Вы находитесь на странице: 1из 7


Battery Charge Controller Characteristics

in Photovoltaic Systems
Steve Harrington
Ktech Corporation
James Dunlop
Florida Solar Energy Center

ABSTRACT intent of soliciting feedback on the information presented from

the battery industry.
Stand-alone photovoltaic (PV) systems are becoming
increasingly viable and cost-effective for remote and/or off- INTRODUCTION
utility grid power requirements. Thousands of PV systems are
being installed annually in the United States and abroad with Sandia National Laboratories (SNL) and the Florida Solar
typical applications including lighting, telecommunications, Energy Center (FSEC) are currently in the second year of an
and other battery charging requirements. In general, a stand- extensive program to evaluate battery charge controllers used
alone PV system consists of a PV array, which converts sunlight in stand-alone PV systems. The primary objective is to gain a
to direct-current electricity, energy storage in the form of better understanding of these battery charge controllers, their
secondary batteries, loads or appliances, and a control system, various algorithms, their voltage control set points, and how
which regulates battery charging and operation of the load. they charge and maintain lead-acid batteries.
While PV module development over the past decade has This report describes the first-year performance results for
resulted in a highly reliable product, the performance of battery eight different models of small (nominal 10 amp PV current)
and charge controller sub-systems have indicated a need for charge controllers under test. A description of the evaluation
improvement. These deficient areas include proper program and preliminary results of the testing have been
specification of charging requirements and control set points reported previously [ 1 1.
for the battery type and environment. A better understanding The controllers are being subjected to comprehensive tests
of how batteries (less than 220 Ah) operate in PV systems is including electrical characterizations at selected temperature
essential for the success of this market. Specifications extremes and operating conditions. After electrical
applicable to the low charge/discharge rates typical in PV characterizations, the charge controllers are divided into
systems need to be published in the battery technical literature. concurrent evaluation paths. One path consists of side-by-side
This paper presents typical strategies for battery charge operational systems tests in which the charge controllers are
regulation and load control used in stand alone PV systems. installed in identical stand-alone PV systems. The other path
Several charge algorithms (methods of controlling current to consists of continuous environmental and electrical cycling in
the battery) are presented, along with terminology used by the which the controllers are subjected to programmed electrical
PV industry for battery charge controllers. Information gained inputs, temperatures, and relative humidities. Characterization
from an extensive evaluation of commercially available charge of operating setpoints of all controllers is performed on a
controllers is discussed along with data collected from tests on periodic basis to detect changes in electrical performance. In
PV systems in the field. Finally, an overview of battery addition, selected custom tests are performed on identical
performance characteristics needed for the successful design models to determine response to current surges, installation
and long-term operation of PV systems is presented with the issues, and system compatibilities.
The evolution of battery charge controllers for PV systems
Presented at the 7th Annual Battery Conference on Applications and Advancer, California will help increase reliability of stand-alone systems in the field.
State University, Long Beach California, January 21-23, 1992
088518985192 $3 00 0 IEEE 1992 By presenting a description of charge controller algorithms,

this information can be used to understand the capabilities and batteries, a widely accepted temperature compensation
effects of different battery charge controller specifications. coefficient is - 5 mv/"C/cell[3]. If the electrolyte connection
These descriptions are universal, explaining various has been adjusted for local ambient temperature (increase in
algorithms, how to recognize them, and how they treat a specific gravity for cold environments, decrease in specific
battery. This program also intends to verify, with actual data, gravity for warm environments) and temperature variation of
what it takes to properly maintain lead-acid batteries in peak the batteries is minimal, compensation may not be critical
operating condition in small stand-alone PV systems. factor.

INTRODUCTION TO CHARGE CONTROLLERS Voltage Regulation Hysteresis (VRH): The voltage span or
difference between the VR set point and the voltage at which
The primary function of a charge controller in a stand-alone the full array current is reapplied. The greater this voltage span,
PV system is to protect the battery from overcharge and the longer the array current is interrupted from charging the
overdischarge. Any system that has unpredictable loads, user battery. If the VRH is too small, then the control element will
intervention, optimized or undersized battery storage (to oscillate, inducing noise and possibly harming the switching
minimize initial cost), or other characteristics that would allow element or any loads attached to the system. The VRH has
excessive battery overcharging or overdischarging requires a proven to be an important factor in determining the charging
charge controller and/or low voltage load disconnect or the effectiveness of a controller.
result will be a shortened battery lifetime and decreased load
availability. Low Voltage Disconnect (LVD):The voltage at which the load
Systems with small, predictable, and continuous loads may is disconnected from the battery to prevent overdischarge. The
be designed to operate without a battery charge controller. If LVD defines the actual allowable maximum depth-of-discharge
system designs incorporate oversized battery storage, and and available capacity of the battery. The available capacity
battery charging currents are limited to safe finishing charge must be carefully estimated in the PV system design and sizing
rates ( U 5 0 flooded or C/ 100 sealed) at an appropriate voltage process. Typically, the LVD does not need to be temperature
for the battery technology, a charge controller may not be compensated unless the batteries operate below 0°C on a
required in a PV system [2,3,4,5]. frequent basis. The proper LVD set point will maintain good
Proper operation of a charge controller should prevent battery health while providing the maximum available battery
overcharge or overdischarge of a battery regardless of the capacity to the system.
system sizingldesign and seasonal changes in the load profile
and operating temperatures. The algorithm of a battery charge Low Voltage Disconnect Hysteresis (LVDH):The voltage span
controller determines the effectiveness of battery charging and or difference between the LVD set point and the voltage at
PV array utilization, and ultimately the ability of the system which the load is reconnected to the battery. If the LVDH is
to meet the load demands. Additional features such as too small, the load may cycle on and off rapidly at low battery
temperature compensation, alarms, and special algorithms can state-of-charge (SOC), possibly damaging the load and/or
enhance the ability of a charge controller to maintain the health, controller. If the LVDH is too large, the load may remain off
maximize the capacity, and extend the lifetime of a battery. for extended periods until the array fully recharges the battery.
With a large LVDH, battery health may be improved due to
Basics of Charge Controller Theory reduced battery cycling, but with a reduction in load
While the specific control method and algorithm vary among availability. The proper LVDH selection for a given system
charge controllers, all have basic parameters and will depend on the battery chemistry and size, and PV and load
characteristics. Manufacturers' data generally provide the limits currents.
of controller application such as PV and load currents, operating
temperatures, losses, set points, and set point hysteresis values. CHARGE CONTROLLER ALGORITHMS
In some cases the set points may be dependent upon the
temperature of the battery and/or controller, and the magnitude Two basic methods exist for controlling or regulating the
of the battery current. A discussion of the four basic charge charging of a battery from a PV module or array-series and
controller set points follows: shunt regulation. While both of these methods can be effectively
used, each method may incorporate a number of variations that
Voltage Regulation Set Point (VR): The maximum voltage that alter basic performance and applicability. Following are
a controller allows the battery to reach. At this point a controller descriptions of the two basic methods and variations of these
will either discontinue battery charging or begin to regulate methods.
the amount of current delivered to the battery. Proper selection
of this set point depends on the specific battery chemistry and Shunt Controller
operating temperature. Temperature compensation of the VR A shunt controller regulates the charging of a battery by
set point is often incorporated in controller design, and is interrupting the PV current by short-circuiting the array. A
particularly desirable if battery temperature ranges exceed * blocking diode is required in series between the battery and
5°C at ambient temperatures (25°C). For flooded lead-acid the switching element to keep the battery from being shorted

16 IEEE AES MAGAZINE, August 1992

25 1 25 1

3 2 0 1-- Irradiance
1 -p. Irradiance -------
---- Battery0 ; =

r --- Battery (V) c

20-- ,, .............. PP VV (O4 -

Time of Day
Fig. 1. Daily Charge Profile for
a Shunt-Interrupting Controller

Irradiance ,_.-------
, ------- pv or)-
when the array is shunted. This controller typically requires a p.
20-- ---- Battery0 { s
large heat sink to dissipate power. Shunt-type controllers are
usually designed for applications with PV currents less than
20 amps due to high-current switching limitations.

Shunt-linear: This algorithm maintains the battery at a fixed

voltage by using a control element in parallel with the battery.
This control element turns on or closes when the VR set point
is reached, shunting power away from the battery in a linear
method (not on/off), maintaining a constant voltage at the
battery. This relatively simple controller design utilizes a Zener
power diode which is the limiting factor in cost and power
ratings of the charge controller.

Shunt-interrupting :This algorithm terminates battery charging

when the VR set point is reached by short-circuiting the PV
array. This algorithm has been referred to as “pulse charging”
due to the pulsing effect when reaching the finishing charge
state. This should not be confused with Pulse Width Modulation
(PWM). Figure 1 illustrates a typical daily charge profile for
a shunt interrupting controller in a system with a 24-hour
continuous load.

Series Controller
There are several variations of this type of controller, all o f
which use some type of control element in series between the
array and the battery.

Series-interrupting :This algorithm terminates battery charging

at the VR set point by open circuiting the PV array. A blocking
diode may or may not be required, depending on the switching
element design. Some series controllers may divert the array
power to a secondary load. Figure 2 illustrates a typical daily
charge profile for a series-interrupting controller in a system
with a 24-hour continuous load.

Series-interrupting, 2-step, constant current: Similar to the

series-interrupting, however when the VR set point is reached,
instead of totally interrupting the array current, a limited,
constant current remains applied to the battery.

IEEE AES MAGAZINE, August 1992 17

provides simple maintenance. As a by-product, when the sub- 103 percent of initial measured capacities after one year of
arrays are switched off charging, the power can be used for a testing at FSEC and 9 months at Sandia. Several batteries
secondary non-critical load. appeared to be sulfated. Some battery capacity recovery was
Series-Linear constant-voltage: This algorithm maintains the observed for the sulfated batteries after repeated controlled
battery voltage at the regulation set print (VR). The series charge-discharge cycles during the quarterly capacity
control element acts like a variable resistor to maintain the evaluations at FSEC.
battery at the VR set point. The series element dissipates the 25
balance of the power that is not used to charge the battery. The
current is inherently controlled by the series element and the 3 20 Irradiance -------
voltage drop across it. This is the recommended charge
algorithm for sealed, valve-regulated batteries. Figure 3
illustrates the typical daily charge profile of a series-linear,
constant-voltage controller in a system with a 24-hour
continuous load.
Series-Linear, constant-current-modified (EMAC): The Energy
Management and Control (EMAC) is a charge controller under 8 0
development at Sandia National Laboratory. This controller I
uses a multi-step charging algorithm with several advanced
features to enhance the amount of charge that will be accepted Time of Day
by the battery. This controller design will be available to the Fig. 4. Daily Charge Profile for a Series-Linear,
industry through the technology transfer program at Sandia Constant Current-Modified Controller
when development is completed.
The EMAC algorithm incorporates several features and steps Table 1. Description of Controllers Under Test
during the daily charging profile. The voltage regulation set at Sandia ((3 2S°C)
point is adjusted for the specific battery type being used. The
regulation set point is temperature compensated at - 5 mv/”C
system I VR I VRH I LVD ILVDH I ZG:,,I luoowll

cell and also compensated for the rate at which the battery is 1 114.361I 1.33 Il1.00 11.48 I No . -

being charged. This allows the battery to receive a full charge,

negating the effects of internal resistance, which produces an
artificially high battery voltage at the terminals during high
charge rates. When this regulation set point is reached, the
controller goes to a preset constant current charge rate to
facilitate controlled charging. As charging continues, the
battery voltage is then clamped to a specific voltage level
determined by the battery chemistry used. The charge rate will
Notes: system 8 has constant current ut 2 Amperes in R q u b t k n
return to constant current linearly as the battery voltage mode with a 15.0 volt maximum battecyvoltage limit
decreases. All of these features are optional and selectable with
M i a Hardware Description
a “personality module,” allowing the units to be easily factory
configured or modified in the field. Figure 4 illustrates a typical Table 1 defines the algorithms, set points, hysteresis, and
daily charge profile for the series linear, constant-current- options for the equipment tested at Sandia [ 6 ] .Figure 5 shows
modified controller in a system with a 24-hour continuous load. the battery capacity at the start of the testing at Sandia along
with capacity tests performed at 3, 4%, 6 and 9 months.
Batteries were not removed from the systems for the capacity
Systems Tests tests unless there were a minimum of three full-sun days prior
Laboratory controlled systems tests are being performed at to the test. This ensured all of the systems had more than enough
both Sandia and FSEC along with evaluations of systems in energy available to charge the battery. The initial drop in
the field to corroborate the data. Specific trends in performance capacity at three months was primarily due to low insolation
can be attributed to regulation algorithms, set points, hysteresis and cold temperatures during winter (Sandia tests began
values, and temperature compensation. Although in different November 1990). This three-month period magnified any
climates, similar trends have been seen at both the FSEC and problems an algorithm may have had in maintaining a high
Sandia evaluation sites. Following are discussions on specific battery state-of-charge (SOC) at lower temperatures. The
issues related to the systems tests. effects of temperature compensation, regulation set points and
Condition of batteries. A key indicator of the performance of hysteresis can be observed here.
a battery charge controller is the general condition and Figure 6 shows the history of battery water additions for each
maintenance requirements of the battery. Although there were of the systems at Sandiaover the same nine month period. As
no catastrophic battery failures at Sandia or FSEC during the battery temperatures increased in the summer months, typically
system tests, measured battery capacities ranged from 30 to 100-120”F, systems with temperature compensation


battery voltage to float up to the value required to connect the
array and return the system to normal operation.

Field Evaluations
Several sites with stand-alone PV lighting systems were
visited to perform evaluations and gain a better understanding
of charge controller operation in real systems. Each system was
completely evaluated to expose conditions that may have been
contributing to poor system performance. In general most
2 " c1 m c3 c4 Is w c7 #a systems were functional, however at a non-optimal
Battery performance. Typically, the systems evaluated had operated
Start 3 Months 4.5 Months for about a year and the batteries were found in generally poor
6 Months a9 Months condition due to lack of maintenance and improper charging.
Fig. 5. Nine Month Battery Capacities at Sandia Both design and hardware-related problems were found in most
40 systems. A few of the more typical problems related to charge
controllers are listed below [7,8,9].
A common problem found in field evaluations was a
mismatch of the charge controller VR set point and the battery.
8 = At one site where identical systems (and controllers) were
0 installed, voltage regulation set points varied from 13.4 to 15. I
S2O volts. Battery capacity tests proved the effects of the different
0 15
set points and showed the lower VR set point systems to have
a lower available battery capacity. Where captive-electrolyte
5 batteries were used, they have been typically overcharged due
0 to a high VR set point. The sealed batteries have less tolerance
#I #2 #3 #4 #5 #6 #7 #0
to overcharging because once excessive gassing has occurred
and electrolyte is lost, it cannot be replenished, hence reducing
3 Months @j6 Months 9 Months battery lifetime.
Fig. 6. Nine Month Battery Water Additions at Sandia Flooded batteries, both lead-antimony and lead-calcium,
were usually found to be undercharged and sulfated. This was
experienced a stabilization of water loss. Those systems without caused by the VR set point being adjusted too low. This results
temperature compensation experienced significant increases in in lower water requirements, although coupled with reduced
water loss. battery capacity and lifetime.
Both under and overcharge situations can be detrimental to Catalytic recombination caps have proven very effective in
the lifetime of a battery. As shown in Fig. 5, system # I is reducing battery water requirements to minimal levels. At one
suffering from severe undercharging and systems # 2 , 5 , 6 and site where half of the systems have recombination caps and
7 are somewhat undercharged. System #8 is overcharging the half do not, no water was required for the batteries with
battery, as shown by the water additions and specific gravity recombination caps over a six month period, while batteries
readings. The cumulative effects of these situations will be without these caps required approximately one-half gallon of
analyzed by an inspection of internal battery parts when the water per 6 - ~ o l t 220
, amp hour battery.
testing is completed. Load controls (timers and photocells) and LVDs have shown
Controllers without low-voltage load disconnects have shown some problems during the field evaluations. Several system
both recoverable and non-recoverable behavior when low- designs with oversized loads with respect to the PV array have
voltage battery conditions occurred. PV systems incorporating relied on the LVD to function as a load control. In these systems
controllers without an LVD are vulnerable to excessive the batteries remained at a low SOC almost continuously, resulting
overdischarge of the battery if either the load exceeds the design in a low load availability. The most trouble free lighting controls
conditions or during extended periods of below average sense array voltage or current to turn the lights on and off at dusk
insolation. Two systems without an LVD experienced and dawn. This eliminates the need to use photocells, which have
overdischarge of the batteries down to 1.5 volts per battery at been found to be unreliable in the field evaluations. All of the
both FSEC and Sandia. One controller allowed the PV array load timers evaluated functioned as expected. Several controllers
to remain connected to the battery without sufficient voltage are commercially available with light control/timersbuilt-in, thus
being applied to operate the controller. After a few days, the eliminating the need for additional components. A few controllers
PV array was able to recharge the battery without user were found with a non-functional LVD. One situation was
intervention. The controller in the other system required a encountered where the LVDH was too small, causing the load
battery voltage of six volts to connect the PV array, which to cycle on and off rapidly at low battery SOC. A small LVDH
without, did not allow the PV array to recharge the battery. may be desirable to increase load availability if battery lifetime
User intervention was required to disconnect the load and allow is a secondary consideration.

IEEE AES MAGAZINE, August 1992 19

Excessive voltage drops (in wiring, switches, fuses, etc.) charging rates. Ratings for minimum charging rates and
have shown to be detrimental to the operation of many of the information on their effect on battery cycle life are needed by
systems evaluated in the field. This has typically resulted in a the PV system designer.
loss of charging effectiveness of the PV array. Because the The discharge termination set point (LVD) should be
controller typically senses the battery voltage at the controller compensated for the low discharge rates found in PV systems.
for the regulation set point, any voltage drops reduce the actual Discharge curves for 20,50 and IO0 hour discharge rates would
charging voltage at the battery. This was a factor in most be more useful. The discharge termination set point (LVD) has
systems evaluated that had low battery capacity. Similarly, any been highly debated and data must be made available to
excessive voltage drops in the load circuit reduce available compensate for low discharge rates. Without this
voltage to the load, possibly reducing performance of the load. compensation, batteries are being overdischarged, resulting in
The use of improper wire sizes and connectors, lack of anti- shorter battery cycle life than expected.
oxidant grease on connections, and not using DC rated fuses Because PV systems typically function in temperature
and switches were the primary reasons for many voltage drop extremes, more published information regarding battery
problems. capacity, cycle life, set points (VR, LVD) and temperature
compensation requirements need to be published for the PV
There is little information available from battery
manufacturers regarding the use of small lead-acid batteries The numbers of stand-alone PV systems are increasing by
(typically a couple hundred amp-hours and less) at low the thousands every year. The success and acceptance of PV
discharge rates. This is due to the fact that manufacturers systems by the consumer is typically judged by the reliability
generally rate their batteries using constant voltage charging and performance of these systems. Performance of the batteries
methods with high initial currents to achieve the rated capacity. in these systems is a key factor in their success. To maintain
Also, high discharge rates are used to define the cutoff voltage a wide acceptance and increasing demand for these systems,
or LVD. The problem here is that high charge and discharge actual service life for the batteries used must be improved. A
rates are not typical in stand-alone PV systems. major problem in stand-alone PV systems is the inability of
The charge rate can affect the way an interrupting-type of the battery to live up to the expectations of the system designer
charge controller operates. If the array is oversized the high and user. The PV system designer currently does not have
charge current may bring battery voltage to a high level sufficient information on how these batteries function in a PV
prematurely, causing array current to be interpreted. The typical system.
and acceptable limits would be no greater than C/20 or C/15 Establishing specifications for batteries with “PV ratings”
for charging with this type of controller. Charge rates greater would be a positive step in achieving improved battery
than this are not typical in PV-only systems due to the high performance in PV systems. Although much of this information
cost of photovoltaic modules. Therefore, setpoints for voltage presently exists, it is not readily available or printed in product
regulation need to be established for various charging literature. Information such as minimum charging current
algorithms and charge rates. battery voltage regulation set points, discharge rated low
Batteries in most PV systems are charged with inexpensive voltage load disconnects, temperature compensated curves for
series or shunt-interrupting type controllers. Laboratory and low discharge rates, and suggestions for when electrolyte
field evaluations of systems have shown that, using the battery specific gravity needs to be adjusted are important parameters
manufacturer’s charge termination set points, these controllers in PV system design. Presentation of this information will give
cannot maintain an adequate battery SOC. These manufacturer PV system designers essential information when choosing a
charging set points were developed for constant-voltage type battery for a particular application.
charging, which has a much different current application to
the battery than experienced in PV systems. The result is ACKNOWLEDGEMENTS
shortened battery lifetime with reduced load availability due
to undercharging. Recommended charging set points for The authors would like to acknowledge several individuals
specific algorithms and battery types need to be published in who contributed to the work presented in this report. Ward
battery specification literature to aid the PV system designer. Bower of Sandia National Laboratories is responsible for
Because battery banks are sometimes oversized to management of the project. Credit is given to Craig Maytrott
compensate for several days without sunshine (autonomy and Leighton Demetrius of the Florida Solar Energy Center
period), there is a question regarding the minimum acceptable for their involvement with the systems tests, characterizations
charging current for the size of a battery. Thicker battery plates and field evaluations. Recognition is also given to Frank
are not always the answer. The batteries in these systems Schalles of EG&G for his work on the EMAC prototype charge
discharge well but have a hard time being completely recharged controller. Finally, credit is given to Joel Morton of New
at low rates. The larger batteries still do not reach the “rated” Mexico Highlands University for the development of test
cycle lifetimes and typically sulfate in systems with low software at Sandia.

20 IEEE AES MAGAZINE, August 1992

REFERENCES CG-D-5-81, United States Coast Guard,
NTIS, Springfield, VA, 1981.
[ I ] Bower, W., J . Dunlop and C. Maytrott, [6] Harrington, S.R.,
Performance of Battery Charge Controllers: An Interim Test Report,” “Charge Controller Testing Status ut SNL as of 6 Months,”
Proceedings of the 21st IEEE Photovoltaics Specialists Conference 1990, Sandia National Labs PV Design Assistance Center, June 1991.
Kissimmee, Florida, May 21-25, 1990.
[7] Harrington, S . R . ,
[2] Vinal, George W., “Photovoltaic Powered Stand-Alone Lighting Systems for Southern
Storage Batteries, California Edison Research Center.“
Fourth Edition, John Wiley & Sons, 1965. Sandia National Labs, April 1991. Internal report, not available.
[3] Kiehne, H.A., 181 Lane, C., Dunlop, J . and W. Bower,
Battery Technology Handbook, “Cecil Field Photovoltaic Svstems Evaluation,”
First Edition, Marcel Dekkar, Inc., 1989 Prepared for Sandia National Labs. August 1990. Internal report,
[4] Linden, D., not available.
Handbook of Batteries and Fuel Cells, [9] Hamrnond, B . and J . Dunlop,
McGraw-Hill, 1984. “Venice Photovoltaic Systems Evaluation, ”

[5] Allen, W.R. et al. Prepared for Sandia National Labs, July 1991. Internal report,
“Evaluation of Solar Photovoltaic Energy Storage for Aids not available.
to Navigation, ”

Steve Harrington’s work includes the investigation of all aspects of combining photovoltaic (PV) energy sources with loads, batteries, and electronic
power processing hardware. He is employed by Ktech Corporation who contracted in-house to Sandia National Laboratories’ Photovoltaic Systems
Research Division.
Mr. Harrington’s specific responsibilities include research into the various methods of charging battery storage systems, power processing hardware
(battery charge controllers), and batteries in PV powered small stand-alone systems. He is primary and co-author on several research papers regarding
the characterization and the effect of PV/battery charge controllers on batteries in PV systems.
He is also involved in the evaluation of stand-alone PV systems in the field, where he provides troubleshooting and technical information to users
to increase the understanding and reliability of those systems. He is familiar with all of the test and evaluation disciplines used in the photovoltaic
industry from solar cells, modules, power conditioning, and control equipment. He has been with the Photovoltaic Systems Research Division since
1990 and has been involved with photovoltaics since 1982.

James P. Dunlop is a senior research engineer at the Florida Solar Energy Center (FSEC) and has been involved with photovoltaic systems development
for 8 years. He has been primary and contributing author of numerous photovoltaic applicationidesign manuals and research papers. Mr. Dunlop
is principal investigator for the Southeast Regional Experimental Station (SERES) project, a multi-year photovoltaic systems research contract with
U.S. Dept. of Energy and Sandia National Laboratory. He is also responsible for a number of other photovoltaic programs at FSEC, including PV
applications development for the Florida Department of Transportation.

IEEE AES MAGAZINE. August 1992 21