How to Drive a TRIAC: Gate Drive, Optocoupler & Snubber Basics

First, Understand What the Gate Needs: Current, Not Voltage

A TRIAC triggers when gate current exceeds the datasheet's I_GT — the trigger current, not a gate voltage. The voltage just has to be enough to push that current. For the JST134Q-600D, I_GT is 5mA in quadrants 1–3 and 10mA in quadrant 4, with V_GT 1.0V max.

Here's the failure chain: an MCU pin supplies about 5mA through a resistor — right at the edge of I_GT for a 5mA gate, and below it for quadrant 4. The part triggers intermittently, or only in some half-cycles, and the load flickers.

That's the classic "it works on the bench, not in the product" triac symptom.

But there's a second half to the story. The gate pulse must persist until the load current exceeds latching current — otherwise the triac drops out the instant the pulse ends.

Short pulses from a one-shot or a mis-sized RC are a real failure mode, confirmed across forum threads on dimmer circuits.

Why You Need an Optocoupler — and Which One

The MCU's ground is not the mains neutral. A bare TRIAC gate sits at the mains-side potential of T1 — drive it from a GPIO and you've connected your logic to the wall.

The opto-triac driver fixes this: the MCU drives an LED, the LED fires an internal triac, and the isolation is guaranteed.

For the LED side: pick R1 so the LED current lands at 10–20mA. With a 3.3V MCU and a 1.2V LED forward drop, R1 = (3.3 − 1.2) / 0.015 ≈ 140Ω — use 150Ω. That's the whole MCU-side design.

For the gate side: the opto's output fires the main TRIAC through R_G. Size it for the mains peak: R_G ≈ (V_peak − V_TM) / I_GT with 2× margin — 100–330Ω covers most sockets from 120V to 230V.

The gate current must return to T1 (MT1), not to the load side of the triac. Get that wrong and R_G carries the full load current — a classic board-fry, and a classic forum question.

Zero-cross versions fire only near the mains null — the load turns on with minimal EMI. Non-zero-cross versions (MOC3021 class) fire at any point in the cycle, which is exactly what phase control needs. Pick the type by job, not by habit.

The other classic opto failure: leakage false-triggering. The opto's internal triac leaks a few µA at temperature, and that leak can fire the main TRIAC. A 47–100Ω resistor from gate to T1 shunts it — the resistor is in the datasheet circuits for a reason.

Quadrants: The Polarity Lesson That Breaks Benches

TRIACs trigger in four quadrants, defined by the polarity of T2 relative to T1 and the gate. And here's the catch: not all quadrants are equally sensitive. On the JST134Q-600D, quadrants 1–3 need 5mA and quadrant 4 needs 10mA.

Real forum cases show the consequence: a "quadrant-3 works but quadrant-1 doesn't" triac turned out to be a part that needed a negative gate drive for half the cycle, or a driver whose polarity only fired some quadrants.

The fix is the same in every case — drive the gate with a source that covers all four quadrants, sized for the highest I_GT (quadrant 4).

An AC-coupled or opto-triac drive does exactly that: the opto fires with whatever polarity the mains presents, and if it can push 10mA+, every quadrant triggers. The D-class gate — 5mA in three quadrants — is why "sensitive gate" parts exist.

Zero-Cross or Phase Control? Pick by EMI and Function

The two jobs a TRIAC does are fundamentally different. Zero-cross switching turns the load fully on or off, near the mains null, for minimum EMI. Phase control holds a partial conduction angle — dimming, motor speed — and generates EMI by design, because you're chopping a sine wave mid-cycle.

Use zero-cross for heaters, resistive loads, and appliance ON/OFF switching. The EMI signature is tiny and the load sees a clean step. Use phase control only when you actually need variable power — a dimmer, a fan speed controller, a soft-start ramp.

And use a snubber with both. Phase control with an inductive load produces the worst case: the load current and voltage are phase-shifted, and at the moment of commutation the TRIAC sees a voltage step while current is still flowing.

That's where the commutation dV/dt rating — 2.5V/µs on the JST134Q-600D — gets violated, and the triac false-fires or fails.

Snubbers and dV/dt: Stopping False Triggering

Every TRIAC has two dV/dt ratings, and confusing them is where designs go wrong. Static dV/dt (120V/µs on the JST134Q-600D) is the rate of voltage rise across the off-state device that would self-trigger it.

Commutation dV/dt (2.5V/µs) is the rate at the moment of turn-off, with load current still decaying — and it's 50× weaker.

The RC snubber across T1–T2 slows both. Typical starting values: 47–100Ω in series with 10–100nF, right across the triac terminals.

The resistor limits the discharge current, the capacitor absorbs the voltage step. For motor loads, size it to the motor's inductance and verify with a scope — "typical" values are a start, not a finish.

The static side has its own fix: leakage from the opto-triac, or capacitive coupling on long gate traces, can fire the gate with no pulse at all. The gate-to-T1 resistor (47–100Ω) shunts that leakage.

Both fixes are cheap, both are skipped constantly, and both show up in every forum thread titled "my triac turns on by itself."

The Drive Decision Framework

✅ Use an opto-triac with phase control (MOC3021 class) when: you're dimming lamps or speed-controlling a motor; the load is resistive or lightly inductive; you accept the EMI and add the snubber.

✅ Use a zero-cross opto-triac (MOC3041/MOC3061 class) when: you're switching loads on/off; EMI matters; the load is heater, resistive, or appliance-class.

✅ Add the snubber and gate-to-T1 resistor when: the load is inductive (motor, solenoid, transformer); the triac is anywhere near a noisy rail; the product must pass EMC testing.

❌ Don't drive the gate directly from an MCU pin when: the mains is involved (isolation); the gate needs more than 5mA (it does); you want the design to survive temperature and aging. The opto costs cents and removes an entire failure class.

❌ Don't use a TRIAC at all when: the load is DC (it latches), the load is under ~20mA (below holding current), or the environment demands a physical gap when off (use a relay).

The one-sentence summary: treat the gate as a current budget with a return path, isolate with an opto sized for quadrant 4, and buy the snubber before you need it. The triac isn't the fragile part — the drive is.