Two Circuits of Switched Mode Power Supply Based on the 555 Timer

Today the switched mode power supplies (SMPS) are widely used in the electronic devices. Compared with the linear ones, they have the advantages to be more efficient, to occupy smaller place and to be cheaper. The only their disadvantage is that they are noisy and in some cases require nice filtering of the output voltage. There are different types of SMPS : step-up (boost), step-down (buck) , inverting, flyback...etc. Here and here cam be found a brief explanation of them - their structure and how they work. The modern SMPS supplies typically contain one dedicated chip. Such chips are produced practically by each semiconductor producer. Some of them are available on the free market, some of them are ASIC's.

Creating this instructable, my main purpose was to develop such circuit of SMPS, which is based on the widely spread, easy to find and available in the parts box of each electronics freak chips. The result of my work are two circuits of boost SMPS - the first one contains one 555 timer and one comparator (opamp - used as comparator) chips; the second one contains two timer 555 chips (can be used the dual timer 556 also). Now I intend to present the circuits and to explain how they work..

On the picture you can see the schematic of the first variant. Before soldering or breadboarding I wanted to be sure that the circuit will work. It is nice practice before bringing any electrical circuit to life, to simulate preliminary its performance by circuit simulator. I did so. I have used LTSpice. It is nice, friendly, FREE and easy to use. It can be downloaded here. Using the simulator, you can prove the whole concept, observe the important signals, tune fine the circuit according your requirements and test it virtually at different conditions - different loads, different ambient temperature...etc. As attachment you can find both circuits of the SMPS, which you can play with - simply install the simulator (if you do not have one installed), unzip the *.ZIP file in one folder and and click twice on the circuit, which you want to simulate. It will opened automatically by the simulator. Then press the "Simulate" button and the simulation will start. Clicking with the pointer over different schematic nodes, you can observe the corresponding signals.

Short explanation of the circuit work: This SMPS works in so called "SKIP" mode. A free running oscillator (555) with enable controls the the gate of the switching MOS transistor. The frequency in this solution is ~ 200 KHz. It can be higher, but the timer has a frequency limit of ~ 300 KHz. MOS versions of the timer chip could work at higher frequencies, but I did not try. Some time ago I was able to bring a BJT timer to oscillate at 950 KHZ, but it was the absolute limit, and choosing between the products of few chip makers, only one was able to do it (I think that this was Texas Instruments chip). A comparator senses part of the output voltage and compares it with reference voltage. When the output voltage reaches the desired value (its part becomes equal with the reference voltage), the voltage comparator changes the state and blocks the oscillator through his RESET pin. In this moment the load current is delivered only by the charged output capacitor, which starts to discharge and its voltage decreases. The voltage comparator has very small hysteresis and when its lower threshold is achieved, it triggers again and enables the oscillator, what as result increases the voltage over the output capacitor. This process is repeated in a never-ending cycle, in which the output voltage is controlled with small oscillations around the needed DC value. If required these output voltage oscillations can be filtered additionally. On the picture I have marked the most important sub-blocks of the SMPS. The hysteresis of the comparator is implemented by a weak positive feedback (resistors R2, R8). When using a resistor divider for a voltage feedback, the comparator hysteresis is multiplied by the ratio of the resistors and appears at the output as amplitude of the voltage oscillations. Another alternative solution is using a level shifter (also presented on the picture, but connected in the 3-rd picture) In this case - no multiplication of the hysteresis exists. R5 is the load resistor.

On the second picture can be seen the most important signals graphically presented: On the top plot pane: the output voltage, on the second pane the input voltages of the comparator, on the bottom pane - the signal at the gate of the output switch.

The second circuit contains two 555 timer chips (or one single 556). Its principle of work is more close to that boost DC/DC converter described in the links given before. Here one of the timers is used to generate a triangle (sawtooth) wave. The second one performs more complicated function - it acts like comparator with latch and PWM controller. As input signals for this comparator act the triangle wave created by the first timer and part of the output voltage taken from resistive voltage divider. To be able to sense properly the levels of these both input voltages the internal reference voltage of this chip, normally created by internal resistor divider, is manipulated externally by the use of two forward connected diodes (it can be used zenner diode; I have 2V in my part collection, but I was not able to find such in the default libraries of LTSpice - that is the reason I used 2 diodes : D1,D3. Using zenner diode would be better solution). To create a linear triangle wave, I have used constant current generator based on the BJT Q1 and Q2. Other solutions are also possible. It can be seen also that the first timer has a simple POR (power on reset) circuit (R8,C3). This solves some problems with the output voltage overshoots. On the next pictures can be seen how this circuit works. On the top pane are presented the triangle wave signal (the green curve) and the signal at the gate of the switching transistor. On the bottom pane is the output voltage. It can be seen how with the increase of the output voltage the duty cycle of the signal controlling the output switch changes. Finally the output voltage reaches the target value (at time "10 ms") and the duty cycle does not change. It can be seen that here the output voltage oscillations practically does not exist - the output voltage has less noise.

To prove the theory and the simulations in practice I decided to make one of the circuits on a breadboard. I have chose the first one. How its implementation looks like, you can see on the first picture. Because I did not have all the devices, drawn in the schematic and taken from the Linear technology libraries, I didsome changes:

Instead LT1716 I have used LM358 as comparator. Because it is a dual opamp, to prevent some undesired oscillations I used both opamps available. One of the opamps was use as buffer of the feedback voltage, the other one as voltage comparator (see the second picture). Instead IRF2907Z I used IRF820. The Schottky diode used - BAT43 or something similar. As voltage reference I have used two 2V zenner diodes in series. R4 was 5.1 K.The resistor divider R6,R7 I replaced with 20K potentiometer. All other parts are more or less the same as in the schematic. In the next step I will present the implementation of the different schematic sub-blocks.

The most interesting in the oscillator - I have used 3 capacitors in parallel (2x470pF+220pF) to reach a value ~1nF (which I did not have :-) )

As described before - two 2V zenner diodes connected in series were used as voltage reference. LM358 replaced the voltage comparator.

As inductance I used a RENCO RL-5472-4 device. I replaced the feedback voltage divider with potentiometer. This made possible to regulate the output voltage.

This short movie shows how the circuit behaves...

As conclusion:

1. These both circuits work in boost mode, but with small modifications, they can be transformed as buck or flyback DC/DC converters. This I leave on you :-).

2. I strongly recommend, that before implementing some circuit physically, it should be simulated virtually. This can protect you of making some errors, could reduce you experiences for parts and PCB's and so on...Normally, If your models are correct - in the practice you will have results very close to those, you received by simulations.

In this context : relations between theory and practice, I would like to ask you for your support. I have started a project at kickstarter, where we want to prove another my theoretical work and you can receive a valuable object, which could have a very high historical significance. Please visit this site and support the project.