TRIAC

What a TRIAC actually is

Strip away the marketing name and a TRIAC is a five-layer NPNPN silicon structure with three terminals: two main terminals, MT1 and MT2 (older datasheets call them A1/A2), and a gate. Unlike a diode or an SCR, it has no fixed anode and cathode — it is symmetric, so it can block and conduct in either direction. The simplest mental model is two SCRs connected in inverse-parallel, sharing one gate. One internal "SCR" handles the positive half-cycle (MT2 positive with respect to MT1); the other handles the negative half-cycle. The word itself is a contraction of TRIode for Alternating Current.

The defining behavior is latching. Apply a small gate current and the device snaps from a high-impedance "off" state into a low-impedance "on" state, dropping only about 1–1.5 V across MT1–MT2 regardless of how many amps flow. Once it is on, the gate loses all control — you cannot turn it off by removing the gate signal. It conducts until the main-terminal current naturally falls below the holding current I_H, at which point it reverts to off and waits for the next trigger. On the AC mains, the current passes through zero 100 or 120 times a second, so the device self-commutates twice per cycle without any effort.

Triggering, latching and holding

Three current thresholds govern a TRIAC, and confusing them is the source of most beginner bugs:

• Gate trigger current (I_GT): the gate pulse needed to fire the device — typically 5–35 mA for small parts, with a corresponding gate voltage V_GT of ~0.7–1.5 V.
• Latching current (I_L): the minimum main-terminal current that must flow while the gate is still applied for the device to stay latched after the gate is removed. Usually a few times I_H.
• Holding current (I_H): the minimum main-terminal current to remain latched once the gate is gone, typically 5–50 mA.

The gate pulse only needs to be a few microseconds to tens of microseconds long — just enough to push the current above I_L. After that the load itself keeps the current above I_H until the AC zero crossing. This is why TRIAC drive circuits sip power: a brief, low-energy pulse controls hundreds of watts of load.

Phase-angle control: the core trick

A TRIAC does not vary like a rheostat; it is a switch. To dim a lamp you do not reduce its peak voltage, you change when in each half-cycle the switch closes. At the start of every half-cycle the TRIAC is off and blocks the mains. A timing circuit waits a controllable firing angle α (measured in degrees of the 180° half-cycle) and then fires the gate. The TRIAC conducts the remaining conduction angle (180° − α) and shuts off at the next zero crossing. Repeat, symmetrically, for the opposite polarity.

The RMS voltage delivered to a resistive load as a function of firing angle α is:

V_rms(α) = V_peak / √2 · √( (1/π) · [ (π − α) + (sin 2α)/2 ] )

where:
  V_peak = peak line voltage  (V)
  α      = firing angle  (radians, 0 → π per half-cycle)

Power into a resistive load scales as V_rms²:
  P(α) = V_rms(α)² / R

At α = 0 the device fires at the zero crossing and conducts the whole half-cycle — full power. At α = 90° (π/2) the load gets exactly half its average power. At α near 180° almost no power flows. The relationship is markedly nonlinear, which is why cheap dimmers feel "crowded" at one end of their travel and why better dimmers shape the control curve in firmware.

The classic DIAC + RC trigger

The textbook dimmer generates its firing angle with a resistor-capacitor network feeding a DIAC (a bidirectional trigger diode). As the AC half-cycle rises, the cap charges through the pot and a fixed resistor. When the cap voltage reaches the DIAC's breakover voltage (typically ~30 V), the DIAC suddenly conducts, dumping a current spike into the TRIAC gate and firing it. Turning the potentiometer changes the RC time constant, which changes how long the cap takes to reach breakover, which changes α.

Approximate firing delay set by the RC network:
  t_fire ≈ R · C · ln( V_peak / (V_peak − V_BO) )

  R    = pot + fixed resistor  (Ω)
  C    = timing capacitor  (F)
  V_BO = DIAC breakover voltage  (~30 V)

Firing angle:   α = 2πf · t_fire   (with f = line frequency)

The DIAC matters: without it, the TRIAC gate would receive a slow, mushy ramp instead of a sharp pulse, and the firing angle would drift with temperature and from positive to negative half-cycles. The DIAC's snap-action and symmetry keep the two half-cycles balanced, preventing a DC component that would saturate transformers and audibly hum motors. The well-known cost of the bare RC+DIAC dimmer is hysteresis ("snap-on" and "drop-out" at different settings) because the cap retains charge between half-cycles; a second RC stage ("double-time-constant" dimmer) fixes most of it.

TRIAC vs. the alternatives

	TRIAC	SCR (thyristor)	Back-to-back MOSFETs	IGBT (+ bridge)	Mechanical relay
Conduction	Bidirectional	Unidirectional	Bidirectional	Unidirectional (needs bridge for AC)	Bidirectional
Control signal	Brief gate pulse, latches	Brief gate pulse, latches	Continuous gate voltage	Continuous gate voltage	Coil current
Can force turn-off?	No — waits for I < I_H	No — waits for I < I_H	Yes — any time	Yes — any time	Yes (with arc/contact bounce)
On-state loss	~1–1.5 V drop	~1–1.5 V drop	I²·R_DS(on), very low	~1.5–2.5 V drop	~contact resistance, mΩ
Switching frequency	Line frequency only	Line frequency only	10 kHz–MHz (PWM)	1–100 kHz	~Hz, wears out
dV/dt sensitivity	High — needs snubber	Moderate	Low	Low	None
Typical power	up to a few kW	kW → MW	up to ~kW	kW → MW	Any (slow)
Best for	Dimmers, fan/heat control	High-power rectifiers, HVDC	Solid-state relays, DC	Inverters, motor drives	Isolated on/off

Worked example: dimming a 100 W lamp

A 100 W incandescent lamp runs on a 230 V RMS, 50 Hz supply. The peak line voltage is V_peak = 230·√2 ≈ 325 V. We fire the TRIAC at α = 90° (π/2 rad) into each half-cycle. What power does the lamp receive?

At α = π/2:
  (π − α) + (sin 2α)/2 = (π − π/2) + (sin π)/2
                       = π/2 + 0
                       = π/2

  V_rms = V_peak/√2 · √( (1/π)·(π/2) )
        = 230 · √(1/2)
        = 230 · 0.707
        = 162.6 V

Lamp resistance (cold-ish, at full power):
  R = V_full² / P = 230² / 100 = 529 Ω

Power at α = 90°:
  P = V_rms² / R = 162.6² / 529
    = 26,439 / 529
    = 50 W

Firing at the half-cycle midpoint delivers exactly half the power — the lamp glows at roughly half brightness (perceptually a bit dimmer, since incandescent output is nonlinear with power). Note the half-cycle timing: at 50 Hz a half-cycle lasts 10 ms, so α = 90° corresponds to a 5 ms delay; the TRIAC then conducts the final 5 ms. The same logic at 60 Hz uses an 8.33 ms half-cycle and a 4.17 ms delay.

Key ratings and how to read a datasheet

• V_DRM / V_RRM — peak off-state voltage: the most a TRIAC can block. For 230 V mains (325 V peak), a 600 V part is standard; 800 V gives margin against spikes.
• I_T(RMS) — RMS on-state current: the continuous current rating, e.g. 4 A, 8 A, 16 A, 25 A. Size for RMS load plus headroom.
• I_TSM — surge current: the non-repetitive single-cycle peak, often 10–20× I_T(RMS), for inrush from lamps and motors.
• I²t — fusing rating: matches the device to a protective fuse so the fuse blows before the silicon does.
• dV/dt and dI/dt — commutation limits: the static off-state dV/dt and the more critical commutating dV/dt (turn-off) that inductive loads stress.
• I_GT / V_GT — gate sensitivity: a "sensitive-gate" TRIAC (I_GT ≈ 3–10 mA) can be driven straight from a microcontroller pin; standard parts need a transistor or optotriac driver.

Failure modes and trade-offs

• dV/dt re-triggering on inductive loads. When current and voltage are out of phase, the device sees a fast voltage step at commutation. Exceed the commutating dV/dt and internal capacitance re-fires it — the TRIAC never turns off. Fix: an RC snubber (classically 100 Ω + 0.1 µF) across MT1–MT2, or a "snubberless"/Hi-Com TRIAC characterized for inductive use.
• Drop-out from low holding current. Very small loads (an LED retrofit bulb, a tiny fan) may not keep current above I_H for the whole conduction window, so the TRIAC drops out mid-cycle and the load flickers. Fix: a bleeder resistor or a minimum-load element to keep I > I_H. This is exactly why many old TRIAC dimmers flicker or buzz with modern LED bulbs.
• Quadrant IV unreliability. The MT2−/gate+ quadrant needs far more gate drive and triggers erratically. Design the gate drive so it never relies on Q-IV; better still, choose a 3-quadrant device.
• EMI from sharp turn-on. Chopping the waveform mid-cycle creates a steep current step rich in harmonics, radiating RF hash. Mains-rated dimmers add an LC filter (an inductor "choke" and an X-capacitor) to meet conducted-emissions limits.
• Thermal runaway / exceeding I_T. The ~1.3 V on-state drop times the RMS current is real heat — a 230 V, 8 A load dissipates roughly 10 W in the TRIAC. Inadequate heatsinking pushes the junction past ~125 °C and the device fails short, often welding the load permanently on.
• No galvanic isolation. The gate shares the MT1 reference, which sits on the live mains. A microcontroller must drive the gate through an opto-isolator (an "optotriac") so the low-voltage logic never touches mains potential.

Where TRIACs show up

• Light dimmers: the canonical application — phase-angle control of incandescent and dimmable-LED lamps from a single wall device.
• Fan and small-motor speed control: ceiling fans, blowers and universal motors, where reduced RMS voltage lowers speed.
• Heater and oven control: often using burst firing (whole cycles on/off) instead of phase control to minimize EMI on slow thermal loads.
• Solid-state relays and contactors: zero-crossing optotriacs switch loads at the zero crossing to eliminate inrush and EMI.
• Soft-starters: ramping the firing angle from near-180° down to 0° gently spins up a motor and limits inrush current.