Small microcontrollers add big functionality
Power-supply designers are often charged with conflicting
tasks: they are required to reduce size and cost while
simultaneously adding features and increasing output
power. Despite the ever-increasing complexity of
analogue power-supply controllers, their features are
limited when the supply is created. To this end, some
designers have turned to purely digital power supplies.
However, for many designers, this is a radical departure
into unfamiliar territory. A worthy compromise comes in
the form of a traditional analogue power supply with a
digital microcontroller as a front end.
The beauty of this design lies in the fact that the power supply is still
controlled using analogue techniques. This allows power-supply
designers to bring added functionality to existing designs without the
need to start from scratch with a digital design. The familiar error
amplifier, current sense and voltage sense are all still present. While
some design elements, such as the compensation network, are still
implemented using discrete components, others are now controlled
by the microcontroller.
The features added by a microcontroller can be grouped into four
categories: control, monitoring, deterministic functions and
communications. These categories will each be discussed in more
detail.
Powerful control
The first group of features falls into the control category and is
dependent on the hardware interface between the microcontroller
and the power supply. It is important to determine the points in the
analogue design when the microcontroller can be connected. Some
power-supply controllers generate control signals, such as voltage
references, internally. These controllers provide very few external
connection points for a microcontroller.
Other power-supply controllers, such as Microchip’s MCP1630, are
designed to be microcontroller friendly and provide a wealth of
connection points. For the scope of this article, it will be assumed
that the supply controller provides us with two control points; a
shutdown input, and the ability to set the reference voltage, as shown
in Figure 1. While these two connection points may seem
insignificant, they can provide very powerful control and complex
functionality.

Fig. 1: The supply controller provides a shutdown input and the ability to set the reference voltage.
Parameter and fault monitoring
Microcontrollers currently find their way into many power-supply
designs with a role in monitoring. Many microcontrollers have on-board
ADCs and analogue comparators. This makes them ideally
suited for monitoring signals, such as input voltage, input current,
output voltage, output current and temperature.
Monitoring this wide variety of signals enables the microcontroller
to perform functions, such as intelligent fault detection. The versatility
of the microcontroller lies in its ability to be programmed and customised
to suit the design’s needs. To this end, not all fault conditions need be
treated in the same way. Brief over-current situations and other non-critical
fault conditions may only set flags. An over-temperature
condition may require the supply to shut down until the fault
condition is clear. Faults that restart the supply can be tallied. Too
many faults in a given timeframe may permanently disable the power
supply.
The processing power of the microcontroller is also utilised through
its ability to acquire complex metrics, such as power. Determining
power in an analogue system requires an intricate analogue
calculation. For the microcontroller, however, this is no problem.
Parameters such as input power, output power, efficiency and power
losses can all be calculated.
Finally, the microcontroller’s monitoring abilities enable more
advanced tasks, such as failure prediction. By comparing current
operating conditions to previously collected data, power-supply
designers can determine conditions that lead to supply failure.
In these cases, the power supply’s
ability to predict its own failure
can provide for cost savings and
higher reliability.
Deterministic functions
Fault detection is not the only
reason for monitoring data. There
are many other actions that can
be taken based on this data.
These tasks fall into the category
of deterministic functions.
Deterministic functions allow
power-supply designers to add
flexibility, functionality and
protection to their designs.
Consider the basic tasks of soft
start or under-voltage lockout.
By allowing the microcontroller
to perform these tasks, the
lockout voltage and soft-start
ramp rates are programmable
and do not rely on analogue
components.
Tasks that are more complex
can also be executed. Consider
power-up sequencing: the power
supply can be programmed to
monitor another voltage, waiting
until the voltage rises fully before starting up. Perhaps two voltages
must rise proportionally or track one another. Any of these
possibilities can be implemented through changes in software,
with no hardware changes required.
Another possibility for a deterministic function may be scaling
the current limit based on temperature. This allows power-supply
designers to use component temperature derating parameters to
ensure reliable operation.
Deterministic functions also make it possible to compensate
components to increase their accuracy. Many datasheets show how
parameters vary with temperature. In these cases, the microcontroller
can be used to perform temperature compensation. This allows the
designer to use lower-cost components and compensate their
results based on temperature. Microchip application note AN1001
(DS01001) describes how a ±6°C temperature sensor can be
compensated and perform as a ±0.1°C temperature sensor.
As a final deterministic function, consider a self-calibrating
power supply, where known voltage is supplied at the output,
sensed through the voltage-feedback circuit and stored. Using this
method, any tolerance in voltage-feedback resistors is eliminated,
and lower-cost resistors can be used without sacrificing accuracy.
Furthermore, the hardware for a 5V and 3.3V power supply is the
same: only the calibration process changes.
These examples of deterministic functions are by no means an
exhaustive list, but show the power, and potential, of the
microcontroller. It is clear that a number of power-supply
parameters can be monitored and controlled via small, inexpensive
microcontrollers. However, the means for storing and retrieving this
information has not been covered. This is where power-supply
communications become important.
Power-supply communications
The real value here is that this communication can happen
remotely. This is especially important for remotely-located
telecommunication and server power supplies. This remote
monitoring also allows operators to increase the reliability of their
systems.
Additionally, remote communication may allow operators to adjust
their voltage and current limits based on anticipated loading. In the
same way, the use of redundant power supplies can increase reliability
and up-time. Should a power supply receive a signal indicating that
conditions for failure are present, the operator can be notified, disable
the failing power supply and activate the spare supply, or the process
can be automated and the failing power supply can activate its own
spare.
Power-supply communications are not just used for monitoring
and setting operating parameters. Many microcontrollers contain
on-board EEPROM that can be used to store data, such as
production information. Should a component failure arise, an
equipment operator can easily determine which supplies are
affected. It is also possible to store repair history. Doing so ensures
that the power supply’s production data, repair history and
operating information will always be present and up to date.
Given the extensive list of microcontroller possibilities, there is a
common misconception about the processor required to perform
these tasks. Designers may believe that only high-end microcontrollers
or digital signal processors can be used. All of the tasks described
in this Design Note can easily be performed with low-cost, 8-bit
microcontrollers. Again, the goal is not to replace the existing
analogue functionality, but merely to supplement it with the
flexibility and processing power that only a digital microcontroller
can provide.