Unijunction Transistors (UJT)

Quick Answer

A unijunction transistor is a bar of silicon with two base contacts and a single junction part way along it. The bar divides the supply, and the junction stays reverse biased until the emitter passes that fraction of it. Then the bar's resistance collapses and the device dumps whatever charge was waiting.

The name is a warning that this is not an ordinary transistor. There is one junction rather than two, there is no useful current gain, and it cannot amplify anything. What it has instead is a threshold it sets for itself, and a violent change of state once that threshold is passed.

Take a bar of lightly doped silicon with a contact at each end and a supply across it. The bar behaves as a resistor, so every point along it sits at some fraction of the supply. Form a junction against the bar part way along, call it the emitter, and that fraction is what the emitter has to beat before anything happens — a fraction with a name, the intrinsic standoff ratio. Below it the junction is reverse biased and the device draws only the small current running end to end through the bar.

Go one junction drop above it and the change is not gradual. Carriers injected into the lower half of the bar make it far more conductive, which lowers the potential at the emitter, which increases the injection. The emitter voltage falls while its current rises — a negative resistance — and the device is fully on within microseconds. A cistern behaves the same way: it fills slowly, does nothing at all for most of the cycle, and then reaches the level at which the siphon takes over and empties in one movement.

That discontinuity is what the part was built for. Charge a capacitor through a resistor into the emitter and it sits doing nothing until it reaches the peak point, at which the device dumps the capacitor through a small resistor at base one. The result is a short, sharp, repeatable pulse, followed by another after the next RC charge — exactly what a thyristor's gate wants, and for two decades the standard way to provide it. It is why the part turns up inside almost every older motor controller and lamp dimmer.

What replaced it won on cost rather than on any failing of its own: a diac does the same job in phase control with two terminals instead of three, a 555 does it with better temperature stability, and a microcontroller does it while also doing everything else. Two notes survive the obsolescence. The threshold is a fraction of the supply across the bar, so the oscillator's period moves with that supply unless it is regulated. And anything built around one to fire a thyristor is usually sitting at line potential, which is the hazard phase control treats at length.