One of the applications of 555 timers is a class D amplifier. In its most simplistic form it can be built with a single 555 and the 200mA current capability is enough to drive a small speaker, making it a good replacement for a low power amplifier. But I wanted more; I wanted to use it to build an amplifier that had enough power to allow listening to music in a small room. Adding a high power stage to a classical 555 class D amplifier was too easy, so I decided to build my own high power 555.
I chose to build the amplifier around two 555s for a better quality. The first is an astable and will generate a short low pulse with a high frequency which should be at least 10 times higher than the maximum desired frequency. The second is a monostable, producing a pulse that should have the width of half the period of the first 555 which would give a 50% duty cycle at its output because of repetitive triggering from the astable.
Controlling the duty cycle then becomes simple and consists of changing the control voltage level by applying the audio signal through a capacitor. I chose to make things better and use of an operational amplifier which also provides feedback. IC3B adjusts the control voltage so that at no input signal the average output (filtered with T5 and C7) is half the supply (set with R8 & R9). IC3A provides a gain stage for the input signal.
IC2 can only provide a limited amount of power in a low impedance load such as an 8 ohm speaker. This is what determined me to build it out of its blocks and make the output push pull stage a high power one.
The home built 555
The datasheet of the 555 provides the internal block schematic so I tried to built it using the components I had available. There’s a LM319 for the comparators, a HEF4013 for the flip flop, an additional HEF4093 NAND gate used as transistor driver and a push pull stage consisting of an IRF520 and an IRF9520.
The schematic tries to replicate the internals of a 555 by using specific ICs. The MOS drivers are the only special part of the schematic that requires attention. They are made using Schmitt trigger NAND gates wired as Schmitt inverters. Each inverter follows an asymmetric filter which has the role of delaying the turn on of each transistor but not the turn off. This is required because the output stage is a puss pull configuration and if both the transistors are on at the same time the supplies will be shorted. As it takes more time to turn off a transistor then on, the filter has the role of delaying the turn on of a transistor in order to allow the other one to turn off. The resulting waveforms at the output of the two gates are shown below, without the transistors connected:
Notice how the top wave that is turning on the N transistor on the high level is never high while the lower signal is high. The same may be expressed for the lower signal: it is never low (turning on the P transistor) while the top is low.
The whole circuit
The 555 presented above was built in place of IC2 in the concept schematic. This way, I can directly drive a 8, 4 or even 2 ohm speaker directly from the output which can supply a higher current now. I’ve only built one channel for the proof of concept, but multiple channels may be built, all sharing the first 555 that generates the trigger pulses.
The output of the whole circuit with the speaker connected looks like this:
On the top is the trigger signal from the first 555 and on the bottom is the output signal from the high power, home built, 555. Due to the inductance of the speaker, there is some over and under shot.
How does it work?
The frequency used was 175 KHz because of the speed of the available comparators. Any higher and they started to miss the trigger pulses. Still, I was able to get a good frequency response and the quality seems reasonable, I actually heard amplifiers built with ICs that sounded worse. With this exact configuration it may be well used for a woofer or subwoofer configuration, albeit the rather low power. Increasing the frequency to maybe 300-500KHz will provide an accurate response throughout the audio bandwidth.
Check the video for more details:
In the first picture, I’ve tried using all the 4 gates of the 4093 as output buffer(no transistors) to drive the speaker. Quality was good, but power was not that much. I’ve then added 3 more 4093 ICs on top of the first (bottom left) to get a total of 16 gates driving the 8 ohm speaker. This little experiment proved that the circuit is working quite well.
After that, I’ve added the power transistors and built the driver