current follows an exponential law, Ic = Is·exp(Vbe/VT), with VT about 26 mV at room temperature. Differentiate and you get gm = Ic/VT, roughly 38 mS per milliamp. A BJT at 1 mA therefore has a transconductance of about 38 mS, regardless of its size.

A MOSFET in strong inversion is square-law: Id = ½·µn·Cox·(W/L)·(Vgs−Vth)², giving gm = 2·Id/Vov. For a typical overdrive of 0.2 V at 1 mA, gm is only 10 mS. That is the reason a BJT amplifier can reach more gain from the same bias current. Deep in weak inversion a MOSFET becomes exponential too, with gm/Id approaching 1/(n·VT), but still a little below a BJT because of the slope factor n of about 1.3 to 1.5.

Key fact: at equal bias current, a BJT always has higher gm than a strong-inversion MOSFET. Analog designers pay for MOSFET gm with width, which costs area and capacitance.

Input impedance: why gates are easy to drive

The BJT input looks into a forward-biased diode. The small-signal input resistance is rπ = β/gm. With β = 100 at 1 mA, rπ is about 2.6 kΩ. That loads the previous stage, and the base current also shifts the DC operating point, a classic source of error in high-impedance sensor circuits.

The MOSFET gate sits behind an oxide a few nanometres to a few hundred nanometres thick, so the DC input resistance is effectively above 10¹² ohms. The catch is capacitance: a small signal MOSFET has Cgs of tens of femtofarads, while a power MOSFET can have a gate charge of 50 nC or more. At high frequency the gate looks like a capacitor, and the driver must source real current to move it.

Speed: minority versus majority carriers

Switching a saturated BJT is slow because excess minority charge piles up in the base and must be removed before the collector current falls. That storage time can run to hundreds of nanoseconds in discrete parts. A MOSFET has no minority storage, so its speed is limited only by how quickly the gate and Miller capacitances are charged.

For small-signal gain, the transition frequency fT is the number to quote. Modern SiGe heterojunction bipolars reach fT above 300 GHz and are still used in mmWave radar and optical transceivers. Advanced CMOS also reaches several hundred GHz for NMOS fT at deep-submicron nodes, but the bulk of its value is that it scales: shorter channels give both higher speed and lower switching energy, which BJTs do not enjoy.

Key fact: a BJT turn-off is gated by stored base charge (storage time), while a MOSFET turn-off is gated by gate charge, which a good driver removes in nanoseconds.

Power handling and thermal behaviour

Temperature behaviour decides how safely a device can be paralleled. A BJT's Vbe falls about 2 mV per degree, so a hotter transistor conducts more current at fixed Vbe, heats further and can run away, ending in second breakdown. Designers add emitter ballast resistors to tame this.

A power MOSFET's on-resistance rises with temperature (often 1.5 to 2 times higher at 125 °C), which pushes current away from the hot die and makes paralleling natural. Conduction loss is I²·Rds(on), so a 5 mΩ part at 20 A dissipates 2 W. A BJT loses power as Vce(sat) × Ic, which has a floor of roughly 0.1 to 0.3 V, so MOSFETs win at low voltage and high current. At very high voltage and current the IGBT, covered in our MOSFET vs IGBT guide, combines a MOS gate with bipolar conduction.

Side-by-side comparison

Where each one is used in practice

• Bandgap references and temperature sensors: the predictable exponential Vbe and the ΔVbe between two unequal currents give PTAT voltages (about 0.086 mV/K per ln-ratio unit). Even CMOS processes include parasitic vertical PNPs for this reason.
• Low-noise and RF front-ends: SiGe HBTs deliver high gm and low base noise in LNAs and mmWave VCOs.
• Digital logic and memory: only MOSFETs. A CMOS gate draws almost no static current, which is what allows billions of transistors on a die.
• Buck converters and motor drives: MOSFETs switch at hundreds of kHz to MHz with low loss.
• BiCMOS: combines both on one die, using bipolars where gm and matching matter and CMOS for logic.

In simulation, you would compare them using Cadence Spectre or Synopsys HSPICE with foundry compact models (BSIM-CMG for FinFETs, VBIC or HICUM for bipolars). Sweeping gm/Id against Id/W is the standard way to size a MOSFET, and the same plot beautifully shows why the BJT sits at the upper bound. Our gm/Id design methodology article walks through this sweep.

Key fact: when a datasheet quotes "β = 100", that number is a function of collector current and temperature, not a constant. MOSFET Vth also drifts, roughly −2 mV per °C.

If you want hands-on practice sizing both device types, the analog track in our CourseTron courses includes SPICE labs. For production design and silicon support, the team at Avecas design services works on analog and mixed-signal blocks.

Ready to design real amplifiers and references with both devices? Join the Analog IC Design course on CourseTron and build your skills with guided SPICE labs.

Frequently asked questions

Why does a BJT have higher gain than a MOSFET at the same current?

Because its collector current is exponential in Vbe, gm = Ic/VT, which does not depend on device geometry. A strong-inversion MOSFET has gm = 2Id/Vov, and Vov is usually 150 to 300 mV, much bigger than VT = 26 mV.

Can a MOSFET replace a BJT in every circuit?

No. Bandgap references, precision matching of differential pairs and very low-noise front-ends still favour bipolars, and MOSFET gate oxide is more fragile against electrostatic discharge.

Why are MOSFETs preferred for paralleling?

Their on-resistance has a positive temperature coefficient, so a hotter die carries less current and the paralleled devices share load automatically. BJTs need ballast resistors to avoid thermal runaway.

What does fT mean for each device?

fT is the frequency where short-circuit current gain falls to unity, roughly gm divided by 2π times the input capacitance. Both device types are compared on this metric.