Where does all this power come from?
by Darren Ashby

I recently had the chance to develop a motor braking system. During this development, I was a bit confused as to why we kept charging buss circuits. I just couldn't figure out where all the power was coming from. At first it didn't make sense to me, but when it did, I began to understand switching power supplies better. Banking on the fact that I'm not the only one that was a little mystified by these type of supplies, I thought I might try to explain them in a way that I wish they would have been explained to me. ⁱ

In the world of power, it often seems that you never have the exact voltage you want. A huge majority of products run off of 120 VAC out of a wall socket. Another huge majority runs off of batteries that are charged from those wall sockets, and another significant majority run off of batteries that you can buy by the case load at Wal-mart. (How many batteries did you by at Christmas?) The problem is that most IC's these days want 5, 3.3, or even 1.5 Volts DC. This is no where near 120 Volts, and definitely not AC! Enter the power supply.

AC rules!

Back when Edison and Tesla argued over what type of electrical power we should have, I'll bet they had no idea of the type of integration that would occur in the world of electricity over the next 100 years. ⁱⁱ One thing they did know about was the transformer. The basis of the transformer is AC current. I won't go into details as I think it is pretty easy to understand. Put AC into one side of the transformer, and depending on the ratio of windings, you get AC out the other side. So, 120 VAC into a 10 to 1 ratio transformer will get 12 VAC out (minus heat losses due to the resistance of the windings). For a long time this was the preferred method of changing power levels. I will call it the 'copper' method. ⁱⁱⁱ Some time ago, however, a different method of changing power levels started becoming popular. It is known as the switching supply. As implied by the name, it regulates power to a load by switching current (or voltage) on and off. For this writing we will focus on the current method. (Don't forget however that current and voltage are invariably linked as Ohm proved so well.) The secret to these supplies is the inductor,^iv and the secret to understanding an inductor for me is to think in terms of current. (If you would like to get inside my head on how I visualize the behaviors of passive components read this). In the same way a capacitor wants to keep voltage across it constant, an inductor wants to keep the current flowing through it constant as well.

DC is what we want

The most basic current switching supply I know of is the buck converter. (I tried to discover why it is called a buck converter,^v to no avail. If any of you out there know, please email me and I'll post it for the rest of us.) Switching power supplies are typically DC to DC conversions. Even those that have an AC input usually create a DC buss, using a rectifier circuit before implementing a switcher. A buck converter will knock a DC voltage from a higher level to a lower level. The following diagram shows the heart of a buck circuit.

Let me identify 4 main parts: the inductor, the switch, the diode and the load.

Flip the switch

Let us start with the load and work our way backward. To begin with, switching supplies like to have a load. Without a load, funny things can happen, but more on that later. What the load wants (in most cases) is a constant voltage. Now if I remember ohms law correctly, one can control the voltage across a resistor (i.e. load) by controlling the current through it. So let's consider the flow of current in this circuit. We will begin with the switch closed. With the switch closed, current will flow through the inductor into the load. The current will rise based on the time constant of the inductor and the impedance of the load. Since the current rises so does the voltage across the load. Assume now that we have a circuit, that is monitoring the voltage across the load, and as soon as it gets too high it opens the switch. Now what happens? (This is where I used to get lost.) First, remember this fact. Just like a capacitor resists a change in voltage, an inductor resists a change in current. When the switch opens, the inductor tries to keep the current flowing. Now if there is no where for it to go, you will see a large voltage develop across the inductor as the magnetic field collapses. In fact, at Time = 0 the value of this voltage is infinite or undefined, whichever suits you.^vi But that doesn't happen in this case. In this case the current flows into the load, and the reason it does so is the diode. Consider it this way, current wants to keep flowing out of the inductor and into the other side of the inductor. Without the diode there would not be a path for this current to follow. However with the diode, this current is pushed through the load. So now the switch is open, and current is still flowing into the load. This current starts out at the same level it was when the switch opened (an inductor wants to keep current the same, remember!) and it decays from there. As the current falls, so does the voltage. Of course we still have a circuit monitoring the voltage across the load, and as soon as it gets too low, it closes the switch again. Viola, the process starts all over.

Now there are two important things I noticed once the pieces fell into place. The first is this control circuit I just described can be implemented with a simple comparator and a little hysteresis. ^vii But that would lead to the frequency of the switcher being determined by the value of the inductor and the impedance of the load. (Not always a desirable trait, it screams instability ^viii problems in my mind. But it is very, very cheap.) The other thing I realized was that when you first turned it on, the circuit would want to slam the switch shut and keep it there for a long time while current built up in the circuit. Are you beginning to see why switchers need a load? Luckily, others much smarter than I, have dealt with these problems already. That is why you hear terms like 'soft start' and 'built in PWM' when you start studying switching supplies. Here is a link to some great links that deal with these various problems.

Some final thoughts

If you just want to get a supply made quick and easy, and your requirements are fairly normal. Try out National's web-bench. There you can design and even order samples online. However, if you have special needs and features, I think the Tiny15 that I used in my TSMC project would make a great controller for the heart of a switching control circuit. It has built in amplifiers to sense current, several ADCs and a PWM output, all in a cute little 8-pin package. (I know, only real nerdy engineers would think an IC could have cute packaging, but I have never denied my nerd-hood.) Oh, and I almost forgot, one of the great things about these type of supplies is the fact you can get by with relatively little copper and attain nearly 100% efficiency. The reason for this is that the decay rate of the current in the inductor depends on the size of it, but if you switch it faster, the average current and thus the average voltage is still maintained. So you can get by with much less copper, especially for larger current draws at low voltages. The efficiency is good because much less power is spent heating the copper in the small inductor than in an equivalent transformer design. But all this comes at a price, switchers are known for their high frequency noise that has disrupted many a sensitive analog design. But who cares about analog anymore, right? :)

----------------------------

ⁱ As I am by no means an expert in this field, I will refer the reader to more in depth links for the formulas and equations involved. I only hope with this article to create a basic knowledge for the engineer who is getting into switching supplies for the first time and feels a bit lost by it all. It is also my hope that the reader will develop an intuitive understanding for the operation of these supplies so he/she will know where to look when the design stops working the way it is supposed to.

ⁱⁱ One major factor in the decision to use AC power was the invention of the AC induction motor by Tesla, a proponent of AC power. Until that time DC motors were the only way to get rotational power. Edison's desire for DC made a lot of sense in that perspective. There was not a good way to transmit DC power without significant losses however. That is why Edison wanted to put the power units in everyone's home. I think it is a bit ironic that the world is swinging back to the idea of a power supply in the home again, and the fact is that DC powered devices are proliferating at an incredible rate. That is not to slight Tesla in the least. He was brilliant, albeit a bit strange. Here is some insight into Tesla's head.

ⁱⁱⁱ This is because you use a lot of copper compared to an equivalent switching supply.

^iv For this article we will focus on regulators that are based around inductors. There are also capacitor based switching supplies, but since I grasped those so much more easily, I couldn't come up with enough material for an article this time, sigh, maybe later.

^v The term 'boost converter' makes sense to me, as it boosts a voltage up. Since I couldn't make the connection between 'buck' and knocking the voltage down, I perused the net for a bleary eyed 2 hours, but no luck finding any history to the term.

^vi This always bothered me a little. Although I have never measured the infinite voltage on an inductor, I have had quite a bit of fun making interesting knick-knacks that happen to shock you when you pick them up. Simply the fact that it can get to infinity (or a singularity as Stephen Hawking calls a divide by zero) seems to insist that there is more here than meets the eye. Maybe that is why I think there is a kernel of truth to this free energy stuff.

^vii I know, a lot of these links go to my own archives, but isn't it human nature to think the way you explain things is always the easiest to understand? :)

^viii If you want some heavy reading with some very large words that describe how switching supplies have chaotic natures (like we didn't know that already) check out this link.

Product Engineering Archive