Digital power control is the new wave, promising improved efficiencies and reduced costs. But, even the most sophisticated engineer can have a sense of “unease” at needing to learn new skills and dealing with new tool sets, while facing ever tighter schedules with shrinking staffs. Some argue that they do not need something as complex as a computer to control their circuit; after all, how can a digital-control chip with millions of transistors ever compete on cost with a dedicated analog controller containing a few hundred transistors?
Is there a future for digital power? The answer is an unequivocal Yes!
A historical review of the development of computers and motor-control systems points the way to the future. Before the arrival of the digital computer, analog computers (1940 – 1960) composed of vacuum-tube amplifiers and switches implemented analog integrators, adders and multipliers to compute functions. The early digital computers filled an entire room with tens of thousands of vacuum tubes consuming tens of kilowatts, versus a tabletop analog computer that was a tiny fraction of the size and used a hundredth of the power; and yet the digital computer won the race. Why? Basically, because of two factors: repeatability and the ability to be programmed to handle ever increasingly complex tasks. There were many engineers who bet against the digital computer and lost.
In the 1960s and 1970s, the development of large-scale integrated circuits really swept away any lingering concerns about the reliability of digital computers. This opened new applications for digital computers, such as motor control. Prior to the 1960s, motor speed control was limited to brute-force methods, such as variable resistors in series with the motor’s power supply; or complex schemes, such as the Amplidyne that controlled the current through the control windings of motor/generator sets. The direct control of motors via vacuum amplifiers was never practical, due to the power losses associated with Class-A amplifiers. The development of high-speed, high-current transistors during the 1970s prompted the widespread use of Pulse-Width Modulated (PWM) methods for controlling motors. The initial PWM amplifiers were composed of discrete analog circuits that implemented velocity control and PWM signal generation.
As microprocessor performance increased and costs decreased throughout the 1970s, motor-control engineers looked at methods to reduce system cost. The position loop was first moved into the digital realm, and then the velocity loop was implemented in software. The result was the elimination of expensive motor tachometers. The next step was the elimination of the analog command voltage from the microprocessor to the amplifier. The microprocessor system directly generated the PWM signals used by the power transistors.
Because the first digital motor controllers were large and expensive, their initial applications were high-end motion control systems. Over the ensuing 15-20 years, the relentless grip of Moore’s Law had reduced the cost and improved the performance of microcontrollers so much that few engineers would even consider using a “dedicated analog motor controller,” instead of a low-cost microcontroller.
The analog motor-control PWM circuits were eventually implemented in integrated-circuit form, to become the basis for modern analog power-supply controllers. Today, there is a wide variety of analog power-supply controllers available from many reputable companies. These analog controllers offer small size and low cost, and engineers feel comfortable working with them.
So, why should power-supply engineers adopt digital-control methods? The reasons are many: reduced component count, improved power efficiency, predictable behavior, ease of manufacture, design, support and the ability to add new features.
Given the inevitability of digital power-supply control, how does one choose an appropriate controller?
There are two basic classes of digital power-supply controllers: dedicated digital pwm controllers, and Digital Signal Controller (DSC)-based products. Both types of digital controllers offer predictable behavior, programmable parameters for filter coefficients, and the ability to simplify the printed circuit board design by eliminating large numbers of passive components.
There are two subgroups of dedicated power-supply controllers: pure dedicated logic and microcontrollers combined with dedicated logic for implementing the control algorithms.
The pure dedicated digital controller offers the lowest-cost and lowest-power approach to implementing a digital control loop. Such devices typically offer some amount of non-volatile memory for the storage of filter coefficients, as well as voltage and current parameters.
The other approach, which uses a microcontroller with dedicated control-loop logic, sacrifices power consumption for improved adaptability. However, it can support more complex power-system monitoring tasks, and may support communications with external devices for remote monitoring and control.
Most of the dedicated digital PWM controllers are designed for a specific application. This is not surprising, since they are designed and sold by the same manufacturers that sell dedicated analog PWM controllers. Hence, both the analog and digital PWM controllers are susceptible to the same issue. If a power-supply designer wants to implement a new power-conversion topology, the designer will either have to adapt an existing controller to meet the needs of the new power-conversion circuit, or the designer forgoes the development of a new and potentially important design. It’s the old “Chicken and Egg” scenario. A dedicated power controller will only be built if the manufacturer thinks there is a large enough market to justify the expense of development. The result is that the bulk of power designs are very similar, making them easy for competitors to “clone,” thus reducing everyone’s profit margins.
The DSC approach replaces the dedicated PWM control logic circuitry, by combining a DSP engine with a microcontroller. The DSC method uses algorithms written in software, rather than dedicated circuitry, to implement control functions. However, the DSC requires more power than dedicated logic because it provides more generalized resources. The DSC has the advantage of being completely flexible and adaptable to new requirements and topologies yet to be developed. Some DSCs also provide security for the embedded software, protecting software intellectual property and making the process of power-supply design cloning more difficult.
Arguably, one of the most important features of the DSC power-supply controller is the complete freedom it gives engineers in creating new circuit designs; to meet specific challenges and requirements for leading-edge circuit designs. Originally, power transistors implemented the “power switches” in the traditional power topologies. Then, as power transistor costs dropped and the need for improved power efficiencies grew, synchronous rectifiers became common. Now, as power transistor cost continues its downward trend, additional transistors are added in power supply circuits to recover energy that would otherwise be dissipated in snubbers. Additionally, many designers are now adding more transistors to switch circuit components in or out to match current operating conditions. The number of PWM outputs required to control all of these transistors continues to grow, as designers find new uses for transistors. The PWM outputs are becoming more than simple PWMs; they should be viewed as programmable transistor control signals. With modern small-geometry semiconductor processes, the cost of adding more PWM outputs is negligible, compared with traditional analog PWM controllers.
DSCs free the power-supply design engineer from the shackles of traditional dedicated PWM controllers, but aren’t they complex to program?
The short answer is no. DSC suppliers have developed Graphical User Interface-based tools to ease the creation of control systems for standard topologies; and they provide numerous application notes and reference designs to further aid the design engineer.
Many power-supply design groups are adding engineers who are comfortable with programming microcontrollers, and know the tools associated with software development. The traditional power-supply design engineer still leads the project because of his/her knowledge of power-conversion techniques, and common issues gleaned from years of experience. The upfront development effort for DSC-based design may be higher than using traditional PWM controllers. However, many designers believe that once the product platform has been designed, the ease with which new product variants are developed is truly startling.
Digital control of power supplies is practical now—it is no longer some futuristic concept. Moore’s Law is constantly driving digital IC costs downward and performance upward. Any last vestiges of cost disparities between the old analog PWM controller and the new digital controller are vanishing. Just as most timing applications can be done less expensively with a microcontroller than the venerable analog 555 timer, the same is happening with digital power controllers versus analog controllers.
