[006] The Digital Power Supply Revolution

Digital power is all the rage now, but don't expect it to shorten your design time, or to eliminate the need for complete optimization of the analog parts of the power supply.

Introduction

At the recent Applied Power Electronics Conference, digital power supplies were featured everywhere, in papers, seminars, and poster sessions. Applications were widespread, in VRMs, power factor correction circuits, inverters, and dc-dc converters.

It’s easy to start feeling overwhelmed with all this information on digital applications. And, in reading the material, to feel like you are perhaps missing the boat with your simple analog solutions. In this article, I’ll examine some of the issues and misconceptions about digital control.

Digital Control is New?

Figure 1 shows the block diagram of a digital controller. At the input side, an a/d converter samples an analog waveform of the power supply, and converts it to a digital value. The analog waveform may be the output voltage, or some pre-processed waveform where it has already been compared to a reference.

Sampling is done carefully, in an effort to avoid switching noise from affecting the results, and this simple sampling process can be quite complex, especially if there are multiple converters operating together in a system.

Figure 1: Block diagram of Digital Controller

The resulting digital signal then enters the processor of the digital controller, which sends the gate drive to the system. The driver circuitry may or may not be included in the controller, and this was discussed in the December 2006 issue of this magazine.

Much of the focus in recent years has been concentrated on solving the issues of the resolution of the digital output pulse. Numerous researchers have come up with solutions, including delay gates, to provide resolution beyond the clock frequency of the digital controller. This is important to avoid numerical oscillation.

Now let’s look at the so-called analog controller—yesterday’s technology if you listen to all the digital papers. Figure 2 shows a typical controller. The output voltage of a converter is processed with an analog amplifier, and discrete analog parts are used to compensate the feedback. The output of the error amplifier then is compared to a ramp with a comparator. The ramp also forms the clock of the converter.

Figure 2: Block diagram of “Analog” Controller

The ramp-reference circuit is nothing more than a simple A-D converter, directly generating a digital waveform, and setting the width of the pulse at the output of the controller. If you look at the input and output waveforms of Figure 1 and Figure 2, one thing should be immediately obvious: the analog controller that we have used for over 20 years in this industry is, in fact, a digital controller!

It has all the features required—and ADC, digital processing circuitry, and digital output. It is also an incredibly elegant solution to the digital controller problem. The clock frequency does not need to be any higher than the desired switching frequency of the power supply, and yet the output digital pulse has infinite resolution. This is something the new digital controllers are still struggling to achieve.

It is very important to recognize that the controllers we’ve always used are digital controllers. It stops us from worrying about whether we are using the latest in technology—analog or digital—when in fact they are both digital with different implementations. And it moves us forward to consider what the real difference is between the old and new, and whether you need it.

The core difference is that the new controllers are programmed with software, whereas the old controllers are hard wired, and not as flexible. As we’ll see in this article, the capability to reprogram may not be as important as claimed for your converter.