The simple Triac light dimmer (or ceiling fan speed controller) has a lot of uses. However the simplicity of design, while perfectly suited to nice, well behaved resistive loads, doesn't work so well for loads that are reactive, or vary a lot during a half cycle (like a transformer driving a spark gap or discharge tube).


Here is the "canonical" TRIAC dimmer schematic. The R and C form a phase shift (delay) network which delays the line voltage. When the voltage across the diac gets big enough, it conducts, triggering the triac ON. In older designs, a neon bulb was used instead of the diac. There are also devices called Quadracs which combine the diac and triac in one package.
Here is the usual "two terminal" dimmer schematic. Note that the voltage across the phase shift network is the voltage across the triac. For resistive loads, this makes almost no difference, but, for inductive or capacitive loads, this might not be true, since the load current and voltage may not be in phase.
This is a conceptual view of how the triac dimmer looks, using the back to back SCR model of a triac. This isn't perfectly accurate, but will serve to illustrate the problem. Note that on a positive going half cycle, when the voltage across the diac gets high enough, it conducts, and D2 will turn on. But, for a negative going half cycle, the path to turn on D1 is through the load. If the load isn't a nice resistive one, this might cause triggering problems.

Reactive loads can sometimes be tamed by the expedient of putting a resistive load in parallel. A small light bulb or a resistor works well. More sophisticated phase control circuits use isolated trigger circuits (using transformers or optoisolators) to generate very consistent trigger pulses. For some loads, preceding the dimmer with a bridge rectifier might help improve the consistency, because the polarity of the pulses will always be the same. Another approach is to put the bridge around the dimmer, as illustrated in the figure below.

hv/dimmer.htm - 1 November 2001 - hv main page - Jim's home page - mail to Jim Lux