NEW SOLUTIONS FOR BRUSHLESS EXCITER RECTIFYING MODULES

REDUCE DOWN TIME DURING OVERHAUL

Authors:
John Demcko
Arizona Public Service

Tim Vachon
Arizona Public Service

Andy Stendin
Larkin Power Components

KEY WORDS AND PHRASES: brushless, diode,

upgrade, Trinistat, exciter balancing, WTA regulator
ABSTRACT: Brushless rotating exciters became
increasingly popular with major OEMs in the late
1960s and early 1970s, as new technologies
offered alternative solutions to the conventional
brush-type exciters. The new brushless exciter
designs eliminated the need for commutators,
brushes, and the main field collector rings and
promised lower maintenance costs associated with
the production of dc required for the field of the
generator. After years of operation, however,
thermal stress and mechanical fatigue can reduce
the reliability of the power semiconductors and
modules, leaving few alternatives but replacement.

Rich Schaefer
Basler Electric

George Ross, retired

Arizona Public Service

steel ring or diode wheel on the outboard side of

the rotating exciter. They are configured as a 3phase full-wave rectifier bridge. The output of the
rectifier diode wheel is connected via insulated bus
bar through the center of the shaft to the main field
of the generator. Depending upon the amount of dc
current required by the field of the generator, the ac
exciter may be quite large to handle the rotor field
requirements of 2000-5000 dc amps. See Figure 1.
The large field current requirement prompted the
need to have many power semiconductors
connected in parallel for each phase to handle the
rotor requirements.

Unfortunately, OEM replacement parts may not

take into consideration new technologies that
improve the mechanical integrity of the product
while making maintenance easier and less
expensive.
This paper discusses the upgrade of the rotating
exciter and excitation system of a 321 MVA
generator at Arizona Public Service Company. It
involved the replacement of the Trinistat WTA
voltage regulator, the diode rectifying modules of
the brushless exciter, and the improved balancing
techniques of the exciter that significantly reduced
the exciter balancing time and improved
serviceability of the excitation system.

I. HISTORY OF THE BRUSHLESS EXCITER

The brushless exciter was the answer to issues
related to the dc brush type rotating exciter. The
problem of commutator wear and replacement
issues regarding brush maintenance for the dc
exciter and generator main field all focus on the
need to improve the process of getting dc power
into the field of the generator. In the late 1960s,
power semiconductors became a mature product
and gained favor for use in rectification circuits for
converting ac to dc. As the trend became popular,
power rectifying diodes became common for use on
rotating exciters. The brushless exciter is an ac
generator with a stationary field winding and a
rotating 3-phase AC output winding. AC output is
rectified by diodes to convert ac to dc into a threephase full wave bridge. The diodes are located on a

Figure 1. Brushless Exciter

In addition to the power diodes, capacitors are
placed across each power semiconductor to
minimize switching transients during the time when
the diodes were commutating in each cycle. The
capacitors provide a filtering to reduce the
magnitude of the switching transients during diode
commutation. These devices are called snubber
capacitors. Figure 2 illustrates a typical application
where the three-phase output of the ac exciter is
rectified by power diodes for each phase. The
cutaway view shows multiple sets of power diodes
in each phase, along with the commutating
capacitor across each diode.

Figure 4. Typical Power Fuse, with expulsion pin

location

Figure 2. Brushless exciter schematic

capacitor failed, the fuse would open. Although the

fuse could be replaced, the failure of the capacitor
rendered the module useless, since the capacitor
was buried in an irrepairable epoxy base.

The age of the diodes, the combined effect of

centrifugal stresses imposed upon the devices due
to spinning at 3600 RPMs, as well as induced
transients from the stator eventually cause
increased rates of semiconductor failure.
The failure of one diode does not force an outage,
since additional power semiconductors are
arranged in parallel.

II. ARIZONA PUBLIC SERVICE CHOLLA

UNIT #3 BRUSHLESS EXCITER ISSUES
As with any system, age eventually takes its toll.
The Cholla #3 unit in Joseph City, Arizona, a 321
MVA generator is no exception. It had problems
with the brushless rotating exciter. The brushless
exciter diode wheel consists of 6 diode modules per
phase, with 2 hockey puck style diodes in each
module. Consequently, the complete 3-phase diode
wheel ring contains 18 modules and 36 hockey
puck diodes, each with its own fuse and snubber
capacitor. See Figures 2 and 3. A noted problem
was a power semiconductor failure causing a power
fuse to open, which was indicated by the fuse
expulsion pin. See Figure 4. Subsequently, the fuse
was replaced. Unfortunately, the root cause of the
problem, a failed diode module, remained. This
lead to the failure of subsequent modules as fewer
diodes carried the bulk of the current in that phase.

Figure 3. Brushless exciter diode module,

Cholla Unit #3
Additional problems arose with the snubber
capacitors used for filtering. Figure 5 shows a
series fuse with the snubber capacitor. When the

Figure 5. Typical rotating diode assembly

Break Away View with Capacitor
When the power semiconductors failed and a repair
was initiated during a scheduled outage, lead time
of the replacement diode was unacceptably long.
This jeopardized the scheduled startup date after
the outage.

III. DECISION TO KEEP THE ROTATING

EXCITER VS. UTILIZING A NEW STATIC
EXCITER
The rotating exciter had other issues, most notably
extremely time-intensive balancing. The original
manufacturers diode modules were often out of
tolerance due to past repairs and a fairly loose
weight tolerance. Balancing time was considerable
since numerous planes were involved to achieve
satisfactory performance in mitigating vibration. For
these reasons and others, there was interest in
reviewing the option of replacing the brushless
exciter for a new static exciter that would work
directly into the main generator field. A full static
exciter option brought additional issues. Brushes,
brush rigging, collector rings, and a means for air
cooling this array must be provided. Power for the
static exciter would need to be derived, and there
was concern whether the station house transformer
was of sufficient kVA to handle the additional static
exciter load. Bus work sizing and concerns about
harmonics were issues needing to be addressed. If

the power were taken from the generator output,

then considerable engineering would be required
for tapping into the isophase bus. The demolition
and rework would be extensive, and a location for
the power potential transformer (PPT) and the static
exciter control cabinet had to be selected. At
Cholla, space is a premium on the turbine deck
floor; hence, finding space for the lineup would be
challenging. Budget estimates were in excess of 2
million dollars to modify the rotor and install and
replace the brushless exciter with a new static
exciter system. This was far more than the project
could afford. Instead, the problem areas associated
with the brushless exciter system were evaluated
and a plan was created to resolve them.

IV. PROBLEMS WITH THE BRUSHLESS

EXCITER SYSTEM
Besides the diode module problems and balancing
issues, the existing voltage regulator had reliability
concerns. The Westinghouse Trinistat WTA voltage
regulator had provided reliable service for the
majority of 30 plus years of service. The past few
years of operation had proved the regulator was
reaching the end of its service life. The WTA was
located in a harsh environment, exposed to coal
dust, high temperature, humidity, and vibration.
Many of the components were no longer supported
by the OEM and, with the extreme heat, card
sockets became hard and brittle, causing them to
disintegrate. See Figure 6. Once the high cost of a
static exciter replacement was established, a plan
was generated to improve the health of the existing
brushless exciter and excitation system:

This included:
Design new Diode Module replacements for the
original OEMs that would be readily available
and cost attractive.
Replace the existing Trinistat for a new digital
voltage regulator equipped with a power system
stabilizer that is designed to work into the exciter
shunt field of the brushless exciter.
Recondition the brushless exciters alternator and
reevaluate the balancing problem to improve the
efficiencies of time.
Develop more efficient methods to balance the
entire diode wheel, alternator, and permanent
magnet generator (PMG).

When the costs were assembled, the dollars to

improve the brushless exciter concerns and replace
the voltage regulator was much more acceptable
than buying and installing the static exciter. The
brushless rotating exciter and voltage regulator
upgrade was estimated at $300,000 versus $2
million for a new static exciter installation.
Redesigning the diode modules for ease of
maintenance, cost, and improved mechanical
integrity injected a certain risk. Management
concurred with engineering that the risk was both
justified and acceptable.

Rectifier Module Improvements

The engineering support of a consultant and
manufacturer was the first step to evaluate the
design of the existing rectifying modules on the
diode wheel and examining the weaknesses of the
old technology. Upon completion of the evaluation,
new rectifier modules were redesigned using the
latest material, components and manufacturing
technologies, to improve reliability and serviceability
of the diode modules. Another important design
criterion was to reduce the procurement and
serviceability costs over the original devices.
The following comparison table describes the
improvements with the new rectifier modules:

Figure 6. Westinghouse Trinistat Voltage Regulator

Old technology diodes

New technology diodes

Base block is the snubber capacitor. Not replacable. Entire module wasted for failed capacitor.

Replaceable snubber capacitor(s) are

readily available.

Weak bus bars subject to breakage.

(see photo above)

Thicker bus bars with center rib for

reinforcement.

10-32 screws for stand-offs

-28 screws for stand-offs to increase strength

to reduce shear stress.

G11 isolation material in stand-offs

G10 isolation material in stand-offs (better specs

than G11) with higher tensile strength

Modules delivered with different weights.

Balancing required after installation.

Modules balanced from factory at 1 g; hence,

reducing balancing time

Modules cannot be made lighter.

Can be made same weight as original or lighter,

if required to reduce bearing fatigue.

Old-type wrapping material on long bus bar.

High voltage isolation material covered with

anti-abrasion weave on long bus bar.

Exciter Improvements
With the new diode modules installed into the
brushless exciter, advantages were significant,
most notably the time taken to remove the old
rectifying components and replace them with new
assemblies. The new assemblies were lighter and
more robust than the existing design.
With the old exciter diode assemblies, the
replacement time was long and involved due to
weight discrepancies from one old diode module to
another. If a portion of old rectifying module was
defective, spare parts were an issue, with very little

time available to disassemble and repair the

assemblies. Issues such as selected fuses, diodes,
capacitors, stand offs and heat sinks would add to
the problems of an outage when these parts were
found to be defective. The capacitors in the base of
the assemblies were not repairable, rendering them
useless. Replacement of the selected diodes and
fuses often resulted in damage to the stand offs
during disassembly. Foreign object damage to the
phenolic stand offs and fatigue cracks in the diode
heat sinks were not uncommon. See Figure 7 for
diode stress cracks. The time and cost to repair
these assemblies made replacement an attractive
alternative.

Figure 7. Fatigue crack on 1200 Amp diode

Exciter Reconditioning
Another portion of the project was to recondition the
brushless exciter. The alternators rotating 3-phase
winding was rebanded, bearings were replaced,
and a power factor test was performed on the
winding insulation to assure a sound ac power
source when brought back to operation.
Voltage Regulator Replacement
The Trinistat voltage regulator problem also
was
addressed. A new digital voltage regulator was
installed. It was a redundant configuration equipped
with excitation limiters to safeguard the
synchronous machine. The excitation limiters
included volts/Hertz, Off-line and On-Line maximum
excitation limiter, under excitation limiter that could
be closely coordinated with the generator capability.
As standby and for use as a commissioning tool, a
field current regulator known as manual control was
provided. Manual control included automatic
tracking of the automatic voltage regulator, such
that, should a transfer to manual control be
necessary, it would be bumpless. A new feature,
voltage soft start was provided that eliminated the
previous requirement to start up in manual control.
The voltage soft start provided a means to control
the rate of generator voltage buildup and eliminated
concern of voltage overshoot as the machine built
voltage. See Fig. 8.

Other state of the art features included

oscillography to capture data of primary parameters
such as field voltage, field current, and generator
voltage, during events to help evaluate
performance and to be used as a diagnostic tool to
focus on resolving problems as they may occur in
the system. Sequence of Event with time stamp
information was also included to report any activity
of the excitation system. Maximum reliability was a
concern; hence, redundant digital controllers were
included in the design with automatic transfer to the
standby controller in the event of a failure of the
primary controller. The redundant controller would
include all the features of the primary unit with
identical settings. The system also would be
equipped with the latest state of the art power
system stabilizer using the Integral of Accellerating
Power to optimize system damping after a fault,
compliant with the WECC (Western Electric
Coordinating Council) that would improve not only
the inter-area stability mode but also local mode
stability. The new excitation system additionally
included redundant 100% rated power rectifier
bridges with automatic transfer to the standby
bridge in the event of primary failure. Two firing
circuits were provided, one driven from the 420
Hertz PMG as the primary power feed to the 6 SCR
rectifier bridge and, in case of an emergency where
the PMG would failed, a backup 60 Hertz station
power supply with appropriate 60 Hertz firing circuit
was provided to allow continued use of the
machine. The rating of the new power rectifier
bridges exceeded that of the Trinistat bridges they
replaced by 100%. Consequently, normal full load
current required from the power rectifier is
approximately 25% of their rating.
Figure 9 highlights the installation of the new
cabinet.

Figure 9. New Digital Voltage Regulator System

To avoid the environmental issues of the previous
system, the new voltage regulator would be located
in an air-conditioned room under the control room.
New cabling for excitation power, field, and control
was run to this location.

Figure 8. Voltage soft start feature

A high power, helical steel resistor with a ceramic

core was connected in series with the exciter shunt
field. It replaced the resistor provided with the
original equipment and was an important element of
the original design. The series resistor is used to
improve the stability of the OEM power rectifier and
lower the alternators field current time constant.
The original resistor was mounted on the frame of
the alternator. It was subject to high vibration
failure. To eliminate this problem, the resistor was
replaced for a new all steel ribbon resistor and
fixed-mounted below the turbine deck and well
away from the machine.

Exciter Balancing
Field experience played a key role in addressing
and minimizing the vibration issues of previous
outages. These included the following:

Low speed balance of the assembly.

Modified sling check to align overhung exciter

to generator coupling #7 bearing loading

Modified PMG alignment.

Replacement of the diode assemblies required
balancing of the exciter rotor. Historically, the
exciter field balancing issue was a time consuming
process that Arizona Public Service wanted to
avoid. The preferred method is a high-speed
balance of the rotor assembly at a facility to reduce
or eliminate the need for field balancing. The unit
minor overhaul time constraints and geographic
distance to a high-speed balance machine resulted
in slow speed balancing the assembly at a local
facility. The exciter utilizes seven balance
correction planes, fully accessible in the field. The
field exciter rotor trim balance history was used to
determine which of the seven planes were most
responsive. The seven planes were reduced to 4
planes for slow speed balancing. The diode wheel,
exciter rotor, and exciter cooling fan were selected
as the most responsive planes. Two of the four
planes were then selected depending on the
unbalance
condition
measured. The
static
component on the diode wheel and exciter rotor
was corrected, while any couple unbalance was
corrected on the diode wheel and exciter cooling
fan.
The match weighing of the new diode assemblies
allowed for no variation in diode replacement. The
old assemblies were out of tolerance due to past
repairs and part replacement that did not take the
time to properly match weigh the assemblies.

Figure 10. Match weights of 18 diode module

reduced the balancing time of the exciter
The sling check historically used the bearing end of
the exciter supported in a sling while the generator
end coupling bolts were selectively torqued to
control run out on dial indicators on the exciter
bearing journal. The time consuming process and
inaccuracies were reduced by adding a linerbearing support similar to those used on soft
bearing balance machines. The support, with the
addition of a proximeter probe to measure
displacement, allowed a more accurate correction
of rotor run out at the coupling.
The exciter journal bearing alignment was modified
to compensate for the overhung exciter rotor mass
relative to the generator. The bearing loading
prevented a lightly loaded bearing contributing to
the vibration of the rotor.
The PMG alignment utilized a truth diameter
machined on the PMG housing and a temporary
shaft on the exciter rotor to mount a laser alignment
tool. The laser alignment allowed real time
measurement of correction shims added to the
PMG, significantly reducing alignment time. The
tachometer shaft was eliminated and replaced with
a temporary shaft used only during assembly. The
tachometer signal was replaced, utilizing the output
of the PMG signal for speed signal. The tachometer
shaft also eliminated the extension shaft that had
caused vibration issues in the past.

V. COMMISSIONING
Cholla #3 startup after a major retrofit represented
a significant commitment of both resources and
manpower. New diode modules, improved
balancing methods, and a new voltage regulator
system all must be combined to provide a new
reliable system expected to reduce the mean time
between failures. The rework of the exciter and
voltage regulator system provided all the desired
benefits that had been anticipated. The new design
of the diode module was lighter, which reduced
bearing stress and helped reduce exciter balance
time by two weeks. Now, if service is required for a
diode module, they are easily maintainable with
100% replaceable parts and without concern of
weight balancing issues. The startup of Cholla #3
was smooth and without incident. On April 20,
2003, generator voltage built up and the machine
was placed back on line. The voltage regulator has
been in service for two years without incidents.
During commissioning, numerous tests were
performed to validate the new excitation system.
These tests included:
Open circuit voltage step changes
Excitation Limiter performance verification
Power System Stabilizer tuning and testing
Figure 11 highlights test results of open circuit
voltage step test to verify gains established for the
digital controller. Here, 2% voltage step change
was generated, which resulted in a slightly
underdamped voltage response with less than 1%
voltage overshoot. Voltage recovery occurred in
less than a second, illustrating the desired gains for
the digital controller.

the setup time for the new PSS has been reduced
to just one hour.
The modern digital excitation system utilizes
operating software that provides a fast and easy
means to set up and commission the voltage
regulator.

Figure 12. (X axis in Seconds)

Left: 1% Voltage Step Change, PSS Disabled
Right: 1% Voltage Step Change, PSS Enabled
Figure 12 illustrates performance with the new
excitation during two conditions. The figure on the
left highlights performance with a 1% voltage step
change with the power system stabilizer disabled;
the figure on the right, with the power system
stabilizer enabled. Notice the MW swings are
quickly dampened in the right figure when the
power system stabilizer has been enabled. Note
than an oscillography function built into the digital
regulator is used with a viewing program to analyze
information.

VI. CONCLUSION

Figure 11. 2% Voltage Step Test Off-Line

The commissioning time to perform open circuit test
has been reduced to one hour. The new power
system stabilizer utilizes the Integral of Accelerating
Power, where the original equipment hardware
implemented a analog transducer frequency input
based design. It was determined that, with the new
implementation, a higher gain could be realized to
improve the system damping. The previous tuning
time of the power system stabilizer was 4-6 hours;

All of the goals established for the Cholla #3 were

met. The best decision proved to be to keep the
brushless rotating exciter and simply address the
issues surrounding it. The money saved by keeping
the rotating exciter was substantial compared to the
costs of retrofitting to a full static exciter while the
reliability of the machine with the new parts has
increased substantially. Where unscheduled
outages due to a failure of an equipment device can
cause severe cost penalties, utilities are challenged
to provide the most economical means to achieve
the greatest reliability of the machine and system.

VII. REFERENCES
[1] IEEE Std. 421.2-1990, IEEE Guide for
Identification, Testing, and Evaluation of the
Dynamic Performance of Excitation Control
Systems.
[2] IEEE Std. 421.4 1990, IEEE Guide for
Specification for Excitation Systems
[3] IEEE Std. 421.5-1992, IEEE Recommended
Practice for Excitation System Models for
Power System Stability Studies.
[4] Tuning a PID controller for a digital excitation
control system, IEEE Trans. On Industry
Applications, Vol. 41, No. 42, March/April,
2005, K. Kim and R. C. Schaefer
[5] Cholla Unit 3 Brushless Excitation System
Upgrade, J. Demcko, T. Vachon, G. Ross, G.
Jones, A. Stendin, and P. Larkin. Presented at
EPRI Generator Maintenance Seminar, June
2003, Romeoville, Illinois

If you have any questions or need

additional information, please contact
Basler Electric Company.
Our web site is located at:
http://www.basler.com
e-mail: info@basler.com

Route 143, Box 269, Highland, Illinois U.S.A. 62249

Tel +1 618.654.2341 Fax +1 618.654.2351

e-mail: info@basler.com
P.A.E. Les Pins, 67319 Wasselonne Cedex FRANCE
Tel +33 3.88.87.1010 Fax +33 3.88.87.0808

e-mail: beifrance@basler.com

1300 North Zhongshan Road,

Wujiang Economic Development Zone
Suzhou, Jiangsu Province, PRC 215200
Tel +86(0)512 6346 1730
Fax +86(0)512 6346 1760
e-mail: beichina@basler.com