The digital power supply: a powerful concept when properly understood
In recent years, semiconductor suppliers and manufacturers of food vaunted concept of digital power .
More and more digital marketing of this technology as a panacea for the increasing complexity of existing
systems. Many users, however, have to wait and deal with the manufacturer of food suspected of being
involved in the promotion of different architectures. To date, the lack of prospects for the industrial
success with the help of numerical methods. This creates confusion in the minds of its designers.
Moreover, in addition to the confusion, manufacturers use different meanings behind the term "digital
power." What does this mean? And what are the immediate benefits?
Not to be confused management and control of power
The distinction between the government and management is a key point in a discussion of digital power
system. Ericsson uses the term "control" for internal monitoring of food, especially management, cycle by
cycle, energy flow in this food-address. This includes the functions of servitude and feedback loops
disadvantages. The plan will be implemented in real time (on a scale coincides with the frequency
switching power supply). These functions are possible with analog or digital techniques. This choice is
often transparent to the user. This means that the use of a diet with a digital control system can not
change on his part.
In contrast, refers to the "current management" in / or communication and control of one or more external
power supplies. This includes functions such as system configuration, power, control and supervision of
individual power and communication errors are detected. The power management features are not
executed in real time. They take place at a pace slower than the switching frequency of power supplies.
These functions now tend to combine analog and digital techniques. Thus, the output voltage is often
determined by resistors, while the sequencing of power lines usually requires special order for each plan.
Such as Ericsson, is the digital management of power to use these functions to make digital technology.
Also, instead of an organization with multiple links to and from the whole sequence of single power
supply and troubleshooting, reducing a structure similar to a data bus for communication complexity of
Power control: the techniques used
In Figure 1, left, provides an integrated circuit Pulse Width Modulation (PWM) in a primary traditional
analog. A resistive voltage divider takes only a fraction of the power supply output. This voltage is then
compared to a reference voltage amplifier with an error. The output of the latter is an analog signal whose
amplitude can be used to correct the output voltage proportional. This error signal is fed back to the PWM,
the output pulse has a width defined by said error signal is complete. This pulse then controls the
direction of the power amplifier semiconductor (typically MOSFETs). Drivers are required to effectively
transform the MOSFET, which have high gate capacitance. A resistor-capacitor network is generally
offset to reach an adequate compromise between speed and stability of the intervention cycle.
The power supply consists of two sections, which are the input filters and output. Without using inductors,
capacitors and resistors, have different functions. The filter protects against input voltage transients input
voltage, stores a certain amount of energy to meet load variations and ensure compliance with specific
questions about the voice re-injected. The output filter reduces the voltage ripple and noise output of the
specific needs, "and saves energy that contribute to circuits called dynamic load. It is important to note
that the filter input and output of the power components are basically the same, control is analog or
The typical structure of a digital control of power in Figure 1, right. The network of measuring the output
voltage is comparable to an analog system. An analog to digital converter (ADC), but replaces the error
amplifier of the analog system, transforming the voltage of a binary socket. It's good to know, in addition
to the output voltage, the value of other parameters, such as the analog output current and temperature
in the diet. ADC can be purchased separately for each parameter. However, it is much more efficient, you
can use a prefix multiplexer (MUX). To measure the switching between the analog inputs and drive one
after another, for the ANC.
The sampling rate (MUX and ADC) is constant, the ANC offers an environment for each sequence of
numbers separated by a fixed time interval. These values are sent to a microcontroller that is targeted.
An integrated program saves the algorithms that perform calculations on the output values of the ADC.
The results of these calculations are defined as the error signal, the pulse length for the control phase,
the optimal values for the delays between the outputs of the driver settings and the necessary
compensation for the cycle. To compensate for the external components for the analog loop is no longer
necessary. Set values for variables such as output voltage, output current and temperature limits are
stored in nonvolatile memory during manufacture or can be loaded into a data store at startup.
Compared with analog control, digital control is much more flexible in their ability to adapt to changing
conditions and the load line. Analog approaches are generally configured with a set of "compromise" for
a particular control. However, digital control systems offer the ability to control the settings under the
terms of the converter to adapt. For example, in a regulation point of load (POL)-type synchronous buck
a timeout ensures that the upper and lower MOSFET never simultaneously. Analog control with a fixed
delay time to set the hour of death, in extreme operating conditions (worst case). But in most cases, the
value of dead time is more than necessary, that the efficiency of the converter. In contrast, a digital
control loop to dynamically change the value of downtime, according to operating conditions and optimize
the performance of the cluster controller.
In addition, the feedback loop compensation necessary result of a compromise between stability and
dynamics of analog systems. On the contrary, it is possible with digital technology, or non-linear adaptive
loop, varying the compensation rate in terms of the transaction. This means that the power pole or
controller responds quickly when needed, and slower in other situations. This approach has other
advantages: it requires a small number of decoupling capacitors for the safety data of output voltage,
space saving and, finally, a digital control allows a batch operation - which is to "jump "to say the ignition
cycle in low load - without the inconvenience of the generally poor performance of dynamic load.
Digital management of power
Digital management of power is the control and supervision of the Board of Directors of the digital power
outside of their channels. An example of a sequence of commands or disable multiple streams, ensuring
that each remains in its notebook line of tension. This type of control is of paramount importance when it
comes to ensuring the proper functioning of components such as DSP and FPGA, various supplies
needed for their heart, their low power consumption and network I / O voltage levels more.
Current systems for digital management of power for energy supply in general, are on a circuit that is
mounted to communicate with a control circuit unit via a digital communication bus. The power modules
can be DC-DC regulators are insulated or isolated post. The central control circuit can take many forms: it
can be given a circuit of power management to a microcontroller or a group of gates in an FPGA. This
control is often called "master" or "host", while the modules are controlled classified as "slaves." In most
cases, the action of the refugee camp for a single card is limited. In some large systems, the host
controller communicates with the highest place in other parts of the system components or even remotely
via extended communication networks installed. Figure 2 shows an example of a network on a single
A practical example
The complexity of the power management increases significantly when the number of voltage level
increases on a map. The order of sequences, the rise time and the time to start or during normal
operation and shutdown are controlled under the conditions of errors. All this is easily done with a digital
management system without the need for analog components to control and delay. You might as well a
sequence of events, such as when it comes, for example, to control the parameters of a module before
the first corner on a second try.
The technical limits of tolerance of the voltage at the output voltage above or below the rated voltage,
ensuring compliance with the specifications defined in any load condition. And "to be used to produce in
order to test these limits in order to test the robustness of the whole. And "perhaps voltage fluctuations of
5% in different combinations. With the digital communication bus, is the work of less than 1 second
without any additional hardware or additional connection. Figure 3 shows examples of sequencing and
In fact, the digital management of power throughout the life cycle of a system is useful and nutrition.
During construction of each module, an automatic test equipment to set up food and storage of data and
serial numbers. During the optimization phase of the system, the digital interface allows designers to use
food to measure temperature, voltage and current on a laptop in order to define the thresholds of
protection circuits and optimize the sequencing power.
During assembly and testing of electronic boards and systems, test equipment capable of the
management interface of digital output voltage tolerances of the test, monitor and adjust the tension is to
use the measure efficiency and serial numbers and codes in the file. And if the designer is a host
controller on board permanent locations, can be easily implemented to start and stop sequences without
the need to add anything. It is easy to monitor the operating temperature to adjust the fan speed. The
effectiveness is tested in real-time damage is reported before a fault. Routine management and fault
detection can be developed taking into account the operating conditions.
Keep in mind that requires a digital setup does not necessarily reflect the presence of a host controller or
a digital bus in the final, or even in the production cycle. If the specifications are known and expected to
remain relatively fixed, the manufacturer of power supplies are simply programmed during production
without substantial changes. The client then uses the power of the same way as traditional analog
Finally, the benefits of CNC convincing enough to make Ericsson the preferred approach in the design of
a power supply and modules in particular. The company has implemented more modules to control the
digital product lines. Some control of digital integrated circuits power supplies are also for system power
management to be useful, most of the equipment needed to manage-and-drop "free" in the hands of the