Feature

I nsulation Testing
as an Electrical Maintenance Tool

Insulation Testing
Electrical testing is essential to the evaluation and maintenance of mo-
tor performance and life. A broad range of testable properties exist, from
the simple and general tests such as voltage measurement, to the highly by Jeff Jowett
complex and specific such as surge testing. (The latter is a technique that Megger
pinpoints turn-to-turn shorts.) These properties are too diverse to be cov-
ered comprehensively as a single topic, but a universal and indispensable
general method is that of insulation testing.
In its basic terms, an insulation tester applies a high dc voltage across
an insulation barrier, measures the amount of current (commonly called
“leakage”) flowing through the insulation, and applies Ohm’s law to cal-
culate the resistance of the insulation. The basic theory is quite simple, no more than half a milliampere!
but the testers themselves are complex – primarily because they must Consequently, testers need output
generate a high voltage and be able to measure accurately a miniscule no more than a few milliamperes
current. to be able to cover the full range
There is no perfect insulator. Apply enough voltage and any material of anything fit to be considered
can break down. (Think of a lightning strike!) However, by applying the as insulation. Once this limit is
test voltage across the insulation, the tester is establishing a highly atypi- reached, voltage will drop, further
cal “circuit” in which an insulating material is being forced to act as a protecting a test item having test
conductor. Obviously, it is not going to conduct very much, so the tester results on the unacceptable end of
must be able to sense currents on the nanoampere level and below. These the scale.
are the principle characteristics of an insulation tester. Because resistance Voltage should rise sharply
is high, measurements are made in units of megohms through the low end of the range,
A caution about voltage supply is in order. Many economy-level testers then maintain full, selected voltage
provide only nominal test voltage selection. That is to say, a given volt- through good values. A poorly
age selection may only provide that voltage over a limited portion of the designed tester will exhibit a slow
resistance range. To avoid this problem, the tester should be provided with rise, not achieving full voltage until
a load curve of output voltage versus load resistance. This curve should well up into the range. The critical
indicate a fast rise up to full, selected voltage at a level of one to five range of values, where the motor
megohms, then maintain that voltage throughout the remaining range. is still operating but is approach-
While dc insulation testing is nondestructive (as opposed to high-po- ing a need for maintenance, will be
tential testing), there are limits where severely deteriorated insulation getting tested at deceptively low
is concerned. To avoid any further damage to insulation already near voltages and yielding potentially
breakdown, test current is limited. Since IEEE Standard 43 recommends misleading results.
never less than two megohms for rotating machinery, a simple applica-
tion of Ohm’s law indicates that at a typical 1,000-volt test, this represents

Summer 2004 1
 The benefit of this testing is that the readings can be Preventive Maintenance
used like the odometer on a car, to give an idea of the The second broad area of testing, preventive main-
working age of the motor, how much can be reason- tenance, is more sophisticated. It is recommended that
ably expected of its performance, and where it is on high quality testers be used because these will have the
its life cycle. From the time a motor is put into service, range necessary to measure equipment in the earliest
insulation picks up moisture, dirt, and other contami- stages of its life cycle, preferably beginning with an in-
nants, suffers wear and tear, is carbonized from voids stallation test. In such applications, a simple pass/fail
and other imperfections, or may be damaged by line tester with limited range will yield only an over range
disturbances. From initially high values (possibly te- result, typically called “infinity.” This is not a true mea-
raohms), these factors will cause insulation resistance surement but merely an indication that the motor has a
to decline, often at a fairly steady and predictable rate. resistance greater than the sensitivity of the tester. Such
An insulation test is a quick and simple maintenance an assurance is satisfactory for pass/fail applications
check that captures the information available from the but does nothing to establish a time line for life expec-
aging process and, by inference, reflects the overall tancy. It may be years before resistance drifts down to
condition of the motor itself. Of course, catastrophic where it can be measured, with considerable loss in
failure can also occur, as in flooding, fire, or lightning flextime for scheduling routine maintenance. There is
strike. An insulation test is typically performed here no better way to evaluate test results than comparison
also, but such events exemplify the separation be- with a previous test. Readings may be quite high and,
tween the two broad areas of testing: pass/fail and if evaluated singly, appear “good.” Yet, if the motor is
preventive maintenance. operating in a hostile environment, possibly exposed
to corrosive fumes or high moisture, readings may be
Pass/Fail Testing dropping rapidly. Early detection would allow time
Pass/fail testing is the more commonly employed for corrective action before significant damage could
testing: a motor has failed, is performing erratically, occur. Conversely, comparatively low readings hold-
or is tripping breakers. An insulation test is performed ing steady probably indicate a widely dispersed low
to look for breakdown. It may appear as a dead short level of harmless leakage, so that critical maintenance
or unacceptably low reading (typically less than two time can be spent on more pressing problems.
megohms). In such applications, a sophisticated tester Standardized test methods can be judiciously
is not required, and the operator may not be concerned employed to help with test time and interpretation.
about the actual readings. Experienced personnel may A number of methods have been devised, and each
look only for direction of pointer travel – toward the evaluates the insulation in a different way. Except for
low or high end of the scale. If low, an electrical prob- the most critical and high-capital equipment, it is not
lem exists, and more specific troubleshooting can com- practical to perform a full regime of test methods.
mence. If high, time might be better spent looking for Rather, judicious selection of the most appropriate
a mechanical problem such as bearing wear. method for a given set of circumstances can maximize
Pointer travel is more significant in insulation test- results while conserving test time. Two broad catego-
ing than with any other type of test. It is also frequently ries of insulation failure are pervasive: contamination
confusing and frustrating to the uninitiated. Motor and localized damage. Two of the most popular test
windings, by their structure, possess a good deal of methods are polarization index and step voltage.
capacitance. Each turn represents two conductors
separated by insulation – a tiny capacitor. The larger Polarization Index
the winding, the greater the capacitance. Secondly, Polarization index (PI) takes advantage of the dif-
insulating material itself will realign molecularly un- ferential between harmless charging currents and
der the influence of an applied voltage field. This also damaging leakage. In healthy equipment, leakage is
represents a flow of charging current, called absorp- small by comparison, so the charging currents domi-
tion. Because of these charging effects, readings start nate pointer travel. A fairly large difference will exist
low and gradually rise as charging goes to completion. between a reading at one minute and at ten. When
What remains after full charge is the current that flows converted to a ratio, typical numbers are seen in the
as a product of the level of imperfection of the mate- range of two to four. On the other hand, deteriorated
rial, called leakage. With the largest windings, this can insulation will be dominated by leakage, which is
take hours – an impracticably long test time. Therefore, constant for a given test voltage, and will mask the
special procedures, called test methods, have been effects of charging. The ten-minute reading will be
worked out to bring the testing into workable time similar to the one-minute reading, and a poor PI of
frames and act as an aid in interpreting results. less than two is the result. The PI is good for detecting
pervasive contamination because the effect on leakage
versus charging current will be readily apparent. It also

2 NETA WORLD
solves a number of practical problems. Dependence imminent, but this type of insulation will often yield
on knowledge of resistance values is eliminated deceptively high readings. An unnaturally high PI
because, whether they are in the tens, hundreds, or will reveal the condition, but can be mistaken for a
thousands of megohms, the result will be the same. good reading by those not familiar with past history.
Therefore, the PI is a good test where no prior records A step voltage test can be expected to draw increas-
are available. ing amounts of leakage through cracks in the dried
PI is also useful on very large motors for two rea- insulation with each voltage increment and readily
sons: time and voltage. Large windings can take a reveal the condition.
prohibitively long time to reach a stable reading, but
that is not necessary with PI. The one- and ten-minute Testing Standards
readings can be relatively low on the time-resistance
The accepted standard for electric motor testing
curve, yet the result is still valid. Caution must be tak-
is IEEE Std 43-2000, Recommended Practice for Test-
en with smaller windings, however, as they may reach
ing Insulation Resistance of Rotating Machinery. This
full charge quickly and yield a low PI even though
standard was revised in 2000 in order to take into ac-
perfectly good. Similarly, the tester must have a good
count the effects that newer types of insulation have
range so that it will not overrange before ten minutes.
on standard testing procedures and interpretation
(Infinity is a concept, not a number and cannot be used
of results. Industry practices, in the interest of time,
to calculate a PI.) The second advantage with large
may interconnect all windings and test them simul-
motors is that they may have high operating voltages
taneously to ground or case. But IEEE recommends
that put “as rated” tests beyond the test voltages of
phases be tested individually, with remaining phases
all but large, expensive testers. PI can be performed
and stator core or rotor body connected to a common
successfully at values below rated.
ground. This will reveal leakage between phases as
well as phase-to-ground. Similarly, testing can be
Step Voltage done through external equipment like motor starters
A step voltage test is performed by increasing the and cables, but it must be understood that the external
test voltage in steady increments and observing the equipment will be included in the measurement and
response of the insulation readings. Voltages and time bring it down. IEEE prefers that external equipment
intervals can be defined by the operator to meet the be disconnected.
demands of test equipment and work schedule, but Testers may also have guard terminals which act
industry standard is five voltages (e.g., one, two, three, as shunts for one or more parallel leakage paths. In a
four, and five kilovolts) in one-minute intervals. Good complex piece of equipment like a motor, there may
insulation will stand up to this treatment, but dete- be several leakage paths. These paths can be examined
riorated insulation will open more leakage paths with more specifically by eliminating one or more through
each increase, and readings will steadily drop. Use guard connections. For instance, stator-to-rotor leak-
insulation resistance and polarization index minimum age can be measured with ground leakage guarded.
acceptable values as a prerequisite for this test if the The 2000 revision of IEEE Std 43-2000, Recommended
higher step voltages exceed the insulation rating. If Practice for Testing Insulation Resistance of Rotating Ma-
the IR and PI test results are below minimum accept- chinery, includes a table of recommended test voltages
able values, clean and/or dry the insulation before based on equipment rating. In response to newer in-
trying a test that will stress the dielectric capability sulation types, the recommended test voltages have
of the insulation. been increased up to ten kilovolts for motors rated at
12 kilovolts and above. Both insulation resistance and
Applications polarization index tests are recommended, with one
or the other performed on equipment rated less than
These two tests, and most other standard meth-
10,000 kilovoltamperes, and both tests suggested for
ods, are not mutually exclusive, but complementary.
equipment rated higher. There are also some interest-
Each will reveal a variety of problems, but each has
ing observations for atypical windings. Large turbine
individual strengths. Step voltage is a good test for
generators may have unencapsulated field windings
recognizing elusive intermittent failures arising from
where little absorption can take place, so a PI is not a
localized damage in otherwise good insulation. A car-
recommended test. It is also advisable for the operator
bon track that has burned through or a pinhole caused
to have knowledge of the basic schematic in order to
by a voltage spike may be lying in an air gap instead of
avoid testing errors. For instance, windings of some
making direct contact to ground. An insulation resis-
induction motors may not be insulated from the rotor
tance test may still yield a high reading and overlook
body, so a PI is not possible.
the critical flaw. But step voltage can arc the gap and
expose the problem. Another specialized problem is
dry and brittle insulation on old motors. Failure is

Summer 2004 3
 Traditional notions about PI testing are also un-
dergoing an upgrade as the result of new insulation
types. Absorption currents may decay to near zero
in much shorter time frames, making the traditional
one-to-ten ratio dubious. Accordingly, newer defini-
tions have incorporated ratios of one minute-to-five
minutes and 30 seconds-to-one minute. An alternative
suggestion is to record the reading every minute until
three consecutive intervals yield the same result. The
caveat for all of these newer methods is that they are
not standardized, and insufficient data has been col-
lected to set new pass/fail values. IEEE requests such
information be forwarded by field sources.
In summary, insulation testing offers a quick and
convenient window into the electrical condition of a
motor and provides a reliable basis upon which fur-
ther maintenance and troubleshooting decisions can
be made.

Jeffrey R. Jowett is Senior Applications Engineer for Megger

4 NETA WORLD