AC Light Dimmer Circuit

• electronics

Intro

One of my new years resolutions is to accomplish something in a side project every week, whether that means a major step in a big project or the completion of a smaller project (another is to blog more, so these work together nicely). Here’s this week’s.

When I moved into my old apartment, I bought the cheapest floor lamp in Ikea. Part of the reason it is so cheap is that the wire runs along the outside of the pole, which doesn’t look particularly attractive. So one evening I got a metal saw and cut a new whole near the base and ran the wire through the inside of the pole, increasing it’s attractiveness dramatically.

One consequence of this was I lost the power switch to the lamp. At the time it was on a wall switch circuit, but now that I’ve moved, the only way to turn it on and off is to use the plug. I was considering a few quick electrical ways to fix this, and at the time I had just started playing with app development, so I decided I’d make a app+bluetooth-controlled switch for it. Then, If I’m going to that much trouble anyway, why not make an AC dimmer while I’m at it?

AC Transistors

It turns out AC dimmers are a little more complicated than I had suspected. This Stack Exchange answer was the main source of design, but I had to do some other research and tweaks to get it to work. Let me break down what I’ve learned.

Electric lights of all kinds (incandescent bulbs, LEDs, etc) typically require some specific voltage to function. In some cases, lowering the voltage will lower the brightness, but you don’t necessarily have much control of the brightness this way. Instead, lights are typically dimmed using Pulse-Width Modulation. Essentially, this means switching the light on and off really fast (so fast that you can’t see the individual pulses) and changing the ratio of how long it spends on to how long it spends off (known as the duty cycle).

Typically switching is done using transistors. But while BJT and MOSFET switching operation is somewhat straightforward, switching AC signals is a bit less to. To do so, you use a TRIAC.

To understand the operation of a triac, you first need to understand Silicon-controlled rectifiers. SCRs are three pin transistors which act somewhat like a gate-controlled diode (as their circuit symbol suggests). An SCR is normally nonconductive, both when reversed biased (like a diode) and even when forward-biased (unlike a diode). However, when a current is applied to the gate and the device is forward-biased, it begins conducting. Here’s were it gets a little strange: after that, the device will remain conductive no matter what happens to the gate, causing it to act like a latch of sorts. The only way for it to return to a non-conductive state is for the bias to fall to zero, at which point it returns to the non-conductive state.

A triac works like two SCRs connected in anti-parallel, allowing the gate to control the conductivity in both directions. Each time the the signal alternates directions, one of the SCRs will be reversed-biased and the device will return to its non-conductive mode.

Dimmer circuit

So to use it as a dimmer, we need to trigger the gate in the middle of each half-cycle. Triggering at the beginning of the half-cycle means the signal is conducted for most of the time, and the light will be bright, while triggering near the end of the half-cycle means the light is switched off for all but a very brief period of time, making it appear dim.

The problem here is you need to know when a new half-cycle starts, which is the purpose of a zero-cross detector. According to the Stack Exchange post, zero-cross detection is done using an optocoulpler such as the TCET1600 that I used. Inside this chip are LEDs and phototransistors. When an AC signal is connected in this configuration, the output is a short pulse at every zero crossing (when the LEDs turn off and the transistor stops shorting the output to ground). It also has the benefit of acting as an optoisolator, protecting the low-power digital circuitry from the power mains.

Dim level control

The post suggests using a microprocessor to control the delay between this signal and triggering the triac. However, since the delay needs to be between 0 and \(\frac{1}{60Hz}\frac{1}{2} \approx\) 8.3 ms for a full range of dimming. With an arduino, you can delay integer multiples of 1ms, which doesn’t give you very much control over the brightness.

Instead, I decided to use a 555 timer in monostable mode. The signal time is \(t=RC\ln(3)\approx 1.1RC\). With a 10K potentiometer, the desired capacitance is \(0.76 \mu F \approx 0.86 \mu F\). I didn’t have one of exactly this value on hand so I used a \(1 \mu F\) instead, which caused the aliasing seen in the video. Since the 555 timer is started by TRIG going low, we need an inverter before it, and because we want the triac to trigger after the 555’s pulse, we need to inverter after as well.

As the Stack Exchange post points out, the circuit isn’t actually optoisolated unless the output is protected as well. For that, it suggests using the MOC3051, which uses LEDs to trigger a triac. Since the MOC3051’s triac is a little sensitive to high voltages, we use it to trigger a second higher power triac (in this case, a BT136).

Next steps

With all of this together, you can see the input and output signals on the oscilloscope in the video below. To finish the project, I need:

• A transformer to step down the voltage into the zero-cross detector
• To replace the resistor with the PWM output of a microprocessor
• A bluetooth module for the microprocessor
• A snazzy case (maybe whip up an Eagle schematic and print a board?)

Video!