Two transistors, two control knobs
Both devices are the same tap from earlier rungs: a small movement of a handle controls a big flow. But they differ in what the handle is. In the BJT the handle is a current — you push current into the base and a much larger collector current follows. In the field-effect transistor, and the MOSFET in particular, the handle is a voltage — you set a voltage on the gate and a channel between the other two terminals opens up. One is current-controlled, the other voltage-controlled, and almost everything else flows from that one distinction.
The first consequence is dramatic: the gate draws almost no steady current. A BJT pays a toll to stay on — it must keep sinking a base current Ib = Ic / beta the whole time it conducts. Switch a 10 A load with a transistor whose beta is 50 and you must feed the base 10 / 50 = 0.2 A, continuously, just to hold it on. A MOSFET asks for no such tribute: once the gate sits at, say, 10 V, the channel stays open while the gate sips essentially zero DC current. The control terminal is, for steady signals, almost free.
The names line up neatly. The MOSFET's gate plays the role of the BJT's base, its source plays the emitter, and its drain plays the collector. The controlling variable is the gate-to-source voltage, Vgs: push it past the threshold voltage Vth and the channel turns on, much as pushing Vbe past ~0.7 V wakes a BJT. The difference is that a BJT's input looks like a forward-biased diode that conducts, while a MOSFET's input looks like a tiny insulated plate that does not.
The honest catch: the gate is a capacitor
"The gate draws no current" is a truth about DC, not about change. That insulated gate plate sitting over the channel is exactly the recipe for a capacitor — two conductors with a thin insulator between them — so the gate is a tiny rechargeable bucket. Holding a steady voltage on a full bucket costs nothing, but every time you want to change that voltage you must shovel charge in or pump it back out. The total charge you must move to turn the device fully on is its gate charge, Qg.
Put a number on it. A power MOSFET with Qg = 30 nC that you want to switch on in 100 ns needs an average gate current of I = Q / t = 30 nC / 100 ns = 0.3 A during that switching instant. So "no gate current" and "needs a third of an amp to switch" are both true — one is the steady state, the other the transient. This is why a real design hands the gate to a gate driver: a beefy little buffer whose only job is to slam charge in and out fast. Switch slowly and the device lingers half-on, burning switching losses — the subject of a later guide in this rung.
As a switch: a resistor versus a fixed toll
When fully on, the two devices drop voltage in completely different ways, and this decides who wins as a switch. A saturated BJT switch sits at a roughly fixed collector-emitter voltage, Vce(sat), of about 0.2 V — a toll booth that charges the same toll whether a trickle or a torrent passes through. A MOSFET that is fully on behaves instead like a plain resistor, its on-resistance Rds(on), so its drop is Rds(on) times I and simply scales with current.
Watch what that means at, say, 5 A. A modern MOSFET with Rds(on) = 10 mohm drops only 5 times 0.01 = 0.05 V and wastes P = I^2 times R = 5^2 times 0.01 = 0.25 W. The BJT, stuck at its 0.2 V toll, wastes P = V times I = 0.2 times 5 = 1.0 W — four times the heat for the same job. At low and moderate currents the MOSFET is the cooler, more efficient switch, which is why power switching is overwhelmingly MOSFET territory today.
Be honest about the crossover, though. Because the MOSFET's loss grows as I^2 times R while the BJT's grows only as a fixed-voltage times I, at high enough current a high-Rds(on) MOSFET can actually dissipate more than the BJT — read the real Rds(on) (it rises with temperature) from the datasheet rather than trusting the headline. The same low-versus-high choice shows up in high-side and low-side switching: an N-channel device switches a load's ground leg (low side) easily, but to switch the supply leg (high side) its gate must be driven above the positive rail — a wrinkle a later guide handles with a charge pump or bootstrap.
Property BJT (bipolar) MOSFET (field-effect) ---------------------------------------------------------------------- Controlled by base CURRENT Ib gate VOLTAGE Vgs Control terminal draws Ib = Ic/beta draws ~0 DC (a capacitor) On-state voltage fixed Vce(sat) ~0.2 V Rds(on) x I (a resistor) gm at same current high (Ic / 25mV) lower (square-law) Gets hotter -> more current (runaway) Rds(on) rises -> self-limits Putting in parallel hard (one hogs current) easy (they share) Hold it ON needs continuous Ib just a held voltage
As an amplifier: who has more gain per milliamp?
For small signals what matters is transconductance, gm — how much output current you get per volt of input wiggle — because that, times a load resistor, is your voltage gain. Here the BJT has a quiet advantage. A BJT's collector current rises exponentially with Vbe, giving the strong gm = Ic / VT you met in the BJT rung. A MOSFET's drain current rises only as the square of the gate overdrive (Vgs minus Vth), a gentler curve, so for the same operating current its gm is usually several times smaller.
Put numbers on it at 1 mA. The BJT gives gm = 1 mA / 25 mV = 40 mA/V; a typical small-signal MOSFET at the same 1 mA might manage only a few mA/V. So the BJT squeezes more gain from each milliamp, which is one reason precision, low-offset, low-noise analog front-ends so often stay bipolar. The MOSFET earns its keep when you build a common-source amplifier or a source follower (the next guide) and need a near-infinite input that loads the source not at all.
That near-infinite input is the MOSFET's killer feature for analog. Because the gate is insulated, the input resistance is astronomically high and the input draws essentially no bias current — perfect for reading a high-impedance sensor, a piezo element, or holding a voltage on a capacitor without it leaking away. One naming trap to remember, though: in a BJT "saturation" means fully on as a switch, while in a MOSFET the saturation region is the opposite — it is the active, flat region where the device amplifies. Same word, opposite meaning.
Heat, paralleling, and the verdict
Now the difference that decides high-power and rugged designs: how each behaves as it heats up. A BJT held at a fixed base drive is treacherous — as it warms, its Vbe falls, which lets through more current, which makes it hotter still, a vicious circle called thermal runaway that can cook the part. A MOSFET does the friendly opposite: its Rds(on) rises with temperature, so a hotter spot carries less current and quietly cools itself. The device has a built-in negative feedback against runaway.
That single trait makes MOSFETs easy to gang together: wire several in parallel and if one starts hogging current it heats, its resistance climbs, and it politely hands current back to its cooler siblings until they share evenly. BJTs do the dangerous reverse — the hottest one grabs ever more current — so paralleling them needs balancing resistors and care. Combine self-balancing, the near-free voltage drive, and the low on-resistance, and you see why MOSFETs dominate power switching and, paired N-with-P, build the CMOS inverter at the heart of every digital chip (the final guide of this rung).
So, the verdict — and it is genuinely a tradeoff, not a winner. Reach for a MOSFET when you are switching power, building digital logic, driving from a weak voltage source, or need a sky-high input impedance; its self-balancing and low on-resistance are hard to beat. Reach for a BJT when you want the most transconductance per milliamp, the lowest offset and noise in a precision or audio front-end, simple low-cost small-signal gain, or some RF jobs. Most everyday tasks either device can do; the table above, plus the honest caveats, is how you choose.