A handle moved by pressure, not flow
In the last rung you tamed the BJT, and its bargain was simple: to keep it on you had to keep feeding a small but real base current. The transistor was a tap whose handle you had to hold down, and holding it cost a steady trickle. The field-effect transistor, or FET, strikes a different bargain. Its handle moves in response to a voltage alone, and once you set that voltage it stays put while drawing essentially no current at all.
The star of this rung is the MOSFET — the metal-oxide-semiconductor field-effect transistor, the workhorse of nearly all modern electronics. Where the BJT is current-controlled (collector current follows base current), the MOSFET is voltage-controlled: the voltage on its control terminal sets how wide a channel opens for current to flow, and that control terminal is sealed behind a sliver of insulator so almost no current leaks into it. Picture a sluice gate raised and lowered by a lever you push through a glass wall — your push sets the gate height, but no water ever touches your hand.
That tiny detail — a control terminal that costs no steady current — is why the MOSFET, not the BJT, became the atom of the digital age. A modern chip holds billions of these switches, and because an idle MOSFET sips essentially zero gate current, they can pack shoulder to shoulder without melting. Hold onto this picture as we open the device up: a voltage in, a controlled channel out, and an insulated handle in between.
Gate, source, drain — and the insulated gate
A MOSFET has three working terminals. Current flows along a channel between the source and the drain — these are the two ends of the pipe, analogous to the BJT's emitter and collector. The third terminal, the gate, is the control handle, analogous to the base. Apply the right voltage between gate and source and the channel between source and drain opens up; the gate-to-source voltage, written Vgs, is the single knob that runs the whole device.
The name spells out the structure: metal-oxide-semiconductor. The gate is a little metal (or polysilicon) plate laid over a wafer-thin layer of insulating oxide, which sits on the semiconductor channel beneath. Look closely and you will recognize an old friend: two conductors separated by an insulator is exactly a capacitor. The gate is a tiny capacitor plate, and that is the whole reason no steady current flows into it — a capacitor passes no DC. We will return to this honestly at the end, because that same capacitor is the MOSFET's one real cost.
MOSFETs come in flavours, and two splits matter. First, N-channel versus P-channel: an N-channel device conducts when the gate is pulled positive relative to the source, a P-channel device when the gate is pulled negative — mirror images, just as NPN and PNP were. Second, enhancement versus depletion mode: an enhancement MOSFET is normally off and you must build (enhance) a channel by applying gate voltage, while a depletion MOSFET is normally on and gate voltage pinches it off. By far the most common part — and our default for the rest of this rung — is the N-channel enhancement MOSFET: off until you push its gate positive.
Threshold, then two regions: triode and saturation
Nothing happens until the gate voltage crosses a line called the threshold voltage, Vth. For an N-channel enhancement MOSFET, while Vgs sits below Vth (a typical logic-level part might have Vth around 2 V) the channel simply does not exist and the device is off. Push Vgs above Vth and a conducting channel forms under the oxide; the further above threshold you go, the wider and more conductive that channel becomes. Vth is the on/off line — the MOSFET's counterpart to the BJT's ~0.7 V turn-on — and, just like that 0.7 V, it is a spec with spread and temperature drift, not a sacred constant.
Once it is on, a MOSFET works in one of two regions, and which one depends on the drain-to-source voltage Vds. At small Vds the channel behaves like a plain resistor whose value you set with the gate — this is the triode region (sometimes called the linear or ohmic region), and it is where you run the device as a switch. Raise Vds enough and the channel 'pinches off' near the drain; the current then levels out and barely grows with further Vds — this is the saturation region, where the MOSFET acts like a voltage-controlled current source and is where you run it as an amplifier. Resistor when squeezed low, current source when stretched high.
N-channel enhancement MOSFET (Vgs is gate-to-source) -------------------------------------------------------------- Region Condition Behaves like Used as -------------------------------------------------------------- Cutoff Vgs < Vth open circuit OFF switch Triode Vgs > Vth, Vds small gate-set resistor ON switch Saturation Vgs > Vth, Vds large current source AMPLIFIER
The MOSFET as a switch
To use the MOSFET as a switch you drive the gate well past threshold so the channel opens as wide as it will go, dropping the device deep into the triode region. There it is simply a small resistor between drain and source, called the on-resistance or Rds(on). A good modern power part can have an Rds(on) of just a few milliohms — a near-perfect closed switch. This is the heart of the MOSFET switch: full gate drive, tiny on-resistance, almost no voltage wasted across the device.
- Say a MOSFET with Rds(on) = 10 mohm (0.01 ohm) switches a 5 A load. Fully on, it is just that resistor carrying 5 A.
- Voltage across it: by Ohm's law V = I times R = 5 times 0.01 = 0.05 V, just 50 mV dropped — the load gets almost the full supply.
- Power wasted as heat: P = I^2 times R = 5^2 times 0.01 = 0.25 W. A quarter of a watt — cool enough to need little or no heatsink.
- Compare a BJT switch: in saturation it drops roughly 0.2 V no matter the current, so at 5 A it burns 0.2 times 5 = 1 W — four times the loss. For switching big currents, low Rds(on) is why the MOSFET usually wins.
Where you put the switch matters. The easy case is low-side switching: the MOSFET sits between the load and ground, its source tied to ground, so to turn it on you just lift the gate a few volts above ground — straightforward. The harder case is high-side, with the switch between the supply and the load; now the source floats up near the supply when the device is on, so the gate must be driven above the supply rail to keep Vgs high enough — which is why high-side switches need extra tricks (a charge pump or a special driver). One more structural fact: every power MOSFET comes with a built-in body diode across drain and source, a parasitic diode that conducts in the reverse direction. It is a free flyback path for inductive loads, but it is slow and can bite you if you forget it is there.
Amplifier, power switching, and the honest reality
Park the MOSFET in saturation instead of triode and it becomes an amplifier, exactly as the BJT did in its active region. The controlling figure of merit is again transconductance gm — how much drain current changes per volt of gate change. The two everyday building blocks mirror the BJT ones: the common-source stage gives voltage gain and inverts (the MOSFET sibling of the common-emitter amplifier), and the source-follower gives near-unity gain but a low output impedance, a buffer (the sibling of the emitter-follower). Guide 3 builds both properly; for now just note the same two shapes reappear.
Two destinations close out the rung. Scaled up, the power MOSFET switches amps or tens of amps in supplies and motor drives — but its large gate is a big capacitor, so each turn-on means shoving in a certain gate charge, and that demands a punchy gate driver to charge it fast. Switch too slowly and the device lingers in the lossy in-between, racking up switching losses as heat (guide 4). And stack one N-channel with one P-channel MOSFET and you get the CMOS inverter: whichever input you apply, one device is off, so almost no current flows except during the flip — the low-power cell that the entire digital world is built from (guide 5).