Concept of switching PSU

Hi guys, I want to understand the concepts behind switching power supplies and have done a bit of reading. I would like to show a schematic that I have drawn up and explain my thinking with the purpose of getting some general constructive feed back on my grasp of the concept.

As far as my understanding goes, energy is stored within the inductor and the voltage drop across the inductor depends on its charge state. If it was a simple RL circuit then the voltage drop would change as a function of time. The switching of the supply is what keeps the voltage drop across the inductor at equilibrium. The diode acts to keep the circuit when the switching transistor is off, so the diode needs to be very fast acting and a schotkky diode is chosen due to this speed requirement.

In my schematic, the switching is controlled with a comparator and a reference voltage. If the voltage across the load is below the reference voltage, then the supply is switched on. When the voltage drop exceeds the reference, the supply switches off and the load is powered only by the energy stored within the inductor. This switching keeps going back and forth as the load discharges the inductor. Switching speed in this instance is dependant on the load, but I understand well designed units would have a fixed switching rate with a dynamic duty cycles, and a microcontroller could be used here to manage this.

I have used a PNP BJT as a high side switch, whereas supplies that handle more current would likely use P channel mosfets.

Where are the input and output capacitors? Can't work without them.

It's more of a concept thing rather than a real circuit, but yes I appreciate a full PSU would require capacitors to store charge.

It can't work without the output capacitor in particular, since thats what defines the output voltage - an inductor has high impedance and can't define a voltage, only a current.

The load is powered from the capacitor part of the time, by the inductor part of the time (which also recharges the capacitor). The sizes of the capacitor and inductor are chosen carefully.

I see, during my reading so far I had seen that the inductor is calculated using some differentiation but not seen anything about the capacitors. It’s something I can focus on as I read some more, I appreciate your feedback.

The circuit topology affects which capacitor does the most work (handles a large ripple current due to the switching). The need for low ESR and high ripple current handling is why capacitor choice can be difficult - real capacitors are not ideal components.

In an ideal world a SMPS would switch at very high frequencies, meaning the inductors and capacitors can be small (less energy to store), but this places high demands on the switching elements. Everything is a compromise, basically.

You also need to consider the maximum amount of ripple voltage thats permissable across the load, and the minimim and maximum current the load can draw, as these determine the inductor and output capacitor value. The maximum current drawn also detemines the inductors maximum saturation current.

Thanks again to both of you, I definitely feel more comfortable with the concepts and how I would need to consider real world application and specifications when looking at component choice. Time for me to look more at the maths now the basic fundamentals are clear.

I'd start by considering the differential equations for L and C:

inductor, L: L dI = V dt capacitor, C: C dV = I dt

Which you can just treat as difference equations most of the time:

L delta-I = V delta-t C delta-V = I delta-t where deltas are changes in the value.

So a 100uF capacitor with 2A flowing into it for 1ms: 1e-4 x delta-V = 2 x 0.001 = 0.002 thus delta-V must be 20V

Clearly 1kHz switching rate would not be good with these values as the voltage ripple is huge.

With 10us for delta-t, you get 0.2V delta-V, ie 200mV.