AVR controlled dimmer switch

Although version 1 works flawlessly, there is still a lot of room for improvements. Instead of using an AT89S2051 microcontroller we can also use a RISC (reduced instruction set) controller called ATtiny2313 which is made by the same company. The ATtiny2313 uses the AVR instruction set and is very powerful. The processor is pin compatible and has lots of advantages over the 80x31 based processor. Here are most of them:

Advantages of the newer processor

• This microcontroller has a 3 wire ISP interface and can be programmed using a simpel interface on a PC printer port.
• The processor detects both rising and falling levels on the interrupt pins, which simplyfies the RC5 decoding routine.
• The processor has a built in Power-On-Reset circuit. (Saves 1 component)
• The processor has a built in RC clock. Which can run on either 4 or 8 Mhz without the need of external parts. (This saves 3 more components)
• The processor has a speed, power and instruction set advantage over the 80x31 based version.
• Free C compiler and simulator (WinAVR and AVR Studio 4).

And of course with so many advantages there must be certainly also be disadvantages, yes but only one.

Disadvantages of the newer processor

• The AVR instruction set uses slightly more memory than the 80x31 instruction set.

The other improvement that can be done is using a transformerless design. That gets rids of the bulky transformer but also creates a few new problems.

Let us look at the new diagram

So how does it work?

The triac TIC206D (BT136-600D) needs only a small current flowing through the gate to start conducting and turn a light on. This is the reason why they can be directly connected to the microprocessor, which can sink at least 10mA. Most digital circuits can sink much more current then they can source, that is why the whole dimmer works with -5 Volts. The resistors R1 to R8, limits the current to -5 mA. That it enough to get the triacs conducting in 3 out of the 4 quadrants. In the 4th quadrant the triacs need 10 mA to start conducting, but that is when the gate current is positive. That is another reason why we are using a negative voltage.

R9 in combination with C1, function as a low pass filter. They remove high frequencies caused by the fase cutting of the triacs. L1 prevents them from going back into the power grid where they spoil your radio reception.

Version 1 used a transformer to generate the -5Volt for the electronics circuitry. Although this is the standard way of making a certain voltage, it is not really necessary. We can use a capacitor too, which is 100 times smaller then a transformer. This has of course a few drawbacks too. First the voltage generated is not galvanically isolated from the power grid, so it is dangerous to touch bare wires. The capacitor is the opposite of a transformer, instead of supplying a fixed voltage, the capacitor supplies a fixed current. Also, the capacitor can only supply small currents, which are in the order of milli Amps. The capacitor works most efficiently if the current supplied is also used. The current needs to be discharged if the circuitry doesn't use it, else the voltage will go up and distroy other parts of the circuitry.

Calculations on the power supply unit

I don't want to make a real long story with lots of calculations, but the dimmer needs less than 20mA to function. C2 is calculated with the formula X_c=U/I=230V/0.02mA= 1/(2πfC). f is the net frequency and is in Europe 50 Hz. Fill in the values and you find C to be 276nF. To be sure that we get at least 20mA we use 330nF for C2. For the Americas C2 needs to be C=0.02/(2*3.14*60Hz*120V)=440nF. So use E12 value 560nF. R11 limits the maximum current though C2. Imagine you plug in just when the power is at it's highest voltage, namely 230V~*√2 = 325V top. Using Ohm's law R11 limits the current to 325V/100Ω=3,25 Amps, whichs is still high but should not destroy the capacitor and the diodes. During the negative cycle, D2 lets the charge in C2 pass and stores it into capacitor C3. This is a small buffer that temperarily stores the charge if more power is needed then is available. At the positive cycle, D1 is needed to charge the capacitor again with a negative charge (confused lol?). D3 is a zener diode that breacks down when the tension reaches 18 Volts. This is to protects IC1 which does not like any tensions over 20 Volts. From the input voltage (which is somewhere between -7 and -18 Volts depending on the total current used) IC1 stabilises the output voltage to -5 Volts. C4 prevents oscillations and is placed close to the microprocessor.

Zero voltage detector

The purpose of the zero voltage detector (sometimes named the zero crossing detector) is to signal to the microcontroller that the triacs need to be fired up again. Depending on the required intensity of the lights, this will happen immediately, or when the lowest intensity is wanted, never. The zero crossing detection circuit is build around R11, R12, D4 and D5 . R11 is choosen so high because P=U²/R=230²/1MΩ=53mW. Any lower value would increase the power wasted. We can use such a high value because the ATtiny2313 datasheet tells us that the maximum leak current I_ih is 1µAmps at 5 Volts. We can now calculate the worst R_i resistance. R_i=U/I=5 Volts/1 µAmps= 5 MΩ. The ATtiny generates an interrupt when the voltage rises above 0.6Vcc and when it drops below 0.3Vcc. Knowing that we can calculate R12. 30% of 5Volt is 1.5Volt. 60% of 5Volt=3Volt. At 1.5V and R12 is 820k I=1.5/820k=1.8µA. U=1.8µA*(1MΩ+820kΩ)=3.3V. Compared to the NULL voltage that is -5-3.3=-1.7 Volts. Another interrupt is generated at 3V/820kΩ=3,6µA =3.6*1.820MΩ=6.65Volts=+1.65 Volt compared to the NULL voltage. So with these values U_grid will cause an interrupt at roughly +1.65Volts and -1.7Volts, this is close enough for our purpose.

Infrared receiver

The infrared receiver is build around standard highly integrated components. TSOP1736 is a receiver that responds best to signals transmitted on a carrier frequency of 36 kHZ like RC5 etc. Some other remote controls use different carrier frequencies, in that case use the appropiate ir receiver module.

Serial interface

The last part of the dimmer is the circuit around OK1 and OK2. These make sure that the serial interface is galvanically isolated from the power grid so that you can savely hook up your dimmer to your computer as long as it has a serial port. Through the serial port we can send or receive commands to or from the dimmer. Don't expect speeds to be much higher then 9600 baud, but this is more then sufficient.

DMX 512 and X10

The microcontroller used in this dimmer is very powerful and is easily adapted to decode both DMX and X10 signals. The software for decoding the DMX512 signal is ready, but you must design your own hardware. For the X10 interface I still need to design some hardware and software. Getting the dimmer ready for decoding X10 signals will still take some time. Not that the software is the problem but the hardware is.

Safety

The two fuses used in the dimmer provide some safety against short circuits and malfunction of the dimmer. Although malfunction of one of the triacs could blow up the microcontroller. There are a few reasons I choose this configuration instead of using opto-diacs. The reasons are size, price and reliability. Nowadays electronics are hardly repeared anymore, but are instead just thrown away and more parts just means lower MTTF (Mean Time To Failure).

Software (firmware)