Stator Insulation Quality Assurance Testing for

Appliance Motors with Aluminum Windings

Daewoong Choi*, Tae-June Kang**, Sang Bin Lee**, Jaegyu Kim*, Jihoon Kim*
* **
Compressor and Motor Business Team, Department of Electrical Engineering,
Digital Appliance Business, Korea University, Seoul, Korea
Samsung Electronics, Suwon, Korea Email: sangbinlee@korea.ac.kr

Abstract—A number of appliance manufacturers have
recently replaced the motor stator winding material from copper
(Cu) to aluminum (Al) to reduce the product cost in the
competitive global appliance market. Although reduction in the
motor cost could be achieved, a noticeable increase in the failure
rate due to stator insulation breakdown was observed. In this
paper, increased failures in Al winding stator insulation is
investigated for 230 V, 50 W single phase induction fan motors
used in the outdoor unit of air conditioners (A/C). Since the
motors with defects or failures could not be screened out with the
quality assurance tests (QA) that were applied to Cu winding
motors, new QA test methods are evaluated for screening out the
units with high risk of failure. The investigation shows that
partial discharge (PD) measurement under surge excitation can
serve as an effective means of screening out the units that are
likely to fail. The claims made in the paper are verified through
comparative experimental surge PD testing on Al and Cu
winding motors before and after accelerated degradation.
I. INTRODUCTION
The appliance industry is facing fierce competition in the
global market, and the manufacturers are putting a lot of effort
on reducing or replacing parts to maintain the cost competitive-
ness. There recently have been cases where the reliability of
the system has been compromised due to excessive cost
reduction efforts. The electric motor used in appliances is a
critical component that directly influences the cost, reliability,
and performance of the overall system. Although the cost of
most appliance motors is low, the impact of failure can be very
harmful for the manufacturer. Failure of the system due to a
defective motor results in a long lasting negative impact on the
manufacturer’s reputation, and can also expose the customer
directly to safety risks. Therefore, implementing a reliable
quality assurance (QA) program for mass-produced motors is
an important requirement for appliance manufacturers [1]-[2].
Replacement of the stator winding copper (Cu) material
with aluminum (Al) in fractional horsepower appliance motors
is one of the options for cost reduction that has recently been
attempted. The cost of the motor can be reduced since the cost
of Al is 9 to 10 times lower per unit volume. However, a
significant increase in the failure rate of Al winding motors was
observed, where most of the failures were related to the stator
winding insulation breakdown, as shown in Fig. 1. The
rigorous QA testing performed during the design stage and
after manufacturing that successfully screened out defective Cu
winding motors was not effective for Al winding motors. This
implies that there is a stator winding insulation failure
mechanism in Al winding motors that is not present or
dominant in Cu winding motors. It is important to understand
the failure mechanism that causes the high failure rate in the Al
winding motors to find potential solutions. Once the cause of
failure is known, design modifications can be made or test
methods capable of screening out the defective units can be
added to the QA program.
The objective of this paper is to investigate the possible
root causes of failure for Al winding motors and suggest an
effective QA test method that can be used for screening out the
defective units prone to failure. An actual case of 230 V, 50
W, single phase induction fan motors with Al windings used in
the outdoor units of air conditioners (A/C) is studied. Since
conventional QA tests were not effective for Al winding motors,
new QA test methods are investigated for screening out the
units with high risk of failure. It is shown that partial discharge
(PD) measurement under surge excitation can serve as an
effective means of screening out the units with weak insulation.
A comparative surge PD test evaluation of Cu and Al winding
motors with identical ratings is given before and after
accelerated degradation to verify the claims made in the paper.
II. MANUFACTURING & TESTING OF AL WINDING MOTORS
A. Aluminum Stator Winding Outdoor Fan Motor
The fan of the outdoor unit of air-conditioners is usually
driven by single phase induction motors since on-off control is
used, and high performance is not required. An appliance
manufacturer recently replaced the Cu stator winding material
with Al to reduce the cost of the motor and A/C system. The
230 V, 50 Hz, 6 pole, 50 W Al winding replacement motor,
shown in Fig. 1, was designed to have identical ratings as the
Fig. 1 Example of stator endwinding insulation failure at the terminal end of
a 230 V, 50 W fan motor with Al windings

978-1-5090-0737-0/16/$31.00 ©2016 IEEE

original Cu winding motor. In the initial stages of • Pressure test: exposure to ambient temperature, relative
development, the possible causes of defective units were humidity, and pressure of 121oC, 100%, 2 atm for 48 hrs
identified from past experience and from design qualification In addition, the motors are put through accelerated aging,
and post manufacturing QA tests. An investigation revealed where a combination of excessive temperature, humidity,
that the defects/failures were mainly caused by galvanic voltage, and load stresses are applied. The worst case
corrosion and winding deformation/ breakage. Several operating conditions that the fan motors are exposed to in
adjustments had to be made to the winding manufacturing service are taken into account in the aging to ensure that the
process to screen out defective units to prevent defects or motor insulation system design can last the expected lifetime.
failure in the Al winding motors. The stator resistance measurement, insulation resistance (IR),
Galvanic corrosion is caused by the electro-potential and AC high-potential (hi-pot) tests are performed after
difference between the Al winding and Cu lead wires at the exposure of the motor samples to the conditions above to
joint where they are connected. This causes oxidation when determine if the insulation design is acceptable [3]-[5].
exposed to air, and increases the contact resistance and local Once the motor design is qualified with the above-
heating, resulting in increase in losses and insulation mentioned tests, QA tests are performed on the mass-produced
degradation. A process for covering the contact area with a stators to screen out defective units. The IR, AC hi-pot, surge
heat shrink tube after soldering or crimping the Al and Cu comparison, and stator resistance tests are performed on all
contact was added to prevent oxidation. There were also cases individual motors, and the pinhole test is performed on selected
of increased resistance/open circuits in the stator winding due samples under the conditions summarized below [3]-[7]:
to the weak mechanical strength of Al. The tensile strength and
elastic modulus of Al is lower than that of Cu by a factor of • IR test: IR measured after applying 500 V to the
3.14 and 1.82, respectively, which increases the likelihood of groundwall (GW) insulation for 1 minute (IR > 100 MΩ)
wire deformation, and lead to localized heating and/or wire • AC hi-pot test: leakage current amplitude observed after
breakage. The winding speed was lowered from 2200 rpm to applying 1800 V to the GW insulation for 1 minute
1900 rpm while adjusting the tension of the wire to prevent Al (leakage current < 3 mA)
wire defects due to deformation. • Surge comparison test: difference in surge voltage
Despite the increase in manufacturing cost due to additional waveform observed after applying a voltage impulse of
processes, the cost of the motor was reduced by 15-20% due to 1800 V with 200 nsec rise-time
the lower cost of Al. The manufacturer was able to eliminate • Pin-hole test: local discoloration (pink) of solution
many of the motor failures with the additional manufacturing observed with stator immersed in solution composed of
steps; however, the failure rate of the Al winding motors was phenolphthalein, 3% NaCl, and 0.2% ethanol with 12 V dc
still higher by a factor of over 4-5 times compared to that of Cu
applied between the conductor and solution
windings when observed over a period of 3 years of service. A
significant percentage of the failures were associated with The above-mentioned insulation tests are capable of providing
stator insulation breakdown observed in the endwinding near indications of severe defects in the insulation introduced during
the terminal end of the winding, as shown in Fig. 1. the manufacturing stage.
B. Design Qualification and Quality Assurance Testing The tests mentioned above were applied for insulation
Rigorous testing is performed on the stator insulation of the design qualification and QA testing for both Cu and Al winding
individual motors and its design to assure that they meet the samples. Although identical design qualification and QA test
quality standards required in the end product. During the procedures were used for screening out defective units, the
design and development stage, a series of design qualification difference in failure rate over a 3 year period was 4-5 times
tests are performed to check if the stator insulation can higher for the motor with the Al winding. This implies that
withstand the severe environmental conditions. This is crucial there is a failure mechanism present in the Al winding motors
considering that the units can be exposed to extreme ambient that does not occur or is not common in the Cu winding motors.
conditions during shipping or after installation all over the The failure statistics show that the unknown cause of failure is
world. The list of the environmental tests performed on the not taken into account in the design qualification and QA tests.
motor and the test conditions for qualification of the stator III. STATOR INSUALTION FAILURE IN SINGLE PHASE AL
winding insulation design are summarized as follows: WINDING INDUCTION MOTORS
• Corrosion test: exposure to 0.9% NaCl, 0.1% CaCl2, 0.25% Most of the failures in the Al winding motors were stator
Na2Co2 under 32oC for 8 hrs/day for 5-10 days insulation failures in the endwinding near the terminal end
• Salt Spray test: exposure to 5% NaCl under 32oC for 8 according to a post-mortem inspection (Fig. 1). This implies
hrs/day for 5-10 days that the failures are caused by breakdown in the turn insulation
• High temperature/high humidity test: exposure to ambient or the insulation between the main and auxiliary windings.
temperature/relative humidity of 60-85oC/85% for 72 hrs The IR and hi-pot QA tests are not effective for finding such
defects since they are intended for finding serious problems in
• Thermal stress test: exposure to 100 thermal cycles of the groundwall (GW) insulation. The surge test is capable of
o o
100 C (30 min.) and -30 C (30 min.)
finding turn insulation problems, but is known to be insensitive
for motors with large number of turns such as the motor sample
under consideration (350 turns). The surge and pinhole tests
are pass-fail tests that provide failure indications for insulation
with serious defects, and are not capable of providing
quantitative indication of the insulation condition. Therefore,
the QA tests described in II.B are not capable of screening out
motors with weak insulation between turns or between the
main and auxiliary winding with high risk of failure.
The operating environment for the single phase induction
motors in the A/C outdoor units can be hostile for the stator
insulation, especially in countries with high temperatures
and/or poor power quality. The motor placed in an outdoor
environment can be exposed to severe thermal and mechanical
stresses due to high temperature, thermal cycling, and vibration
of the compressor and/or fan. In addition, the electrical stress
can also contribute to degradation and failure of insulation.
The motor is turned on and off with a mechanical relay that is
switched up to 3000 times a year with cables that can be long,
which is a favorable condition for high amplitude, steep
fronted voltage surges. It has been reported in a number of
resources that such voltage surges can result in 1) high voltage
peaks at the motor terminal due to voltage reflection, and 2)
non-uniform voltage distribution with higher voltage in the line
end turns [5], [8]-[10]. This can potentially cause partial
discharge (PD) or puncture that can lead to rapid degradation
or failure of low voltage stator terminal end insulation [8]-[9].
The design of the Al winding motors is similar to its Cu
winding version that has the same ratings. The insulating
materials used between the turns and between the main and
auxiliary windings are identical for the two types of stators that
operate at the same temperature (Class B insulation). The
stators of both types of motors are impregnated by spraying
insulating varnish, and the insulation design qualification tests
are identical. Considering that the failure rate of Al winding
motors is 4-5 times higher despite the similar insulation design,
identical operating environment and screening tests, it is
logical to suspect that the possible root causes of failure are
related to the difference in the material characteristics. The
difference in the material properties between Al and Cu that
can have an impact on the reliability of the winding insulation
are related to mechanical and thermal properties. Al is softer
and easier to deform, and it has a thermal expansion coefficient
higher than that of Cu by a factor of 1.45.
Since Al windings are easier to deform, separation between
the conductor and insulation is more likely to occur during
motor manufacturing. There are a number of steps in the
assembly process of motors such as winding, insertion, taping,
crimping/soldering, heat shrink tubing that require handling of
the Al wires that can potentially cause conductor-insulation
separation. The non-uniformity in the quality of the thin Al
wire and the human error introduced in the winding and
handling process can increase the variance in the degree of
separation between the conductor and insulation. Separation
can also occur due to the larger differential thermal expansion
coefficient between the conductor and insulation in Al winding
motors, particularly in the bent part of the endwinding [5].
When the motor is operating, the winding temperature can
reach up to 120oC under worst case conditions.
With separation between the Al conductor and insulation
material, the stator insulation is more likely to degrade and fail
due to a combination of thermal, mechanical, and electrical
operating stresses [2], [5], [8]-[11]. Long term wear of
insulation due to mechanical stress is one of the main failure
mechanisms in low voltage motors along with thermal stress,
and loose windings provide a favorable condition for
mechanical wear. In addition, introduction of air (voids) in the
insulation system due to conductor-insulation separation
increases the likelihood of discharge/arcing due to short rise-
time voltage surges. Since the enamel insulation used in low
voltage random wound motors is not PD resistant, PD activity
can cause rapid degradation of insulation leading to failure [8]-
[11]. With weak insulation between the turns or main-
auxiliary winding exposed to the operating stresses, motor
failure due puncture of the insulation can occur due to steep
fronted voltage surges produced by switching transients, faults
in the voltage distribution system, or lightning strikes.
IV. QA TESTING BASED ON SURGE PD TEST
Considering that the main cause of the increased failure rate
in the Al winding motor units is endwinding insulation
breakdown, a new type of screening test is required for
preventing failures. There recently have been active research
efforts on finding new test methods for turn insulation
assessment, inverter-fed motor insulation testing, and in-
service insulation testing [12]-[17]. Among the new test
concepts presented, testing based on impulse or high frequency
signal excitation of the windings showed potential for
providing a quantitative assessment of the turn or phase
insulation condition [12]-[14], [17]. This work focuses on
evaluating the surge PD test as a screening test for monitoring
the endwinding insulation condition.
A. Surge and Surge PD Tests
The purpose of the surge test is to apply short rise-time and
relatively high magnitude voltage impulses to stress the
terminal end insulation to check if the motor can withstand the
surges encountered in service [6]. With a short rise-time
impulse voltage applied, a significant percentage of the voltage
is applied across the turn, phase, and GW at the terminal end of
the stator unlike the IR or hi-pot tests. The schematic of the
surge generator and motor test setup is shown in Fig. 2. The
capacitor is charged to a predetermined level and discharged
into the series RLC circuit that includes the stator winding.
This results in an under-damped oscillation at the resonant
frequency given by
(1)
݂ ൎ ͳȀሺʹɎξ‫ܥܮ‬ሻ,
where L, C are the equivalent inductance and capacitance of
the series circuit. If the voltage applied to the weakened
terminal end turn insulation is high enough, it will result in
arcing (shorting) between the turns, and decrease in L. The
increase in the resonant frequency results in a shift in the surge
waveform to the left as shown in Fig. 3 as the impulse voltage
magnitude is gradually increased, providing an indication of
weak turn insulation. It is also capable of providing indications
of weak phase and GW insulation near the terminal end.
However, it is a pass-fail test, and cannot provide quantitative
indication of terminal end insulation condition.
The surge PD test observes PD activity in the stator
insulation with a short rise-time voltage impulse applied to the
stator. This test is considered in this paper since PD activity in
the terminal end turn and phase insulation, where the voltage
stress is applied, can provide indirect indications of weak
insulation with air pockets. It was shown in a number of recent
publications that it is capable of providing indications on
endwinding turn and phase winding insulation defects or
degradation leading to failure [2], [17]-[18]. It is evaluated
under the expectation that it can provide a quantitative
indication of the turn or phase insulation condition, which was
not possible with any of the QA tests described in II.B.
The surge PD test is mainly applied for qualifying inverter-
fed low voltage motors, as described in IEC STD 60034-18-41
[19]. It is crucial to check that PD is not present in the stator
insulation when the motor is driven by an inverter, since PD
activity due to PWM switching can cause rapid insulation
degradation and failure in low voltage motors [2], [5], [8]-[11].
The test specifies details on how the turn, phase, and GW
insulation system used in inverter-fed motors can be qualified
for PD free operation. PD activity in the stator insulation can
be measured off-line under surge excitation using capacitive
couplers, electromagnetic (EM) couplers, high frequency CTs
(HFCT), or antennas, as outlined in IEC TS 61934, and shown
Fig. 2 Schematic of surge test circuit and setup
Fig. 3 Surge test waveform measured under surge test failure; the waveform
is shifted to the left due to increase in the resonant frequency
in Fig. 4 [20]. The measured signal is high pass filtered if the
impulse signal is included in the measurement, to observe the
low amplitude PD signal with high sensitivity after eliminating
the relatively high amplitude surge voltage. The PD signal
buried in the impulse waveform can be separated from the
surge voltage as shown in Fig. 4 to observe the presence of PD.
B. Application of Surge PD Testing to QA of Stator Insulation
When applying the surge PD test to the single phase AC
motor, the surge generator is connected to the motor to apply
stress to the turn or main-auxiliary winding insulation. The
voltage impulse is produced between terminals P and G
(ground) of the surge generator, and the terminals of the main
and auxiliary windings are accessible from the outside through
M and A, as shown in Fig. 2. Terminal N provides access to
the other end of the main and auxiliary windings, which are
connected inside the motor, and terminal F is the frame of the
motor. If the impulse is applied between M (P) and N, F (G)
with A floating, the voltage stress is applied to the turn, main-
auxiliary, and GW insulation at the terminal end of M. The
insulation at both terminal ends of the main and auxiliary
windings are tested, where the connections between the surge
generator and motor for the four different test configurations
are summarized in Table I.
With the voltage impulse applied as shown in Table I, PD
occurs in the air (void) of the insulation, if the voltage exceeds
the electric breakdown voltage. If the surge voltage amplitude
is gradually increased, PD will start to occur above a certain
voltage defined as the PD inception voltage (PDIV). If there is
separation between the conductor and insulation material on
the terminal end of the insulation under test, PD is likely to
occur at lower surge voltage. Therefore, PDIV measurements
can provide an indication of the presence of air pockets within
the terminal end insulation, and can be used to identify motors
with separation between the conductor and insulation with high
Fig. 4. Schematic of surge PD test setup for single phase induction motor
according to IEC TS 61934 with surge and PD waveforms shown
Table I. Test configurations for testing terminal end insulation of main and
auxiliary windings for single phase induction motor shown in Fig. 4
risk of failure. In addition to PDIV, the PD magnitude or PD
pulse count at given impulse voltage levels can also be used as
an indication of motors with air pockets in the insulation.
The following requirements are considered important when
implementing a QA test for screening out defective motor units:
• Test must be capable of providing indications of
endwinding insulation problems in Al winding motors and
show clear differentiation with that of the Cu winding
motors with lower failure rate
• Test measurements must be repeatable and provide
quantitative indication of problem to enable statistical
screening of weak motors
• Cost of investment for implementing the test must be low
• Test must be simple with short test duration
Measurement of PDIV or PD magnitude under surge excitation
can be time-consuming (costly) and subjective, considering
that the voltage has to be adjusted, and the PD signal is
irregular. This makes PDIV and PD magnitude measurements
difficult to automate and implement as a QA test according to
the test trials by the authors. On the other hand, measurement
of the PD pulse count for multiple impulse voltage levels for a
given time duration is expected to be easier to implement if a
PD pulse count antenna device is used. This test can be
implemented using the existing surge test setup by adding an
antenna, and the test can be designed to have a short test
duration. This test concept is evaluated in this paper as a QA
test for screening out motors with high risk of insulation failure
for Al winding motors.
V. EXPERIMENTAL RESULTS
A. Experimental Test Setup
To perform a comparative evaluation of the surge PD test,
40 new Al and Cu winding single phase induction motor units
were tested. The 20 Al winding motors were split into two
groups, samples A1-A10 and B1-B10, and the 20 Cu winding
motors were also split into two groups, C1-C10 and D1-D10.
Accelerated aging was performed on Al and Cu winding motor
samples A1-A10 and C1-C10, and the aged motors are referred
to as samples AA1-AA10 and CC1-CC10, respectively, as
summarized in Table II. Both types of A/C outdoor fan motors
are rated at 230 V, 50 Hz, 6 pole, 50 W.
The surge PDIV and IR measurements were made, and hi-
pot and surge breakdown tests were performed for new and
aged Al (samples B1-B10, AA1-AA10) and Cu (samples D1-
D10, CC1-CC10) winding motors. A commercial 12 kV surge
test equipment was used for generating the variable amplitude
voltage impulses at 5 pulses per second. For PDIV
measurement, a commercial electric discharge detector antenna
with bandwidth of 80 to 300 MHz was used with the test setup
configured as shown in Fig. 4 and Table I. The repetitive
PDIV, which is defined as the impulse voltage at which more
than five PD pulses occur for ten applied voltage impulses, was
recorded. The IR value was measured after applying 500 V dc
voltage for 1 minute using a commercial IR tester, and the dc
hi-pot test was also performed by applying 2 kV to the GW
insulation for 1 minute to confirm that the GW insulation is not
damaged for the 40 motor units. The last test performed on the
40 motor samples (AA1-AA10, B1-B10, CC1-CC10, D1-D10)
was the surge breakdown test. The impulse voltage magnitude
was gradually increased until surge test failure, and the
breakdown voltage was measured. It was considered a surge
test failure if the pulse-pulse error area ratio (PPEAR)
exceeded 10%. The PPEAR is the percentage error between
the integral of the present and previous pulses when increasing
the impulse voltage in uniform voltage steps. A shift in the
surge waveform as shown in Fig.3 due to arcing between turns
would result in a noticeable increase in PPEAR [21].
The number of PD pulses under surge excitation was
recorded for the Al and Cu winding motors before (A1-A10,
C1-C10) and after (AA1-AA10, CC1-CC10) accelerated
degradation to verify the capability of the proposed QA test.
The commercial surge tester equipment was used for excitation
and a UHF antenna was used for counting the PD pulses. The
number of pulses was recorded while applying 50 impulses at a
given voltage level set between 1200 and 2400 V with the
voltage increased in 150 V intervals. For accelerated aging, the
motors were placed in an environmental chamber, as shown in
Fig. 5, with the ambient temperature and relative humidity
regulated at 85oC and 85%, respectively. The motors were
operated at rated voltage under no load conditions for 3240
hours. The stator winding temperature was monitored with
thermocouples installed, and the motor was turned off when
the temperature reaches 160oC and turned back on when the
motor cools down to 130oC.
B. Experimental Results:Cu-Al, New-Aged Motor Comparison
The PDIV measurements are shown in Figs. 6-7 for new
and aged Al (samples B1-B10, AA1-AA10) and Cu (samples
D1-D10, CC1-CC10) winding samples. The tests were
performed with the impulse applied to stress the insulation of
both terminal ends of the main and auxiliary winding (MA,
NM, AN, NA windings), as summarized in Table I. The PDIV
was measured 3 times for each of the 40 motor samples and the
average values are shown in Figs. 6-7 for each group. It can be
observed that the new Al winding motor units have a PDIV
value 230-660 V (15-30%) lower than that of the Cu winding
motors. Lower PDIV implies that air pockets are present in the
winding insulation, and makes Al winding motors more
Table II. Summary of Al and Cu winding motor samples used for testing
Fig. 5. Environmental chamber for accelerated aging of motor samples
vulnerable to electrical and mechanical aging. Therefore, low conditions described in V.A. does not cause change in the IR,
PDIV under surge excitation can provide information on motor and that all 40 motors passed the DC hi-pot test, indicates that
units with high risk of failure. The lowest PDIV measured in there was no significant degradation in the GW insulation.
the 10 new Al winding motors was 1060 V on the N side of the
auxiliary winding (NA) of motor B3. Considering the variance The average values of the surge breakdown voltage
in the PDIV, the motor with the lowest PDIV in the hundreds measurements for the 40 new and aged Al and Cu winding
of volts range have high risk of going through electrical aging motors are shown in Figs. 9-10. The reason two sets of 10 Al
resulting in early failure. (A and B) and Cu (C and D) winding motors were used in this
paper is because the new units that are tested for surge
The PDIV measurements were repeated for the aged Al and
Cu winding samples AA1-AA10 and CC1-CC10. There was a
noticeable decrease in the PDIV for both motors at the terminal
ends of all windings, as can be seen in Figs. 6-7. The PDIV
decreased to below 1000 V in 3 of 20 motor samples, where
the lowest PDIV was 750 V. This shows that PDIV decreases
as the motor insulation degrades due to operating stresses. The
likelihood of motor failure would therefore increase with
degradation especially for the motors with low PDIV.
The IR and DC hi-pot tests were performed on all of the 40
B1-B10, D1-D10, AA1-AA10, and CC1-CC10 motor samples
to evaluate the GW insulation condition. The average values
of the IR measurements of the new and aged Al and Cu
winding units are summarized in Fig. 8. The results show that
there is no meaningful difference between the IR values of the
Al and Cu winding motors whether they are aged or unaged. Fig. 8. Average of IR measurements for Cu and Al winding motors units
The fact that 3240 hrs of accelerated aging under the before (B1-B10, D1-D10) and after (AA1-AA10, CC1-CC10)
accelerated aging

Fig. 6. Average of PDIV measurements performed on MN, NM, AN, NA Fig. 9. Surge breakdown voltage measurements of MN, NM, AN, NA
windings of Al winding motor units before (B1-B10) and after (AA1- windings of Al winding motor units before (B1-B10) and after (AA1-
AA10) accelerated aging AA10) accelerated aging

Fig. 7. Average of PDIV measurements performed on MN, NM, AN, NA Fig. 10. Surge breakdown voltage measurements of MN, NM, AN, NA
windings of Cu winding motor units before (D1-D10) and after windings of Cu winding motor units before (D1-D10) and after
(CC1-CC10) accelerated aging (CC1-CC10) accelerated aging
breakdown voltage (B and D) cannot be used for further testing
or aging since the turn insulation is likely to be punctured. For
the new Al and Cu motors, the average value of the surge
breakdown voltage was lower for the Al winding motors by 96-
560 V. This indicates the Al winding insulation is inferior in
withstanding voltage surges, which is consistent with the PDIV
measurements shown in Figs. 6-7. The difference between the
Al and Cu winding motors’ surge withstand capability
increases after the 3240 hour accelerated degradation. It can be
observed in Figs. 9-10 that the surge breakdown voltage is
lower for the Al winding motors by 714-1168 V after the
accelerated aging, which indicates faster degradation with
aging. The results shown in Figs. 6-10 show that surge PDIV
can be used as an indicator of Al motors with high risk of
failure by screening out the units with low PDIV.
C. Experimental Results – QA Testing
Although the PDIV can provide a good indication on the
presence of air pockets in the endwinding insulation to predict
the condition, as shown in Figs. 6-7, it is a time-consuming test
since the antenna output must be observed carefully while
adjusting the voltage. Therefore, it is difficult to automate as a
QA test, and it can increase the cost of the manufacturing
process and motor. To verify if the surge PD test can be used
for QA testing, the number of PD pulses were measured and
recorded for the 40 Al and Cu winding motors before (A1-A10,
Fig. 11. Number of PD pulses for 50, 1200-2400 V impulses in 150 V steps
for NA winding of Al winding motor before (A1-A10) and after
(AA1-AA10) accelerated aging (thick lines are average values)
Fig. 12. Number of PD pulses for 50, 1200-2400 V impulses in 150 V steps
for NA winding of Cu winding motor before (C1-C10) and after
(CC1-CC10) accelerated aging (thick lines are average values)
C1-C10) and after (AA1-AA10, CC1-CC10) aging. The PD
pulse count for 50 applied impulses between 1200-2400 V in
150 V steps for the 40 motors are shown in Figs. 11-12 for the
N side terminal of the auxiliary winding (NA). The
measurements performed on the 10 individual motors are
shown as thin lines and the average values are shown as thick
lines. The results show that PD activity can be observed at
lower voltages (more PD pulses for a given voltage) for Al
winding motors, which is consistent with the results shown in
V.B. The increase in the PD pulse count with aging is also
noticeable for both motors, as expected. The PD pulse count
on the individual motors increases monotonously for most
motors, as the voltage is increased. The average values of the
PD pulse counts for the MN, NM, AN, NA windings before
and after aging are summarized in Figs. 13-14 for Al and Cu
winding motors, respectively. The trend in the increase in the
PD pulses with increase in applied impulse voltage before and
after accelerated aging is consistent with that of Figs. 11-12
and the results shown in V.B.
It should be noted that the voltage levels, and the number of
impulses applied under each voltage condition in Fig. 11-14
were chosen for evaluation of the proposed method for QA
testing. It can be chosen to measure the number of PD pulses
at one or multiple impulse voltage levels for any number of
applied impulses depending on the degree of accuracy desired.
It is a trade-off between test time and accuracy, since the
Fig. 13. Average number of PD pulses for 50, 1200-2400 V impulses in 150
V steps for the MN, NM, AN, NA windings of Al winding motor
before (A1-A10) and after (AA1-AA10) accelerated aging
Fig. 14. Average number of PD pulses for 50, 1200-2400 V impulses in 150
V steps for MN, NM, AN, NA windings of Cu winding motor before
(C1-C10) and after (CC1-CC10) accelerated aging
repeatability can be improved with larger number of impulses
and number of voltage levels, which also takes longer time. In
any case, the advantage of the test is that it can be easily
performed with the present manufacturing QA test setup by
adding an antenna since the surge test is included in most QA
processes. The motors with high pulse count for a given
impulse test voltage are the units with air pockets that are
vulnerable to mechanical and electrical degradation and failure.
This test can therefore be used to screen out motors with high
risk of failure at low added cost.
VI. CONCLUSION
Potential root causes and solutions for the increased failure
rate in single phase induction motors that employ Al windings
were investigated in this paper. The increased failures were
mainly due to endwinding insulation breakdown near the
terminal end, which implies that the turn or main-auxiliary
winding insulation is susceptible to failure. Considering that
existing qualification and QA tests could not screen out units
with weak insulation, the surge PD test proven to be capable of
providing indications on turn and phase insulation condition
was evaluated as a QA test.
The surge PDIV, IR, hi-pot, and surge breakdown tests
were performed on 40 Al and Cu winding A/C outdoor unit fan
motor samples before and after accelerated aging. The test
results show that the surge PDIV and surge breakdown voltage
are significantly lower for the Al winding motors when
compared to Cu winding motors with similar design and
insulation material. The results also show that PDIV and surge
withstand capability of the Al winding motor degrades faster
with accelerated aging, while there is no noticeable change in
the GW insulation IR or hi-pot test results. It can be deduced
from the results that separation between the Al conductor and
insulation is more likely due to the material properties, and
result in formation of air pockets in the endwinding insulation
and low surge PDIV and breakdown voltages. Since the units
with low PDIV have a higher risk of failure due to electrical
and mechanical aging, the surge PD test can provide
quantitative information on units with weak insulation. In
addition to the evaluation, a QA test based on the PD pulse
count at given impulse voltage levels is suggested as a simple
means of screening out Al winding motor units with high risk
of failure. Adding the proposed method in the QA process is
expected to help decrease the failure rate with minimal increase
in the cost of testing.
VII. ACKNOWELDGMENT
We would like to thank Mike Teska of SKF Condition
Monitoring Center, Fort Collins, CO, USA, for donation of the
12 kV surge generator and antenna based PD detector used for
obtaining the test results in the paper.
REFERENCES
[1] K. Lee, J. Hong, S.B. Lee, and S. Lee, “Quality assurance testing for
magnetization quality assessment of BLDC motors used in compressors,”
IEEE Trans. on Ind. Appl., vol. 46, no. 6, pp. 2452-2458, Nov./Dec.
2010.
[2] D. Bogh, J. Coffee, G. Stone, and J. Custodio, "Partial discharge
inception testing on low-voltage motors," IEEE Trans. on Ind. Appl.,
vol. 42, no. 1, pp. 148-154, Jan./Feb. 2006.
[3] IEEE Recommended Practice for Testing Insulation Resistance of
Electric Machinery, IEEE Standard 43-2013, 2014.
[4] IEEE guide to induction machinery maintenance testing and failure
analysis, IEEE Standard 1415-2006, 2006.
[5] G.C. Stone, E.A. Boulter, I. Culbert, and H. Dhirani, Electrical
insulation for rotating machines – design, evaluation, aging, testing, and
repair, IEEE Press Series on Power Engineering, John Wiley and Sons,
2nd Edition, 2014.
[6] IEEE guide for testing turn insulation of form-wound stator coils for
alternating-current electric machines, IEEE Standard 522-2004, 2004.
[7] D.J. Barta, “Salt water pinhole testing of magnet wire,” Proc. of
Electrical Insulation Conference and Electrical Manufacturing Expo,
pp.171-174, Oct. 2005.
[8] M. Kaufhold, G. Boerner, M. Eberhardt, and J. Speck, “Failure
mechanism of the interturn insulation of low voltage electric machines
fed by pulse-controlled inverters”, IEEE Electr. Insul. Mag., Vol. 12,
No. 5, pp. 9-16,. 1996.
[9] W. Yin, “Failure mechanism of winding insulations in inverter-fed
motors, IEEE Electr. Insul. Mag., Vol. 13, No. 6, pp. 18-23, Nov./Dec.
1997.
[10] N.K. Ghai, “Design and application considerations for motors in steep-
fronted surge environments,” IEEE Trans. on Ind. Appl., vol. 33, no. 1,
pp. 177-186, Jan./Feb. 1997.
[11] G.C. Stone, B.K. Gupta, M. Kurtz, and D.K. Sharma, “Investigation of
turn insulation failure mechanisms in large AC motors,” IEEE Trans. on
Pwr. App. and Syst., vol. PAS-103, no. 9, pp. 2588-2593, Sept. 1984.
[12] S. Grubic, J. Restrepo, J.M. Aller, B. Lu, “A new concept for online
surge testing for the detection of winding insulation deterioration in low-
voltage induction machines,” IEEE Trans. on Ind. Appl., vol.47, no.5,
pp.2051-2058, Sept./Oct. 2011.
[13] P. Neti, S. Grubic, “Online broadband insulation spectroscopy of
induction machines using signal injection,” Proc. of IEEE ECCE, pp.
630-637, Sept. 2014.
[14] P. Nussbaumer, M.A. Vogelsberger, T.M. Wolbank, “Induction machine
insulation health state monitoring based on online switching transient
exploitation,” IEEE Trans. on Ind. Electron., vol.62, no.3, pp.1835-
1845, Mar. 2015.
[15] K. Younsi, P. Neti, M. Shah, J.Y. Zhou, J. Krahn, K. Weeber, C.
Whitefield, “On-line capacitance and dissipation factor monitoring of
AC stator insulation,” IEEE Trans. Dielectr. Electr. Insul., vol. 17, no.
5, pp. 1441-1452, Oct. 2010.
[16] P. Zhang; K. Younsi, P. Neti, “A novel online stator ground-wall
insulation monitoring scheme for inverter-fed AC motors,” IEEE Trans.
on Ind. Appl., vol. 51, no. 3, pp. 2201-2207, May/June 2015.
[17] J. Yang, T. Kang, B. Kim, S.B. Lee, Y. Yoon, D. Kang, J. Cho, H. Kim,
“Experimental evaluation of using the surge PD test as a predictive
maintenance tool for monitoring turn insulation quality in random
wound AC motor stator windings,” IEEE Trans. Dielectr. Electr. Insul.,
vol. 19, no. 1, pp. 53-60, Feb. 2012.
[18] M. Tozzi, A. Cavallini, and G.C. Montanari, “Monitoring off-line and
on-line PD under impulsive voltage on induction motors - part 1:
standard procedure,” IEEE Electr. Insul. Mag., vol. 26, no. 4, pp. 16-26,
July-Aug. 2010.
[19] Rotating electrical machines – Part 18-41: Partial discharge free
electrical insulation systems (Type I) used in rotating electrical
machines fed from voltage converters – qualification and quality control
tests, IEC International Standard 60034-18-41, 2014.
[20] Electrical insulating materials and systems – Electrical measurement of
partial discharges (PD) under short rise time and repetitive voltage
impulses, IEC Technical Specification 61934, 2014.
[21] J. Wilson, “Current state of surge testing of induction machines,” Proc.
of Iris Rotating Machine Conference, pp. 1-13, 2003.