I've often wondered this but never followed up.

Take a 12v supply (not car for argument sake) drop the voltage down to say 3 volts, obviously there's going to be a lot of heat dissipation, so you use PWM... But what's the logical steps of converting 12v > 3v..

12v > pwm, but that's going to give 0>12v>0v how do you get 3volts from 12volts using pwm? what's the process? put a linear voltage regulator coupled with a big capacitor? I don't need a circuit just what's involved to get 12v>3v using pwm... I can understand LED's because you could simply use a resistor to control the current, but what about a regulated 3v supply?

 

First, it only works when current is flowing out the 3 volt end. That's why the pwm switcher doesn't get a result of "12v...0...12v...0". That would be the result if zero current was flowing and there was no final capacitor. Add the capacitor and you'd just get 12...........

OK. Got it now? The final capacitor is always being emptied, at least a little bit, and the switching circuit is reloading it, and stopping, so fast that the ripple voltage is a minor problem.

 

You are basically describing a SMPS. Switching mode power supplies always have some ripple, it is part of the design concept. Like analog, SMPS also have some sort of feedback to measure and correct the output voltage.

 

It's not just PWM-ing to a capacitor; there is also an inductor in the loop.

Without the inductor, there would be a very high current flow to quickly charge the capacitor when the switch closed, likely burning up the switch.

The inductor resists any change in current flow. So on a basic level, PWM applied through an inductor to charge a capacitor is how a switching supply works. There is more to a practical SMPS design than that, but those are the basics.

 

I left out the inductor because the question seemed to be about the simplest part of the theory of operation. The inductor does slow down the start of the current increase when the circuit switches on, and the inductor makes current flow persist after the switcher turns off, but that seemed like a detail that wasn't necessary to answer the question.

 

hmm, ok so the capacitor is fed via an inductor, but how do you get a regulated voltage out? Do you /have/ to have a linear regulator or is there another way? Smps? From 12v i thought that was used with > 120v sources?

 

There needs to be feedback to the switching circuit.

There are voltage feedback PWM controllers, and current feedback PWM controllers.
Both have their own unique set of problems to overcome when designing.

Current feedback is generally much more responsive than voltage feedback.

One problem is during start-up; the output voltage is zero (or nearly so), and so the voltage across the inductor will be quite large. If the switch is simply controlled by a constant PWM, the target output voltage will be overshot by a considerable amount.

 

In simpler terms, the control section is all kinds of complicated but, in general, it checks the voltage on the output and adjusts the switcher for more or less current, trying to keep the voltage stable.

 

It's not just PWM-ing to a capacitor; there is also an inductor in the loop.

Without the inductor, there would be a very high current flow to quickly charge the capacitor when the switch closed, likely burning up the switch.

The inductor resists any change in current flow. So on a basic level, PWM applied through an inductor to charge a capacitor is how a switching supply works. There is more to a practical SMPS design than that, but those are the basics.

Also, it might seem like a good idea, just in your mind, to replace the inductor with a resistor. After all then, PWM is not wasting energy, the resistor is just limiting the charge current. Well, it turns out this is no more efficient than a linear regulator if you actually work it out, which kind of makes sense.

 

Resistors don't store energy. When current passes through a resistor, it dissipates power as heat.

Inductors store energy as the current through them increases, and releases the energy when the current flow tries to slow down. It's not a 100% efficient process, but far more efficient than a simple resistor.

 

alright so all those example circuits which use a 555 with a resistor connected to a led.... is no more efficient than just using say linear regulator????

 

It depends.

The 555 PWM-type circuits can actually be pretty efficient. Using PWM is generally more efficient than using linear regulation. However, in the case of LEDs, the peak current flow must be limited by some means.

 

The concept of the current limiting resistor seems to be hard for some folks. I have seen several designs where their was PWM but no current limiting resistor. Then they ask why the LED burned out.

Where PWM saves a lot of heat is in the solid state switching device. A transistor operating in the linear mode gets hot, very hot, while a properly designed switching transistor stays cool.