Full text of DOI:10.1109/2943.877842

Alex McEachern
he most common electric power-quality
■ Immunity: a low-speed transient to 150% of
nominal power line voltage for 0.1 cycles is
delivered to a microwave oven, but the oven
continues to function normally.
events —s ags, swells, and tran-
sients —h ave been identified and quan-
tified in recent national studies by U.S.
T
utilities and the Electric Power Research Institute
ERPI). Utilities are making efforts to improve
■ Tolerance: a voltage sag to 60% of nominal
for 5 s is delivered to the microwave oven.
The cooking power is greatly reduced during
that interval, but the timing functions con-
tinue to operate normally, and the oven pro-
ceeds with its programmed sequence.
■ Graceful failure: a voltage swell to 170% of
nominal for 5 min is delivered to the micro-
wave oven, and a user-replaceable fuse or cir-
cuit breaker opens.
(
power quality; however, it also makes sense to de-
sign loads, such as appliances, process control com-
puters, and semiconductor manufacturing
systems, to tolerate common power-quality events.
This article suggests some design criteria for
“
power-quality-tolerant” designs, and suggests
ways of achieving them. Warranty and service
costs will be reduced and customer satisfaction
increased.
All three modes of survival provide useful customer
benefits and reduced service and warranty costs.
Immunity, Tolerance, and Graceful Failure
In discussing how to design for power-quality
events, we first need to define “survival.” It is use-
ful to distinguish three different modes of survival:
immunity, tolerance, and graceful failure.
Electric Power-Quality Studies and Standards
The electric power industry, led by ERPI [1], has
been studying power quality for at least 15 years.
Immunity means that the user is unaware that a
power-quality event has taken place. Tolerance
means that the user may be aware that a
power-quality event has impacted the device, but
is not required to take any action to restore the de-
vice to its normal operating mode. Graceful failure
means that the device may fail during a
power-quality event, but it is relatively easy for the
user to restore operation.
Here are three examples of survival modes:
Fig. 1. The Electric Power Research Institute used
PQNodes like this one to gather power-quality data
at sites all over the United States. The three-year
project provides a strong, statistically valid base of
information about common power disturbances.
Alex McEachern is with Power Standards Lab of
Emeryville, California. McEachern is a Senior Member
of IEEE. This article appeared in its original form at the
1998 International Appliance Technology Conference.
66
IEEE Industry Applications Magazine
■
November/December 2000
1077-2618/00/$10.00©2000 IEEE

The most extensive study is EPRI’s recently com-
pleted National Survey of Distribution Power Quality
vated at a nonzero instantaneous distribution volt-
age. The transient may typically rise to as much as
180% of the nominal voltage, but will decay
within less than a single fundamental cycle.
Momentary interruptions are generally caused
by proper operation of the utility distribution sys-
tem. A fault (such as a tree branch, squirrel, etc.)
causes an overcurrent; the utility’s circuit breaker
trips, then, the circuit breaker automatically
recloses, typically in 5 to 15 s.
Lightning-induced impulsive transients can rise
to as much as 6 kV for 1 or 2 ms. There are indus-
try-standard waveforms for testing susceptibility.
Neutral-failure-induced sustained overvoltages
can, in theory, be as high as 200% of nominal volt-
age, but in practical situations, rarely rise above
170% of nominal voltage. They are caused by a
combination of unbalanced loads and a neutral
connection failure in split-single-phase environ-
ments, the most common residential environment
in the United States. A complete description of the
causes and characteristics of this problem can be
found in PQ Today [5].
[
3
2], which recorded power-quality events at over
00 carefully chosen locations throughout the
country for three years. Your local utility, if it is a
memberofEPRI, canprovideacopyofthisstudy.
The IEEE has several working groups that set
recommended practices on power quality, includ-
ing IEEE-519, which covers harmonics, and
IEEE-1159 [3], which covers monitoring electric
power quality. IEEE-1100 [4] covers powering
and grounding for computers and similar equip-
ment. It is a useful guide for thinking about power
for devices that contain electronic controls. ANSI
C84.1 defines various voltage tolerances, generally
for longer intervals.
The semiconductor manufacturing industry
has published an excellent standard for voltage sag
immunity called SEMI F47. Free copies can be re-
quested at http://www.PowerStandards.com.
The statistics in these national studies can be
used to evaluate the economic return on improving
the power-quality tolerance of a design. The statis-
tics, combined with power-quality compatibility
testing, will indirectly quantify the number of ex-
pected failures without modification; an economic
benefit from reduced warranty cost and increased
customer satisfaction can then be calculated.
Suggested Power-Quality Design Goals
Standards for power-quality tolerance (or electri-
cal system compatibility) have not yet been devel-
oped and published. The suggestions below are
based on the author’s experience and judgment
and are offered as a reasonable goal. It is likely that
a design that meets these goals will have greatly
reduced power-quality-related service and war-
ranty costs.
Immunity: 80% to 120% of nominal voltage
for an indefinite period; 150% of nominal voltage
for a single cycle; 2-kV impulsive transient.
Tolerance: 0% to 120% of nominal voltage for
Common Power-Quality Events
Although a huge variety of power-quality events
can take place, the most common in a residential
environment are:
■
Voltage sags from a variety of sources
each source has an associated set of sag
(
characteristics);
■
Power-factor-correction-induced, low-fre-
quency oscillatory transients;
Momentary interruptions;
Lightning-inducedimpulsivetransients; and
Neutral-failure-induced sustained
overvoltages.
1
3
5 s; 180% of nominal voltage for a half-cycle;
-kV impulsive transient.
■
■
■
PQPager/3100 PQNode Waveshape Disturbance Three Phase Delta
200.0
If you design to survive these events, you will
have dealt with almost all practical problems; rarer
events like bursts of high-frequency noise, and
rarer problems like distorted voltage waveforms,
can probably be ignored. Sustained power inter-
ruptions are generally not considered to be
power-quality “events,” and customers generally
do not expect their devices to operate during a sus-
tained interruption.
2 V
0.0
Voltage sags have two common causes: local
loads (e.g., motor starting currents, heaters) that
generally will not sag below 70% for more than 2 s
and faults on the utility distribution system that
generally will not sag below 80% and will gener-
ally be cleared by reclosers within 15 s.
Power-factor-correction-induced, low-fre-
quency oscillatory transients appear whenever a
switched power-factor-correction capacitor is acti-
−
200.0
0
.00 ns
3.33 ms/div
66.67 ms
06/12/97 08:08:19.00 AM
Arnold Sub
Fig. 2. A typical small disturbance. Power-factor-correction capacitors
can induce low-frequency oscillatory transients like this one, the second
most common disturbances. By far, the most common are brief reductions
in rms voltage, called voltage sags.
IEEE Industry Applications Magazine
■
November/December 2000
6
7

means that a simple, inexpensive
diode/capacitor combination may
be sufficient.
In general, you will not want to
approach this section of the prob-
lem by increasing the size of the
power-supply filter capacitor be-
cause that will generally lead to in-
creased power-line harmonic
currents. Instead, the energy stor-
age should be incorporated within
or downstream from the power-
supply regulation circuits.
It is important to analyze the
behavior of those parts of the sys-
tem that do not receive power from
the energy-storage device. Two
common problems: those parts of
the system may lapse into a state
that is not expected by the micro-
processor or there may be an unin-
tended power leakage path
through signal or control lines to
the “unpowered” sections.
Fig. 3: Even a brief voltage sag (upper graph) can lead to a huge momentary
current surge (lower graph) at the end of the sag. Subtle problems like this
can be diagnosed with a sag generator.
Sags are, by far, the most common
power-quality event, so it makes
Graceful failure: less than 80% or more than
20% of nominal voltage for an indefinite period;
-kV impulsive transient.
sense to focus most of your efforts on this problem.
Two often-overlooked problems occur at the end
of a voltage sag, when the voltage returns to nor-
mal. A discharged filter capacitor may suddenly
cause a huge current surge —s oft-start circuits are
often disabled at this point —w hich, in turn, can
disrupt other circuits. Otherwise, a power-on-reset
circuit may interpret the sudden increase in volt-
age at the end of a sag as requiring the entire system
to be reset. Be careful, and test your design with a
sag generator [6].
1
6
How to Achieve Power-Quality Design Goals
First, ensure that you have the ability to test your
design for power-quality tolerance and compati-
bility or engage a group such Power Standards Lab
to perform your testing. You need to know how
your existing designs behave.
Overvoltage Tolerance and Immunity
In general, this is simply a matter of selecting com- Impulsive Transient Tolerance and Graceful
ponents that can tolerate brief and sustained Failure
overvoltages. Incremental costs are minimal. In
some designs, an overvoltage sensor may need to be
desensitized.
In general, inexpensive suppression devices such as
metal oxide varistors (MOVs) are used to accom-
plish this goal. These devices are placed across the
power conductors after the fuse. They have a high
impedance at normal voltages, but rapidly switch
to a low impedance when the voltage rises above
some threshold.
Coordination with internal devices, fusing, and
(surprisingly) other loads is the key to proper MOV
application. You must select an MOV with a volt-
age threshold that is lower than a voltage that will
damage other components in your design, such as
transformer insulation or electronic switches, and
one that can dissipate the available energy in the
expected impulsive transient. Yet, you want to se-
lect an MOV with the highest possible voltage
Sag and Interruption Tolerance and
Immunity
This requires some form of energy storage in the
design, typically a capacitor. The key here is to se-
lect and electrically isolate the parts of the design
that will draw power from the energy-storage de-
vice. For example, in a washing machine, the elec-
tronic controls should be able to draw power from
the storage device, but the motor, pumps, and
valves should probably be ignored. In fact, there
may well be sections of the electronic controls that
can also be ignored, such as displays.
This approach implies rectifier isolation be- threshold because all MOVs in a residence are es-
tween power supplies for various parts of the de-
sign. Note that the increasing availability of dual
sentially in parallel, and the one with the lowest
voltage threshold is the “low bidder” during a
lightning strike and draws more energy.
5-/3.3-V logic parts, including microprocessors,
68
IEEE Industry Applications Magazine
■
November/December 2000

MOVs are applied downstream from fuses, so
the fuse characteristics become important. For
large impulsive transients, it is acceptable to blow
the fuse; but for smaller transients, fuse operation
is not necessary and annoying to the customer.
These transients typically last for less than a milli-
second, a time interval not covered by typical fuse
data sheets. Air core inductors can be used to tune
the MOV-fuse interaction.
ifications will reduce service and warranty costs
and increase customer satisfaction.
References
[
1] Electric Power Research Institute, Marek Samotyj,
Power-Quality Program Manager, +1 650 855 2000,
msamotyj@epri.com, 3412 Hillview Ave., Palo Alto, CA
9
4304.
[2] “An Assessment of Distribution System Power Quality,”
vol. 2, Electric Power Research Institute, Palo Alto, CA,
1
996. See also, D.D. Sabin, T E. Grebe, and A. Sundaram,
“Quality enhances reliability,” IEEE Spectrum, pp. 34-41,
Feb. 1996.
Conclusions
[
3] IEEE Recommended Practice for Monitoring Electric Power Qual-
ity, IEEE Std 1159, 1995.
[4] IEEE Recommended Practice for Powering and Grounding Sensi-
tive Electronic Equipment, IEEE Std 1100, 1992,
5] PQ Today Newsletter, Summer 1997. Available:
http://www.dranetz-bmi.com or call +1 800 372 6832.
6] For more information on sag generators, including tutorials,
see http://www.PowerStandards.com, or call +1 510 596
1784, fax +1 510 655 3902.
Electronic designs can often be inexpensively mod-
ified to tolerate unavoidable power-quality events.
The final design should be tested with disturbed
power, such as that provided by a sag generator [6].
Utility power-quality statistics, now available
from national studies, will allow economic justifi-
cation of these modifications and tests. These mod-
[
[
IEEE Industry Applications Magazine
■
November/December 2000
69