Refresh! Digital Power

WHAT IS DIGITAL POWER?
In any discussion of digital power, a distinction must be made between two key concepts: power control and power management. Ericsson uses the term power control to address the control functions that are internal to a power supply, especially the cycle-by-cycle management of the energy flow within the unit. This definition includes the feedback loop and internal housekeeping functions. The power control function operates in real time in comparison to the switching frequency of the power supply. Control functions can be implemented using analog or digital techniques, with power supplies using either approach often appearing identical to the end user. That is, the use of digital power control may not require any changes or new design on the part of the end user.

By contrast, power management refers to communication and/or control outside of one or more power supplies. This includes functions such as power system configuration, control and monitoring of individual power supplies, and fault detection communication. The power management functions are not real time—they operate on a time scale that's slower than the power supply's switching frequency. Presently, these functions tend to combine analog and digital techniques. For instance, resistors often program the output voltage, while power sequencing typically requires dedicated control lines to each power supply. According to Ericsson's definition, digital power management implies that all of these functions employ digital techniques. Furthermore, rather than using multiple customized interconnections to each power supply for sequencing and fault monitoring, some type of data communications bus structure is used to minimize interconnection complexity.

POWER CONTROL
On the left-hand side of Figure 1, a pulse-width-modulation (PWM) IC provides the primary control in a classic analog control loop. A resistive voltage divider samples the power supply's output voltage, which an error amplifier compares with a DC reference voltage. The error amplifier's output is an analog signal with a magnitude that's proportional to the required output voltage correction. This error signal feeds back into the PWM chip, which produces an output pulse whose width is defined by the error signal. The PWM output pulse then controls the on time of the power-stage semiconductors, which are typically MOSFETs. Because MOSFETs have large input gate capacitances, driver circuits are necessary to turn them on and off efficiently. A fixed resistor-capacitor network typically compensates the loop to ensure the proper balance between dynamic response and stability.

The two other major sections of the power supply are the input and output filter networks. Built from inductors, capacitors and resistors, these sections provide several functions. The input filter helps protect the power supply from transients on the supply voltage, provides some energy storage for operation during dynamic load changes, and includes filter networks to allow the power supply to meet its input conducted emissions specifications. The output filter smoothes the output voltage to ensure that the power supply meets its ripple and noise specifications, and also stores energy to help service the dynamic current requirements of the load circuits. Importantly, the input and output filters and the power devices will remain essentially the same for either an analog or a digital control structure.

The structure of a typical digital power supply control system appears on the right-hand side of Figure 1. The output-voltage sensing arrangement is similar to that of an analog system. However, an analog-to-digital converter (ADC) replaces the analog system's error amplifier, converting the sense voltage to a binary number. In addition to output voltage, it's useful to know the value of other analog parameters such as output current and temperatures in the power supply. While separate ADCs could sense each parameter, it's typically more efficient to use a single ADC and precede it with a multiplexer. The multiplexer (MUX) will then switch between the analog inputs to be measured and route each one in turn to the ADC.

Since the sampling rate of the MUX and ADC is fixed, the ADC outputs a series of numbers for each parameter, each of which is separated by a known time period. These values supply a microcontroller that provides the processing for the system. Onboard program memory stores the microcontroller's control algorithms that perform a series of calculations on the ADC's output values. The results of these calculations are parameters such as the error signal, the desired pulse widths for the driver stage, optimal delay values for the various driver outputs, and the loop compensation parameters. The analog system's external loop compensation components are no longer necessary. Reference values of parameters such as output voltage, output current, and temperature limits are stored in non-volatile memory during manufacture, or can be downloaded into data memory at system start-up.

Compared with analog control, digital control is far more flexible in its ability to adapt to changes in line and load conditions. Generally, analog approaches are configured with only one "compromise" setting for a given control parameter, whereas digital control systems have the ability to change the control parameters as a function of the converter's operating conditions. For instance, in a synchronous point-of-load (POL), buck type regulator, a dead-time value ensures that the top and bottom MOSFETs never simultaneously conduct. Analog control systems use a fixed timing network to set this dead time for worst-case operating conditions. But under typical operating conditions, this dead time value is longer than necessary, which reduces converter efficiency. By contrast, a digital control loop can dynamically vary the dead time value as a function of operating conditions to optimize the POL regulator efficiency.

Also, feedback loop compensation in analog systems is necessarily a compromise between stability and dynamic response performance. Using digital control techniques, it is possible to construct non-linear or adaptive control loops that vary the compensation factor as a function of operating conditions. That is, the power supply or POL regulator responds quickly when it needs to, and more slowly in other situations. This approach has other benefits, too. Fewer output decoupling capacitors are necessary for a given voltage tolerance, saving cost and component space. Digital control can allow operation in discontinuous mode—that is, when the power supply "skips" switching cycles under very-low load conditions—without the usual disadvantage of poor dynamic load performance.

DIGITAL POWER MANAGEMENT
Digital power management refers to digitally controlling and monitoring a board's power supplies from outside of the internal workings, for instance, controlling the sequence in which multiple power supplies turn on and off, while ensuring that each unit meets its voltage margin specifications. These issues are crucially important to ensure the correct operation of devices such as DSPs and FPGAs that require multiple power supplies for their low-voltage cores and multiple-voltage-level I/O banks.

Today's digital power management systems typically assume a basic architecture consisting of board mounted power supplies (BMPS) that communicate with a central control device via a digital communications bus. The BMPS can be isolated DC/DC power modules or non-isolated point-of-load (POL) regulators. The central control device can also take many forms, such as a dedicated power management IC, a microcontroller, or spare gates in an FPGA. The central control element is often called the "master" or "host," while controlled BMPS are normally referred to as "slaves." For the great majority of systems, the host has a control domain that consists of a single system board. In some large-scale systems, this host will interact with higher-level controllers elsewhere in the system, or perhaps even with remote devices via long distance communications networks. Figure 2 shows an example of a single-board power system.

PRACTICAL EXAMPLES
As the number of voltage levels on a board increases, the system power management complexity increases considerably. For voltage sequencing, the sequencing order, ramp times, and delays must be controlled for normal start up and shut down operation as well as for fault conditions. All this is quite easy to accomplish with digital management without resorting to analog control and timing components. Event-driven sequencing can also be easily configured, for example, verifying the performance parameters of BMPS #1 before turning on BMPS #2.

Voltage margining is used during the final stages of production for "corner testing" to verify the robustness of the unit. Voltages are varied by perhaps ±5% in different combinations. Using the digital communications bus, this can be accomplished in less than a second without any additional hardware or interconnections. Figure 3 shows examples of sequencing and margining.

In fact, digital power management is useful throughout the power supply and system life cycle. During initial power supply manufacture, ATE (automatic test equipment) can configure parameters such as output voltage trimming, trip points for over current, over voltage, and over temperature, and load date codes and serial numbers. During power system optimization, designers can exploit the digital interface to the power supplies by connecting the bus to a laptop computer to measure temperature, voltages, and output currents, to set the trip points for fault protection circuits, and to optimize the power sequencing.

During assembly and testing of the board and system, ATE can use the digital power management interface for voltage margin testing, voltage monitoring and trimming, measuring conversion efficiency, and recording serial numbers and date codes. And if the designer places a permanent host controller on the board for use during normal operation, it's easy to implement sophisticated start-up and shutdown sequencing without extra components and interconnections. Operating temperatures are easy to monitor to regulate system fan speeds. Efficiency may be monitored in real time and degradation detected before failures occur. Fault detection and management routines can be developed that take into account conditions elsewhere in the system.

Importantly, the use of digital configuration does not necessarily require a host controller or digital bus in the user's end system, or even in the manufacturing process. If the configuration requirements are known and will remain relatively fixed, the power supply manufacturer can easily program them during manufacture without any hardware changes. The customer can then use the power supply in the same way as a conventional analog unit.

Per-Johan Wiberg is a Strategic Product Marketing Manager at Ericsson Power Modules