DIGITAL EXCITATION TECHNOLOGY

INTRODUCTION
In today’s power generation systems, digital excitation technology has become the choice
for many excitation systems. Digital technology is being utilized in both rotary excitation
and static excitation systems. It is being used on emergency standby power systems and
on prime power systems ranging from small diesel generators to large hydro and steam
installations. Digital excitation systems offer more features than were economically feasible
with their analog counterpart. These features are typically integrated into one controller in
digital excitation systems, resulting in a potential improvement in performance over the
analog counterpart. This paper will describe how digital excitation systems work and will
discuss their advantages.

BACKGROUND
Excitation systems for synchronous machines evolved from manually controlled rheostats
providing the excitation to rotary dc exciters. Electromechanical voltage regulator replaced
the manually controlled rheostat and eventually solid state analog electronics replaced the
electromechanical voltage regulator. The advent of digital excitation is just the next logical
step in the evolution of excitation technology.

Digital excitation technology was first discussed in the mid 1970s. The early digital excita-
tion systems suffered from the limitations on computing power of embedded microproces-
sor and the resolution of existing analog to digital (A/D) converters. As technology ad-
vanced, more practical systems started to appear in the mid 1980s. Digital excitation was
first utilized on large utility generators where the increased cost of the basic excitation
system could be offset by the increased features that were incorporated. As technology
further advanced, digital excitation systems were used on smaller generation systems.
Today, digital excitation control is being used on generators as small as 100kW. The incor-
poration of additional controls like speed governing, prime mover controls, and system
protection along with communications will allow digital excitation to become more com-
mon on even smaller generators in the near future.

FUNDAMENTALS OF DIGITAL CONTROL

The design of the digital excitation system resembles the basic block diagram for an ana-
log excitation system. The main difference between the two is that the digital excitation
system replaces some of the analog blocks with a microprocessor. A block diagram of a
basic digital excitation system is shown in Figure 1.

1
 Figure 1: Digital Excitation System Block Diagram
As the block diagram shows, the digital excitation system is a closed loop feedback sys-
tem. The output voltage of the synchronous machine is fed back to the controller. The
controller consists of a microprocessor, an Analog to Digital converter, and some form of
Digital to Analog converter. The reference input is incorporated into the microprocessor
and is adjusted through the use of human interface. This interface can range from a so-
phisticated personal computer keyboard input to an elementary pushbutton or toggle
switch input. The summing point and digital controller are implemented via software in the
microprocessor. The digital output of the microprocessor is converted to analog and ampli-
fied to the appropriate voltage and current levels for use by the synchronous machine’s
field winding. The flux generated by the field winding, along with rotation, causes the
synchronous machine to produce an ac voltage at its output terminals. The generator’s
output voltage is then fed back into the A/D converter, closing the loop on the digital exci-
tation system.

ITEMS SPECIFIC TO DIGITAL EXCITATION CONTROL

The digital controller is essentially an embedded system that consists of a microprocessor
and associated peripheral interface circuitry. The microprocessor performs the data han-
dling and mathematical calculations required to perform the basic task of voltage regula-
tion and excitation control. This action is performed based on the sensed terminal voltage
fed back from the generator output.

To more fully understand the operation of the digital excitation system, we will look at each
of its building blocks in detail.

ANALOG TO DIGITAL CONVERSION

The generator output voltage is reduced to a level compatible with the A/D converter input.
This scaling circuitry may also process the voltage waveform by rectifying, filtering, RMS
conversion, etc. prior to providing it to the A/D converter. The A/D converter takes the
analog signal representative of the generator output voltage and converts it into a digital
“word” for use by the microprocessor. The A/D converter might have a number of signals
multiplexed (combined) into its input, such as: 3 phase line voltage, 3 phase line currents,

2
field voltage, field current, accessory voltage inputs, utility line voltage, etc.

The A/D converter shown in Figure 2 is a simple example of one used to sense the
generator’s analog output voltage. It converts the analog output voltage, after scaling, into
a digital binary representation that the controller can use in its regulation process.

Figure 2: A/D Converter Block Diagram

The A/D converter shown in Figure 2 is configured as a unipolar device. With +VREF
connected to 5 Vdc and - VREF connected to ground, the device will only convert positive
analog signals. This method of A/D conversion requires some processing of the signal to
convert it from an ac signal to dc. This signal may also be filtered to provide a relatively
smooth signal to the A/D converter. Figure 3 displays a typical twelve bit A/D converter’s
binary output. As the figure shows, a 12 bit unipolar A/D gives a hexadecimal output rang-
ing from 000 to FFF.

3
 Figure 3: 12 Bit A/D Conversion Table
The resolution of an A/D converter can be calculated by knowing what the maximum input
signal is and also by knowing the maximum range of its binary output. The A/D converter
and scaling circuit in Figure 2 has a maximum range of twelve bits and a maximum input
signal of 150V. Using the following formula, the resolution for the twelve bit A/D is calcu-
lated to be 0.037V/bit.

A/D Resolution = Maximum Input ÷ [2 (Number of bits) - 1]

= 150 ÷ (2(12) - 1)
= 150 ÷ 4095
= 0.037 Volts/bit

The result of the above equation indicates that, as the input changes by 0.037V, the output
of the A/D will change by one bit. If the input changed from 119.963 Volts to 120.000 Volts,
the output would change from 3275 to 3276 decimal. In digital control systems, this topic
of A/D resolution becomes quite important. In simple terms, the output of the digital con-
troller can only be as accurate as its input signals. Thus a system with a higher resolution
A/D converter will generate a more accurate and stable output. A typical digital excitation
system utilizes A/D converters with 10 to 16 bits of resolution.

Some systems utilize ac sampling of the incoming waveform. The signal is sampled at a
rate at least twice the highest frequency contained in that waveform to avoid a problem
know as aliasing. For 50 or 60 Hz systems, that rate could be as low as 100 or 120Hz
respectively. Typical systems have waveforms that contain harmonics of the fundamental
signal. These systems typically contain a low pass filter to limit the high frequency content
of the signal to avoid aliasing. Practical filters require sampling of the ac waveform at a
somewhat higher rate to avoid aliasing. An example of an ac sampled waveform for 12
samples per cycle is shown in Figure 4 .

4
 Figure 4: AC sampled waveform with 12 samples / cycle
An additional consideration is how the sampled data is to be processed. The samples of
the ac waveform may be used to determine its RMS value when an RMS responding volt-
age regulator is desired. Some systems utilize a discrete Fourier transform to separate the
fundamental component from its harmonics. Control can then be performed based on the
fundamental signal, ignoring the harmonics.

Another method of analog to digital conversion is through the use of a voltage to frequency
(V/F) converter. The ac signal is applied to the V/F converter, which provides a variable
frequency signal to the microprocessor. The microprocessor measures the frequency (or
period) of this signal and uses it to determine the magnitude of the incoming signal. This
method may be a lower cost solution than utilizing an A/D converter.

Regardless of the type of sampling and waveform processing, the purpose of the A/D or
V/F circuitry is to convert the incoming signals into a digital format for use by the micropro-
cessor.

REFERENCE INPUT
The digital controller uses a digital word stored in its random access memory (RAM) as the
reference value. This reference can be adjusted via local controls, remotely mounted raise/
lower switches or via a serial datalink to a personal computer. This reference can also be
modified by other portions of the program to facilitate voltage setpoint changes, voltage
reduction during underfrequency operation, voltage setpoint changes to perform reactive
droop or reactive differential compensation, etc. This reference can also be changed by
higher level supervisory controls like Discontinuous Excitation Controls, VAr or Power
Factor controllers, Limiters, Power System Stabilizers (PSS), or externally applied acces-
sory input voltage.

5
SUMMING POINT
The summing point is a mathematical operation within the microprocessor that simply adds
the various signals required to perform the desired operation. The addition process is
sensitive to the sign of the signal so it can also perform the subtraction function. The simple
controller shown in the block diagram sums the reference to the negative of the sensed
output voltage. The resulting “error signal” is directed to the digital controller routine.

DIGITAL CONTROLLER
This block performs the complex calculations required to implement the control algorithm
that has been designed into the microprocessor. The control algorithm is a series of instruc-
tions that provide the required output to control the generator in a desirable manner. The
control algorithm may implement a form of classical control methodology like Transient
Gain Reduction, Rate Feed Back or Proportional, Integral and Derivative (PID) control. More
complex control algorithms may also be performed like Adaptive control, Neural Networks,
Fuzzy Logic or a combination of these styles of control. The transfer function of a PID con-
troller is shown in Figure 5.

Figure 5: PID Controller Transfer Function, Gc(S)

The microprocessor’s implementation of this transfer function is via a series of instructions
in its program memory. These instructions are derived from the continuous time domain
equations for the PID controller as follows:

Gc(s) = U(s) / E(s) = KP + KI/s + Kds

This continuous-time equation is transformed into a discrete time equivalent using the z
transform. The integrator (1/s term) is represented via trapezoidal integration as:

1 / s ⇔ T (1 + z-1) / 2(1 - z-1) , where T is the sampling time of the controller.

6
The derivative term, s, is represented via first difference as:

s ⇔ (1 - z-1) / T

The resulting equation, after the following substitutions relating discrete PID settings to
analog PID settings
KPD = (KP - KI T / 2)
KID = KI T
KDD = KD / T
is given as:

Gc(z) = U(z) / E(z) = KPD + KID / (1 - z-1) + KDD(1 - z-1)

This equation can be simplified as:

Gc(z) = U(z) / E(z) = (a - bz-1 + cz-2 ) / (1-z-1)

Where a = KPD + KID + KDD ; b = KPD + 2KDD ; c = KDD

The difference equations relating the error e(n) to the controller output u(n) as found from
the transfer function above are:
x(n) = e(n) + x(n-1)
u(n) = ax(n) - bx(n-1) + cx(n-2)

where e(n) = Z -1{E(z)}; u(n) = Z -1{U(z)}; x(n) = state variable

Once in this form, the code for the microprocessor can be written. This code may be in
assembly language or a higher level language like C. Examples of a simple PID control
algorithm in MatLab language is shown below:

%%% Calculation of regulator output

err = Vref - Vs;
x3 = x2;
x2 = x1;
x1 = err + x1
Vr = x1*(Kpd+Kid+Kdd) - x2*(2*Kdd+Kpd) + x3*Kdd

The digital controller gives unprecedented adjustability to the user over the many control
parameters. This feature allows the user to closely match the voltage regulator to the
specific generator it is controlling, resulting in superior performance over a typical analog
voltage regulator.

7
D/A CONVERTER AND POWER STAGE
The D/A converter takes the digital data from the microprocessor and converts it into an
analog signal for eventual use by the exciter or main field. The D/A converter may be a
discrete device as shown in Figure 6 or may be a series of timing signals supplied to the
power stage as shown in Figures 7 and 8. The power stage takes the low level analog or
timing signals and amplifies them to the levels required by the generator system. These
power stages can be thyristor or transistor controlled, phase fired or pulse width modu-
lated, shunt, compound, auxiliary winding or PMG fed.

Figure 6: 10 Bit D/A Conversion Table

Figure 7: PWM Power Stage

8
 Figure 8: 3-Phase Semiconverter Power Bridge

COMPARISON OF ANALOG & DIGITAL EXCITATION SYSTEMS

In any discussion of digital excitation systems, a comparison between an analog system
and a digital system must be reviewed.

Analog controllers process the signals continuously and with infinite resolution. Digital
controllers process signals in discrete time intervals with a resolution determined by the A/
D conversion circuitry. Modern digital controllers use high performance microprocessors
and A/D converters so that the excitation system performs essentially similar to their ana-
log counterparts.

Component aging and temperature drift can affect the performance of analog circuitry.
Since analog excitation systems use analog circuitry throughout, this can be a significant
problem. A perfectly designed controller may start to exhibit undesired characteristics as
temperature varies or after a period of time. In digital controllers, once the signal is con-
verted into digital format, the microprocessor’s accuracy is not affected by age or tempera-
ture.

Analog controllers are hardwired solutions; this makes modifications or upgrades in the
design difficult. Digital controllers are programmable, thus making them easy to upgrade
and maintaining design investment. Newer controllers utilize “Flash” memory chips, that
can be reprogrammed through a serial communications port without physically changing
any parts.

Analog systems are typically limited to simple control techniques with limited adjustability.
Although a well-designed analog controller can provide good performance, they typically
are designed to allow adjustment of only a few of the control parameters. Therefore, the
transient response of the generation system may not be optimal. Digital controllers can

9
easily implement multiple stability “networks” that allow precise matching of the control
parameters to the specific generator. This precise matching can result in improved perfor-
mance over analog controllers with limited adjustability, as seen in the figure below.

Figure 9: Analog vs. Digital Performance Comparison

Load Application Transient Response
Digital controllers can also be used to implement more sophisticated techniques from
modern control theory, such as adaptive controllers neural networks and fuzzy logic con-
trollers. These control techniques may result in better performance due to their ease of
adjustment at time of commissioning, resulting in a controller that is closely matched to the
generator it is controlling.

Remote control of analog controllers requires special care to avoid introducing noise in the
electronics, sometimes requiring Motor Operated Potentiometers or Reference Adjusters.
This adds to system complexity and cost. In addition, if the excitation system has more
than one mode of control, multiple Reference Adjusters may be required. Digital controllers
inherently use contact inputs to raise or lower a control functions setpoint. It is easy to
provide for local and remote control of these setpoints for many modes of operation.

Communications is one of the biggest advantages of digital controllers. Remote communi-

cations ability is not typically available in analog excitation systems. Digital controllers
typically contain one or more serial interface ports. These controllers can be designed to
implement many communications protocols, to ease the interface with other automated
systems in the power plant.

10
Event Recording / Data Logging is becoming a very desirable feature. This function has
traditionally been implemented digitally by a separate device. Many digital controllers are
incorporating event recorders and data logs internally. This feature is very valuable when
debugging a system fault.

Self testing and system testing is another feature that has become popular in digital con-
trollers. This can be as simple as a watchdog timer to a sophisticated diagnostics routine.

Off line setup is difficult with analog controllers. Typically the generation system must be
run to adjust and calibrate the controller’s parameters. Digital systems can be adjusted
with generator de-energized or at least off-line, easing setup and reducing the risk of
stressing the system.

Security is a growing concern with excitation systems. Analog systems contain potentiom-
eters that can easily be misadjusted unless locked in a cabinet. Digital systems often
provide a password security, sometimes multilevel, to provide the commissioning engineer
with the desired security of settings.

Analog systems use many parts per function and, therefore, may be more expensive and
less reliable for sophisticated multifunction systems. Once the appropriate signals are
provided to the digital controller, they can implement many different styles of control func-
tions. Some of the more popular control functions are:

VAr and Power Factor control

Excitation Limiting
Power System Stabilizer (PSS)
Discontinuous Excitation Control (DEC), Transient Excitation Boosting (TEB)
Stator Current Limiting (SCL), Short Circuit Current Limiting

CONCLUSION
Digital excitation control systems have significant advantages over their analog counter-
parts. If the excitation system requirements are for simple, minimum function controls, then
analog may be a lower cost solution. For systems that require multiple functions or where
precise matching of the control system to the generator is needed, digital is likely to be the
more cost effective solution. The logical choice of controllers for simple single function
voltage regulators may still be analog. For complex multifunction excitation control sys-
tems, the logical choice is evolving toward digital.

11