Performance evaluation of commercially available stand-alone photovoltaic area lighting systems

Material Information

Performance evaluation of commercially available stand-alone photovoltaic area lighting systems
McNutt, Peter F
Publication Date:
Physical Description:
vii, 126 leaves : illustrations ; 29 cm


Subjects / Keywords:
Photovoltaic power systems -- Evaluation ( lcsh )
Lighting ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 125-126).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Electrical Engineering.
General Note:
Department of Electrical Engineering
Statement of Responsibility:
by Peter F. McNutt.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
34042625 ( OCLC )
LD1190.E54 1995m .M36 ( lcc )

Full Text
Peter F. McNutt
B.S., University of Colorado at Denver, 1991
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Electrical Engineering

This thesis for the Master of Science
degree by
Peter F. McNutt
has been approved for the
Graduate School
William R. Roemish
Miloje S. Radenkovic

McNutt, Peter F. (M.S., Electrical Engineering)
Performance Evaluation of Commercially Available Stand-Alone Photovoltaic
Lighting Systems
Thesis directed by Professor Pankaj K. Sen
Six commercially available, stand-alone photovoltaic (PV) area lighting systems
were installed at the National Renewable Energy Laboratory (NREL) in Golden,
Colorado, in 1993 for long-term independent evaluation and testing under real-world
conditions. This paper describes the systems, the testing that was conducted, the
experiences gained during installation, system safety issues, and the data gathered to
evaluate the systems' performance. This data includes battery and PV array voltages
and currents; temperatures of the PV array, the controller, the batteries, the ballast
and the lamp; solar irradiance; relative lamp illuminance; ambient temperature; and
other weather parameters. The results of this study are being used to evaluate nightly
system operation and long-term reliability under various solar and weather condi-
tions, as well as to intercompare different types of lighting systems. This ongoing
study is helping to evaluate the performance of these commercially available PV
lighting systems, information which will, in turn, help both the manufacturers and
users in the PV community.
This abstract accurately represents the content of the candidates thesis. I recom-
mend its publication.
Panfchj K. Sen

1. Introduction ................................................... 1
1.1 PV Area Lighting System Operation .............................. 4
2. The Lighting Systems Under Evaluation at NREL .................. 7
2.1 System Descriptions ............................................ 7
2.2 Testing Performed at NREL ............................................ 23
3. Lamps Used in PV Area Lighting Systems ......................... 32
3.1 Lamp Characteristics ................................................. 32
3.2 Compact Fluorescent Lamp Operating Methods ........................... 38
3.3 Ballasts ............................................................. 40
3.4 Effects of Lamp Housing and Pole Height on Illumination Levels 41
4. Mechanical Design Considerations ............................... 42
4.1 Wind Loading on PV Arrays ............................................ 42
4.2 Snow on PV Arrays .................................................... 42
4.3 The Battery Box ...................................................... 43
5. The PV Power System ............................................ 44
5.1 The PV Array.......................................................... 44
5.2 The Batteries ........................................................ 46
5.3 System Autonomy ...................................................... 48

5.4 The Controller....................................................... 49
6. Other PV Area Lighting System Considerations .................. 52
6.1 Documentation ....................................................... 52
6.2 Safety and the National Electric Code (NEC) ......................... 52
6.3 Installation Requirements ........................................... 52
6.4 Specifying and Procuring a PV Area Lighting System .................. 53
6.5 System Operation .................................................... 54
6.6 System Maintenance .................................................. 55
6.7 Vandalism and Theft ................................................. 55
6.8 Warranties and Manufacturer Support ................................. 55
6.9 Aesthetics .......................................................... 56
7. Results of Testing at NREL .................................... 57
7.1 Installation Experiences ............................................ 57
7.2 Potential System Problems ........................................... 57
7.3 Indoor Lamp and Ballast Bench Test .................................. 58
7.4 Outdoor I-V Curve Traces of the PV Arrays .......................... 64
7.5 Lamp Output and Run-time ............................................ 69
7.6 System Voltage Drops................................................. 84
7.7 Daily Systems Data ................................................. 85
7.8 Battery Maintenance ................................................ 100

7.9 Battery Capacity Test............................................ 100
8. Conclusions and Recommendations ................................. 123
9. References ...................................................... 125

At the University of Colorado at Denver I thank Dr. P.K. Sen for his project guid-
ance, and Dr. W.R. Roemish and Dr. M.S. Radenkovic for serving on my committee.
At NREL I thank Roger Taylor and John Thornton for their financial support and
project guidance; Dick DeBlasio, Laxmi Mrig, and Troy Strand for the use of the
outdoor photovoltaic test site and their technical expertise; Joe Burdick for his
project guidance; and Daryl Myers for providing weather data. I thank Chris
Thompson at Public Service Company of Colorado for arranging to have NREL test
a PV area lighting system, and Mike Thomas at Sandia National Laboratories for his
help, especially on the subject of batteries. Most of all I thank my wife, Tania, for
her patience while I finished my thesis.

As more organizations and individuals realize the benefits of photovoltaic (PV) area
lighting, they are approaching the National Renewable Energy Laboratory (NREL) in
Golden, Colorado, seeking information needed to make informed decisions when
purchasing these systems. This thesis addresses issues important to selecting and
purchasing, as well as designing, a PV area lighting system, including:
what a stand-alone PV area lighting system is and how it operates
testing of commercially available PV area lighting systems at NREL
lamps used in the context of these lighting systems
issues unique to lighting systems of this type
specifying and procuring PV area lighting systems
Since stand-alone PV area lighting systems produce their own power, the high costs
associated with extending overhead or underground power lines to remote or diffi-
cult-to-access locations is eliminated. With power line extensions costing from
$10,000 to $50,000 per mile [1] a $1,500 to $8,000 PV lighting system can be cost-
effective when located only a short distance from an existing power grid. Other
benefits include: elimination of utility bills; undisturbed landscaping, pavement, or
underground utilities; and the fact that stand-alone PV systems can be relocated
easily if necessary [2].
The following list briefly highlights some of the key issues that should be considered
before a PV area lighting system is designed or deployed. This list was compiled
from systems experiences at NREL, from manufacturers information, and from
conversations with experienced PV systems designers. Further information on each
issue can be found in this thesis.
Before purchasing a PV area lighting system, weigh the cost of a PV lighting
system vs. extending a utility line to a conventional lighting system at that loca-
Low-Pressure Sodium (LPS) Lamps
Low-pressure sodium (LPS) lamps are the most efficient lamps presently avail-
able, but colors are difficult to distinguish under their pure yellow light.

Compact Fluorescent (CF) Lamps

Cold temperatures, below 10C (50F), substantially reduce the light output of
compact fluorescent (CF) lamps. Air flow across them can also cool and reduce
their light output. CF lamps configured for rapid-start operation may have longer
service life, but instant-start operation reduces power consumption with no re-
duction in light output.
Wind Loading
Because of the tremendous forces high winds can exert upon the PV array, area
lighting system foundations need to be deeper than those of conventional lighting
Good documentation can mean the difference between a working and failing PV
lighting system.
Lead-Acid Batteries
High temperatures substantially shorten the life expectancy of lead-acid batteries.
Cold temperatures substantially reduce their usable capacity and necessitate
higher charging voltages. Excessively deep discharges may damage the batteries.
Serviceable batteries, compared to sealed-gelled or captive-electrolyte batteries,
are: less expensive, more forgiving of being overcharged (if serviced regularly),
more forgiving of operating at high temperatures (if serviced regularly), more
prone to giving off potentially explosive gases, and more prone to leaking or
spilling battery acid. Hydrocaps can reduce the service requirements of service-
able batteries.
A temperature-compensation sensor built into the charge controller, as opposed
to a remote sensor attached to the batteries themselves, may lead to undercharged
batteries in cold climates.

PV Arrays
High module temperatures decrease the operating voltage and power output of
crystalline-silicon modules. Arrays mounted at steeper tilt angles yield more en-
ergy during winter months and shed snow faster, but wind loading is greater.
The actual power draw of a particular lamp/ballast set can be substantially differ-
ent from the power rating of the lamp and should be verified by the system de-
signer. Electronic ballasts operating at high frequencies (10-20 kHz) yield higher
lamp and ballasts efficiencies. Ballasts operating below 15 kHz may produce
audible noise.
Lamp Housings
Lamp housings with clear, flat lenses, compared to domed, diffused lenses, will
yield higher illumination levels, but will cover a smaller area with light. The
lamp housing for a CF lamp should be sealed tightly enough to prevent wind
from cooling the lamp and decreasing its light output, especially in colder cli-

1.1 PV Area Lighting System Operation
Figure 1-1 is a simplified block diagram of a PV lighting system. The PV array
directly converts sunlight into electrical energy to charge the batteries during the day.
The batteries store the energy and then power the lamp during the night. The con-
troller maximizes battery life by protecting the batteries from being overcharged
during the day, and from being over discharged at night while the lamp is operating.
The controller may also serve to turn the lamp on and off automatically by sensing
the array output. Many controllers have built-in lamp timers to conserve battery
energy. The ballast is an electrical device that converts the low voltage from the
batteries to the high voltage required to operate LPS or CF lamps, two lamps com-
monly used in many PV area lighting systems.
Figure 1-1. Simplified block diagram of a typical PV area lighting system

Figure 1-2 shows the hours between sunset and sunrise at three different locations:
the equator, Miami (latitude 26 north), and Seattle (latitude 48 north). Away from
the equator winter nights represent worst case operating conditions for PV lighting
systems since there are fewer hours of sunlight for the array to charge the batteries
while at nighttime the lamp is operating longest.
Figure 1-2. Hours between sunset and sunrise for three locations winter nights are longer at
locations further from the equator [3]

Figure 1-3 shows the typical daily operation for one of the lighting systems under
test on a clear, sunny day. The battery voltage increases as it is charged by the array
during the day and then decreases as the lamp operates at night. An important
feature on a PV lighting system is system autonomy, which is defined as the number
of days the lamp must continue to operate if the PV array does not receive full


Figure 1-3. Typical daily operation on a clear, sunny day for System #3 on September 20-21,

The Lighting Systems Under Evaluation at NREL
This chapter describes the PV area lighting systems, the test site, the testing that was
conducted, and the data acquisition system.
2.1 System Descriptions
To gain a better understanding of commercially available stand-alone PV area
lighting systems, NREL purchased (but did not design) six systems in early 1993.
These systems were purchased from different manufacturers with various hardware
and lamp configurations, for evaluation and long-term testing under real-world
conditions. Figures 2-1 through 2-6 are photographs of the systems. Tables 2-1
through 2-6 summarize the features of the six systems under test at NREL. The
array, battery, and lamp information was supplied by the manufacturers. The data in
the controller and wiring tables was measured at NREL. Schematics for the six
systems can be seen in Figs. 2-7 through 2-12.

Figure 2-1. System #1
Figure 2-2. System #2


Figure 2-6. System #6

System # of PV PV Array Controller # of Battery Battery Hours Lamp Lamp Pole System
PV Array Material Tilt Type Batteries Capacity Type Autonomy Power Type Height Cost
Modules Output (Ah) (Watts) (ft) (early
(Watts) 1993)
1 3 180 m-Si 55 shunt 3 300 captive electrolyte 60 35 LPS 22 $2,871
lead acid
2 2 106 x-Si 48 series 3 270 lead acid 72 18 LPS 15 $2,438
constant voltage flooded CF
3 5 245 m-Si 52 4 440 lead acid 72 39 rapid 26 $5,233
with hydrocaps start
sealed CF
4 2 140 x-Si 44 shunt 2 230 flooded 60 36 instant 16 $2,078
lead acid start
sealed CF
5 3 144 x-Si 0 series 3 135 lead acid 72 24 rapid 18 $2,695
gel-cell start
sealed CF
6 2 106 x-Si 50 shunt 2 172 lead acid 48 18 rapid 15 $1,575
gel-cell start
m-Si = multicrystalline silicon
x-Si single crystalline silicon
CF compact fluorescent
LPS = low-pressure sodium
Table 2-1. General summary of the six commercially available PV area lighting systems under test at NREL(data provided by the

I PV Module PV Array
System Manufacturer Material P max Imax Vmax sc Voc Aperture Aperture Aperture # of P r max Imax I.C Aperture
and (W) (A) (V) (A) (V) Length Width Area Modules (W) (A) (A) Area
Model (m) (m) (m2) (m2)
Solarex MSX-60 m-Si 60 3.50 17.10 3.80 21.10 1.11 0.50 0.56 3 180 10.50 11.40 1.67
2 Siemens M55 x-Si 53 3.05 17.40 3.40 21.70 1.29 0.33 0.43 2 106 6.10 6.80 0.85
3 Duravolt DV50 m-Si 49 2.96 16.56 3.05 20.70 1.00 0.45 0.45 5 245 14.80 15.25 2.26
4 Solec S-70 x-Si 70 4.25 16.50 4.81 21.30 1.20 0.53 0.64 2 140 8.50 9.62 1.28
5 Siemens M75 x-Si 48 3.02 15.90 3.40 19.80 1.22 0.33 0.40 3 144 9.06 10.20 1.21
6 Siemens M55 x-Si 53 3.05 17.40 3.40 21.70 1.29 0.33 0.43 2 106 6.10 6.80 0.85
Aperture area measured inside frame
Table 2-2. Module and array specifications (supplied by manufacturers)

System Manufacturer and Model Type Nominal Voltage Amp-Hrs #Of Batteries Battery Bank Amp-Hrs
GNB Sunlyte sealed
1 captive electrolyte lead acid 12 100 3 300
2 Exide SP27 sealed gelled lead acid 12 90 3 270
3 Trojan T-105 serviceable flooded lead acid 6 220 4 440
with hydrocaps
A Delco sealed 12 115 o 230
4 2000 flooded lead acid
Sonnenschein sealed 12 75 o 150
O 8G24 gellled lead acid
MK sealed 12 87 o 174
o 27GEL gelled lead acid
Table 2-3. Battery specifications (supplied by manufacturers)

System Manufacturer Type LVD LVR VR VRR ^lamp-on ^lamp off Timer
and Model (V) (V) (V) (V) Setting
1 SunAmp PBRT-12-18A-15 Shunt 11.75 12.35 14.40 13.56 4.57 8.40 dip switches
2 Bobier PLC-1 Series 10.85 (temperature sensitive) 12.73 14.32 timed 0.95 8.10 trim pot
3 Polar Power SLC 12/20 ATEC Constant Voltage 11.36 12.75 14.74 constant voltage (lPv = 0A) (lPv > 0A) delay-to-on and lamp-run jumpers
4 SCI ASC-12/12-E and SCIT-8 (timer) Shunt (and Timer) 11.49 13.10 14.40 13.48 trim pot
Bobier 10.73
5 PLC- Series (temperature 12.44 14.37 timed 0.95 8.00 trim pot
(customized) sensitive)
6 SunAmp PVRT 12-15S-12 Shunt 11.71 12.32 14.16 13.32 4.49 8.24 dip switches
LVD = low-voltage disconnect
LVR = low-voltage reconnect
VR = regulating battery charging voltage (maximum)
VRR = regulating battery charging recovery voltage (minimum)
ViarTp-on = array threshold at which lamp comes on
Vian*M>ff = array threshold at which lamp goes off
Table 2-4. Controller specifications (measured at NREL)

Ballast Bulb Fixture
System Manufacturer Type Manufacturer Watts Type Manufacturer Type
and and and
Model Model Model
1 Bodine Tran-Bal 12LPS35E solid-state Osram Na35WA3Y1 35 LPS Western Lighting Lumastar 2 1835/LPS shoe-box directed
2 Bodine Tran-Bal 12LPS18E solid-state Osram Na18W A3z6 18 LPS Western Lighting Lumastar 1818LPS shoe-box directed
3 Bodine Tran-Bal 12R24-36E solid-state rapid-start Osram Dulux L36/39W RS/4100 39 CF Western Lighting OmegaSol 1824/36/FLU shoe-box directed
4 Gill Manufacturing 12BP4008 solid-state instant-start Osram Dulux 39W/4100K 36 CF Solec shoe-box directed
5 Thin-Lite 12BX27 solid-state rapid-start Philips PL-L 36W/41/4P 24 CF GE M-400A2 Powr/Door cobra head domed diffused
6 lota for bulbs Dulux L 36W or F39BX 39W solid-state rapid-start Philips PL-L 18W/41/4P 18 CF MagnaRay W1PL 18DC 12VDC 1.7A directed
Table 2-5. Lamp specifications (supplied by manufacturers)

Module to Module Array to Controller Controller to Battery Battery to Battery Controller to Lamp
System Size (AWG) Length (feet) Type (if exposed) Size (AWG) Length (feet) Size (AWG) Length (feet) Size (AWG) Length (feet) Size (AWG) Length (feet)
1 10 3 TC pair 10 25 10 3 10 1 10 25
2 12 1.5 inside LiquidTight conduit 12 20 12 2 6 2 12 20
3 10 1.5 sunlight- resistant TC pair 8 30 10 3 4 3 10 30
4 4 unmarked 16 20 16 3 16 2 16 20
5 not exposed 10 2 2 10
6 10 2 USE-2 10 15 10 2 8 2 12 15
Table 2-6. Wire sizes and approximate lengths

Western Lighting fixture
Lumastar 2 1835/LPS
Figure 2-7. Schematic of System #1

PV Module Siemens M55
Module 53W 3.05A 17.4V
+ +
*12 wire
#12 wire
Manufacturer Supplied Components
Crimp Splice
Terminal Block Connection
Screw Connection
m Nut or Bolt Connection
A Wire Nut Connection
Solder Splice
Plug Connector
T Switch
Added for Tut Purposes
|eiw\ Cunent Shunt

^ Connection to Oata Acq Sys
Western Lighting fixture
Lumastar 1818LPS

Charge/Light Controller
Bobier PLC-1
Batt+ Batt-
#12 wire
+ Ballast
. Tran-Bai
#12 wire
llamp shunt
10A <
#6 wire
2 T
+ Battery -
Exide SP27
12V. 90 Ah

6' gnd rod
12V, 270Ah
Figure 2-8. Schematic of System #2.

PV Module + PV Module + PV Module + PV Module + PV Module Duravott DV50 49W 3A 16.6V +
#10 wire TT. < I th-- U
#8 wire
/ --
#10 wire
Western Lighting fixture
OmegaSol 1824/36/FLU
+ Ballast 8odine Tran-Bal 12R24-36E ----- Lamp Osram Dulux L36/39W RS

Rapid-Start Fluorescent
Charge/Light Controller Polar Power SLC 12/20 ATEC
PV- Lamp-
Lamp+ Batt+ Batt- Gnd
#10 wire
I ' V
r----1------box connection
#4 wire
+ Battery - M + Battery " t
Trojan T-105
6V, 220Ah
12V, 440Ah
+ Battery + Battery
8' copper
ground rod
Figure 2-9. Schematic of System #3


#16 wire
\ /
Manufacturer Supoliad Comoonanta\
Crimp Splice
Terminal Block Connection
Screw Connection
Nut or Bolt Connection
A Wire Nut Connection
Solder Splice
m MoJcx Connector
iS* Fuse
T Switch
Added ter Test Purootes
W\* A n Current Shunt
ox_ V Connection lo Oats Acq. Syi.
Solec fixture
Charge Controller SCI ASC-12/12-E (shunt)
PV- Com
Batt+ Batt-
#16 wire
" \
m --
+ Ballast ----- m - Lamp
Gill Manu Osram Dulux
-12BP4008 - A 39W/4100K
#16 wire
Light Timer
Batt+ SCI SCIT-8
L- / PV+ Lamp-
1 1 1 t Lamp+ PV- Batt+ Batt-
-box connection
+ Battery - + Battery -
Delco 2000
12V, 100 Ah
8* gnd rod
12V, 200Ah
Figure 2-10. Schematic of System #4

Figure 2-11. Schematic of System #5

MagnaRay fixture
Figure 2-12. Schematic of System #6

2.2 Testing Performed at NREL
The six commercially available PV area lighting systems were installed at NRELs
outdoor PV test site in Golden, Colorado, and were attached to the data acquisition
system (DAS) between July and November, 1993. Table 2-7 lists the system,
weather, and DAS parameters monitored. Data is collected every five seconds and
averaged every hour. The lamp illuminance is used to measure the relative light
output of each individual lamp with respect to time. The photometers, used to
measure illuminance, are positioned 30 cm (1 ft) in front of the center of each light
fixtures lens.
Measurement Units
Array Voltage V
Battery Voltage V
Array Current A
Battery Current i A
Back-of-Module Temperature C
Controller Temperature C
Battery Temperature C
Ballast Temperature C
Lamp Temperature C
Plane-of-Array Irradiance W/m2
Lamp Illuminance* lux
Ambient Temperature C
Wind Speed m/s
Relative Humidity %
DAS Battery Voltage V
DAS Case Temperature C
* illuminance measurement used to determine the relative brightness of each lamp with respect to time
Table 2-7. System, weather, and DAS parameters being monitored

2.2.1 The Test Site
NRELs outdoor test site in Golden, Colorado, is located at latitude 39.74 north and
longitude 105.12 west, at an elevation of 1737 m (5700 ft). Figure 2-13 displays the
monthly expected irradiance for a PV array at a fixed tilt angle and horizontal along
with the expected monthly temperatures for this site. During the winter months,
when lamp run-times are longest, the output of an array at a 55 tilt array will
provide more energy to the batteries.
Figure 2-13. Expected solar energy falling on an array at two different tilts, along with ex-
pected ambient temperature (data for Boulder, Colorado, approximately 30
miles north of Golden) [4]

2.2.2 The Test Plan
The testing was divided up into four parts: installation, baseline tests, intermediate
tests, and final tests. The test plan was developed attempting to cover as many
important PV area lighting system aspects as possible.
The installation phase included:
taking an inventory of each system
installing the systems
connecting each system to the DAS
The baseline tests included:
calibrating the DAS
determining system losses
taking I-V curve traces on the arrays
noting and adjusting the array orientation
measuring the voltage set-points of the controllers
measuring the power and efficiency of each lamp and ballast
The intermediate tests included:
taking DAS measurements (Table 2-7) every 5 seconds and averaging and storing
the data every hour
inspecting the systems periodically
taking I-V curve traces on the arrays
performing manufacturer-recommended periodic maintenance
producing reports for NREL and UCD
The final tests included:
performing a final system inspection
taking I-V curve traces on the arrays
performing a battery capacity test
producing a final NREL report and a UCD thesis

2.2.3 The Data Acquisition System (DAS)
This section gives a description of the equipment that comprise the data acquisition
system (DAS). The DAS includes the data logger, voltage dividers, current shunts,
thermocouples, pyranometers, and photometers.
The Data Logger
The data from the six systems was collected by a single Campbell Scientific CR10
datalogger connected to three 32:2 multiplexers (Fig. 2-14). The CR10 is a pro-
grammable datalogger accessed by a 486 computer through an RS-232 interface.
The CR10 has enough built-in memory (64K RAM) to store six days and 23 hours of
data. The CR10 data is manually loaded to a SM-192 storage module that is capable
of storing 24 days and 12 hours of data. The data from the SM-192 is manually
loaded into a harddisk file. The power supply for the CR10 is an ac connected
rechargeable battery to minimize data loss due to ac power loss. Specifications of
the CR10 include [5]:
Analog Inputs Accuracy: 0.2% of full scale
Full Scale Range / Resolution: 2500 mV / 333 pV
250 mV/33.3 pV
25 mV/3.33 pV
Differential Input Voltage Noise
with 60 Hz Rejection:
Common Mode Range:
DC Common Mode Rejection:
Normal Mode Rejection
0.18 pV RMS
2.5 V
> 140 dB
(60 Hz, slow differential measurement):
Input Resistance:
Input Current:
70 dB
200 GQ
3 nA maximum

Figure 2-14. Simplified block diagram of the data acquisition system
DAS Input Voltage Dividers
Voltage dividers, approximately 20:1, made up of 1% wire wound resistors, were
used to limit the array and battery voltages to the 2.5 V range that the datalogger
can accept. The voltage dividers were calibrated using an HP 3456A 5-digit DVM
and an EDC 330 DC Reference Voltage Standard.
Current Shunts
System currents were measured using 1% current shunts. Full-scale reading for
these shunts is 50 mV.
Type-T Thermocouple Temperature Sensors
System temperature measurements were made with Type-T (copper vs. copper-
nickel) thermocouples working on the principle that when two dissimilar metals are
joined together they create a voltage dependent upon the junction temperature and
composition of the metals (the Seebeck voltage [6]). Thermocouples can be con-
nected directly to the CR10 which converts the input voltage to a temperature. The
temperature range of the thermocouples is -200C to 350C and the maximum error
is specified to be less than 1.5%. To accurately measure the voltage from the ther-
mocouples, the temperature of the connectors where the thermocouples enter the
DAS (another metal junction) must be known. This is accomplished by a 10TCRT
thermocouple reference sensor (from Campbell Scientific, designed specifically for
the CR10), a thermistor that is calibrated to measure the temperature of this interface.

The maximum thermistor error is 0.4C over the -33C to 48C range and its lineari-
zation error is less than 1.0C. The 10TCRT is designed to connect and be measured
directly by the CR10 datalogger to facilitate thermocouple measurements. The
results of a test performed to determine the accuracy of a thermocouple measured by
the CR10 and by an electronic thermometer (Table 2-8) indicate close agreement.
CR10 Electronic Difference
Reading Thermometer (C)
(C) (C)
1.1 1 0.10
25.5 25.4 0.10
29.8 29.8 0.00
Table 2-8. Comparison between CR10 and electronic thermometer temperature readings
Licor LI-200SA pyranometers were mounted to the edge of each systems PV array,
in the same plane-of-array, to measure the available solar irradiance. The LI-200SA
utilizes a silicon PV detector mounted in a fully cosine-corrected miniature head.
The LI-200SA pyranometers can be mounted in any plane without affecting their
performance. For clear, unobstructed daylight conditions, the LI-200SA compares
favorably with more expensive thermopile-type pyranometers. The pyranometers
spectral response (350 1100 nm) does not cover the full range of the solar spectrum,
but the error introduced is less than 5% under most conditions of natural daylight.
The pyranometers were calibrated at NREL against a thermopile pyranometer in
March, 1993. Specifications for the Licor pyranometers include [7]:
Absolute Calibration:
Response Time:
Temperature Dependence:
Cosine Correction:
Operating Temperature:
Relative Humidity
high stability silicon photovoltaic detector (blue
5% maximum, 3% typical
80 pA per 1000 W/m2 typical
maximum deviation of 1% up to 3000 W/m
<2% change over a 1 year period
10 ps
0.15% per C maximum (calibrated at 25C)
corrected up to 80 angle of incidence
-40 to 65C
Oto 100%

Licor LI-210SA photometers were positioned 30 cm (1 ft) in front of the middle of
each lamp housing to measure the relative light output of the lamp over the course of
each night and many months of testing. The LI-21 OS A photometer utilizes a filtered
silicon photodiode to provide a spectral response that matches the CIE curve within
5% under most light sources. The spectral responsivity curve of the standard
human eye at typical light levels is called the CIE Standard Observer Curve
(photopic curve), and covers the waveband of 380-770 nm. The human eye responds
differently to light of different colors and has maximum sensitivity to yellow and
green. In order to make accurate photometric measurements of various colors of
light or from differing types of light sources, a photometric sensors spectral respon-
sivity curve must match the CIE photopic curve very closely. Each LI-21 OSA
photometer connects to a calibrated connector that converts the current of the sensor
into a voltage. The outputs were calibrated for 10 pV per lux. The photometers and
their connectors were calibrated by Licor in March, 1993. Licor recommends
calibration every two years. Specifications for the LI-21 OSA photometer include [7]:
Absolute Calibration:
Response Time:
Temperature Dependence:
Cosine Correction:
Operating Temperature:
Relative Humidity
high stability silicon photovoltaic detector (blue enhanced)
5% traceable to NIST
20 pA per 100 klux typical
maximum deviation of 1% up to 100 klux
<2% change over a 1 year period
10 ps
0.15% per C maximum (calibrated at 25C)
corrected up to 80 angle of incidence
-40 to 65C
Oto 100%
At a temperature of -16C, the coldest measured nighttime temperature measured at
the NREL test site (Fig. 7-7), the maximum drop in the photometers output could be
6.15%. The measured drop in illuminance for the CF lamps was approximately 85%
on that particular night (November 24,1993).
Signal Grounds and Shields
The negative terminal of the battery bank was chosen to be the ground reference for
each system. All voltage and current instrumentation wiring was shielded twisted-
pair, the shielding being connected only on the CR10 end to prevent ground loops in

the shielding. The shielding is meant to minimize data errors due to electromagnetic
and radio-frequency interference.
DAS Calibrations
After the systems were connected to the DAS, the accuracy of the voltage and
current sensors was measured. The data in Table 2-9 indicates very close agreement
between the calibrators that were connected outdoors to each system and read
indoors by the CR10. A calibrator is essentially a high accuracy, programmable
voltage or current source.
Other Test Equipment
The following is a list of test equipment that was used for various tests along with
each instruments specified accuracy:
Fluke 87 True RMS DVM:
Fluke 80T-150U Temperature Probe:
Philips PM3394 200 MS/s Digital Oscilloscope:
EDC 330 DC Reference Voltage Standard:
EDC 3200C AC-DC Current Calibrator:
0.1% (dc)
Daystar portable I-V curve tracer:
Spire 240A Module Tester:
Outdoor HP Module Testing System:

Array Voltage Battery Voltage Array Current Battery Current
Calibrator Setting CR10 Reading % Difference Calibrator Setting CR10 Reading % Difference Calibrator Setting CR10 Reading % Difference Calibrator Setting CR10 Reading % Difference
0.00 0.00 0.00 0.00 0.00 0.00 -4.00 -3.97 -0.75%
1.00 0.99 -0.80% 2.00 1.99 -0.60% 1.00 0.98 -2.40% -3.00 -2.96 -1.33%
3.00 2.99 -0.27% 4.00 4.00 -0.02% 2.00 1.98 -1.00% -2.00 -1.97 -1.50%
6.00 5.99 -0.17% 8.00 8.00 0.01% 3.00 2.98 -0.67% -1.00 -0.98 -2.40%
9.00 9.00 -0.01% 9.00 9.00 -0.03% 4.00 3.97 -0.75% 0.00 0.00
10.00 10.00 -0.03% 10.00 10.00 0.00% 5.00 4.98 -0.40% 1.00 0.98 -2.40%
11.00 10.99 -0.09% 11.00 11.00 0.00% 6.00 5.97 -0.50% 2.00 1.98 -1.00%
12.00 11.99 -0.08% 12.00 11.99 -0.08% 7.00 6.95 -0.71% 3.00 2.97 -1.00%
13.00 12.99 -0.08% 13.00 13.00 0.00% 8.00 7.97 -0.38% 4.00 3.97 -0.75%
14.00 14.00 0.00% 14.00 13.99 -0.07% 9.00 8.96 -0.44% 5.00 4.98 -0.40%
15.00 14.99 -0.07% 15.00 15.00 0.00% 10.00 9.97 -0.30% 6.00 5.97 -0.50%
17.00 17.00 0.00% 16.00 16.00 0.00% 7.00 6.95 -0.71%
19.00 18.99 -0.05% 17.00 17.00 0.00% 8.00 7.97 -0.38%
21.00 21.00 0.00% 9.00 8.96 -0.44%
23.00 23.00 0.00% 10.00 9.97 -0.30%
Table 2-9. Verification of DAS voltage and current accuracy for System #3 using voltage and current calibrators.

3. Lamps Used in PV Area Lighting Systems
This chapter provides an overview of the lamps used in many PV area lighting
systems, as well as ballasts and lamp housings.
3.1 Lamp Characteristics
Figure 3-1 is a summary of characteristics of the most popular lamps used for area
lighting. LPS and CF lamps are presently used in many PV area lighting systems,
due to:
their high lumen/watt rating (unlike mercury vapor, halogen, and incandescent
their long service life (unlike halogen and incandescent lamps)
their availability in smaller lamp sizes (unlike metal halide lamps)
the availability of inexpensive DC ballasts (unlike high-pressure sodium lamps)
Low- High-
Pressure Pressure Metal Mercury
Sodium Fluorescent Sodium Halide Vapor Halogen Incandescent
250 i---------------------------.------------.-------------.------------.-------------.------------
j Lumens/Watt Color Rendering Index (%) Service Life Figure 3-1. Comparison of lamps used for area lighting (maximum values based on manufac-
turers' information |8], |9])

Marks on the LPS and CF Lumens/Watt columns in Fig. 3-1 indicate the specified
rating for the lamps under test at NREL. In general, the smaller the lamp size, the
lower its efficiency will be [10]. Efficiency is the ratio between the total light a lamp
produces and the power consumed by the lamp. High lamp efficiency is important
due to the expense of adding PV modules and batteries to a PV lighting system.
Color Rendition
Although not having as high an efficiency, nor as long a service life as LPS lamps,
CF lamps are popular due to their high color rendition index (CRI), indicated in the
middle columns of Fig. 3-1. The higher a lamps CRI (100% being the best), the
easier it is to distinguish colors under the light of that lamp. The CF lamps under test
at NREL have a CRI of 85% (data provided by the lamp manufacturer). The maxi-
mum specified CRI for LPS lamps is 20%. Under the light of LPS lamps all colors
are difficult to distinguish, appearing yellow, gray, or black. Figure 3-2 is the
spectral distribution (the colors that comprise the light) for the CF lamps under test
while Fig. 3-3 is the spectral distribution for LPS lamps. From Fig. 3-3 it is observed
that LPS lamps produce monochromatic, or pure yellow, light. This characteristic of
LPS lamps limits their use. For example, LPS lamps would be a poor choice at an
automobile dealership; however, astronomers prefer LPS lamps around observatories
since their light is easily filtered out.

Figure 3-2. Spectral distribution for a CF lamp with a CRI of 85% {11]
UV Blue
Visible Light
700 nm
Red IR
Figure 3-3. Spectral distribution for a LPS lamp with a CRI < 20% [11]

Temperature Effects on Lamp Illuminance
The light output of CF lamps is dependent upon the ambient temperature (Fig. 3-4).
As seen in the figure, the average monthly illuminance for the CF lamp under test at
NREL closely followed the ambient temperature, dropping to 50% of its maximum
monthly illuminance. The illuminance of the LPS lamp was not influenced by
ambient temperatures, dropping only about 10% during the test period. The illumi-
nance in Fig. 3-4 was normalized with respect to the maximum monthly output of
each lamp during the test period.
Figure 3-4. Temperature effect on lamp illuminance monthly averages for a CF and an LPS
lamp under test at NREL

The effect ambient temperatures have on CF lamps illuminance is documented by
lamp manufacturers (Fig. 3-5). The light output of CF lamps is determined by the
minimum bulb-wall temperature, usually the point farthest from the lamp base. As
shown in the figure, lamp orientation also affects the illuminance versus temperature
because the base, the warmest part of the lamp, may heat the rest of the lamp.
Figure 3-5. Data from one lighting manufacturer illustrate the relationship between ambient
temperature (as well as lamp orientation) and light output for compact fluorescent
lamps [12]

LPS and CF lamps generate light by discharging an electric arc through a gas mix-
ture. A certain amount of heat is required to maintain the arc brightness. LPS lamps
utilize an outer bulb over the inner, light producing bulb (Fig. 3-6). A vacuum
between the inner and outer bulbs in LPS lamps serves to insulate the inner bulb
from external ambient temperatures.
outer bulb
inner, light producing bulb
arc brightness dependent
upon bulb wall temperature
Figure 3-6. Low-pressure sodium lamp construction minimizes the effect of cold temperatures
on illuminance [10]
Although the operating brightness of LPS lamps is not effected by cold temperatures,
the amount of time required to reach full brightness will increase with colder tem-
peratures. Normally, an LPS lamp requires 7-15 minutes to reach full operating
At least one lamp manufacturer offers a CF lamp that is designed to maintain greater
than 90% rated light output in ambient temperatures from -5C to 54C (23F to
130F) [13]. This lamp type was not tested at NREL in the present study.

3.2 Compact Fluorescent Lamp Operating Methods
The CF lamps tested at NREL are configured for rapid-start or instant-start operation.
Cathode heaters, built into the base of most fluorescent lamps (Fig. 3-7) facilitate
lamp operation. With rapid-start operation, the cathode heaters operate during lamp
start-up, as well as during normal operation, consuming approximately 3 W of
power, continuously. With instant-start operation, a higher starting voltage is
applied to the lamp to facilitate arc ignition, eliminating the power drawn by the
cathode heaters.
Rapid Start Operation
4 Wires Base Lamp life = 10,000 Hours
1.5W ' 7
39W> 4W 32W | Cathode Heaters
I 1.5W i i
Instant Start Operation
2 Wires Base Lamp life = 7,500 Hours
Ballast Tr
36W > 4W 32W | Cathode Heaters not used
No decrease in light output
Figure 3-7. Two modes of CF lamp operation (10)

A bench-test determined that a CF lamp configured for rapid-start operation con-
sumed 39 W, but when the cathode heaters were disconnected the input power
dropped to 36 W, an 8% decrease. Furthermore, it was determined that there was no
decrease in the lamps illuminance (Fig. 3-8).
Light Output

cathode heaters operating
input power = 39 W
cathode heaters disconnected
input power = 36 W

_ ^lamp base temperature lamp bulb temperature. ooo llO IDO 90 '80
_ 1 '70
ballast temperature^ I 60 So 40
III 30 20
O 8 H O O UJ
* H w % s M


Figure 3-8. Results of bench-test demonstrating the effect cathode heater operation has on light
output of a CF lamp note the drop in lamp base (maximum) temperature, and the
lack of change in the lamp bulb (minimum) temperature. The spike in the light
output was caused by covering the photometer momentarily to indicate when the
cathode heaters were disconnected.
Due to the higher starting voltages required to reliably strike the arc of a CF lamp
configured for instant-start operation, the lamp life will be reduced [10]. A lighting
company representative stated lamp life is typically derated 25% [14]. So instead of
operating for 10,000 hours, lamp life may be reduced to 7,500 hours, representing a
decrease of approximately six months of 12-hour lamp operation.

3.3 Ballasts
The ballast converts the low voltage of the batteries to the high voltage (Table 7-1)
required to operate LPS or CF lamps used in many PV area lighting systems.
Electronic ballasts are different from inverters in that they typically operate at higher
frequencies for increased lamp and ballast efficiency [10]. One ballast operating at
9.6 kHz was found to emit an audible noise that might annoy people if it were to be
used indoors.
To properly size the array and batteries in any PV system, the power draw of the load
(the lamp and ballast in the case of PV area lighting systems) must be accurately
determined. Long-term monitoring of the systems at NREL (Fig. 3-9) revealed that
the lamps and ballasts consumed from 25% less to 45% more than the power rating
of the lamps. The design of the ballast determines the power delivered to the lamp,
as well as the power loss in the ballast itself. Although there are special ballasts that
are designed to start CF lamps at very cold temperatures (-20F), they may not
necessarily increase the light output, and may shorten lamp life due to higher starting
Figure 3-9. Measured lamp and ballast input power for two 18 W lamps under test at NREL

3.4 Effects of Lamp Housing and Pole Height on Illumination Levels
The lens and reflector of a lamp housing can influence a lamps light output. A
clear, flat lens will deliver higher lighting levels, but will not cover as wide an area
as a domed, diffused lens. A good reflector will increase the amount of light pro-
jected from the lamp. A well sealed lamp housing is important for CF lamps in cold
or windy locations as air blowing across the lamp can cool and decrease the light
output (Fig. 3-10). A short pole will position the lamp closer to the ground increas-
ing lighting levels, but ground coverage will not be as wide.
Figure 3-10. Results of bench-test demonstrating the effect blowing air has on light output of a
CF lamp note the drop in the lamp bulb (minimum) temperature. The spike in
the light output was caused by covering the photometer momentarily to indicate
when the fan began operating.

Mechanical Design Considerations
This chapter points out some of the major mechanical differences between PV and
conventional area lighting systems.
4.1 Wind Loading on PV Arrays
High winds can exert tremendous loads on an array mounted at the top of a tall pole.
A PV area lighting system should be able to withstand the maximum expected winds
for a particular site. Unless specified by the system manufacturer, it may be wise to
consult with an engineer to determine the proper foundation depth. At NREL, a
7.92-m (26-ft) tall system was installed with a 1.22-m (4-ft) deep foundation, the size
normally installed with a conventional lighting system of the same height. After
several high-wind days at the test site, the pole has begun to lean noticeably. It was
later learned that the minimum specified foundation depth is 1.52 m (5 ft). High
winds can also spin an array that is not adequately attached to the pole out of align-
ment so that the array no longer points south. At NREL, high winds spun two arrays
out of alignment. Some manufacturers specify the maximum winds their system is
designed to withstand. In general, the larger the array or steeper it is tilted, the
greater its wind load will be.
4.2 Snow on PV Arrays
Snow sitting on a PV array reduces the power it can produce. The steeper the array,
the quicker snow will clear. It has been observed at NREL on several occasions that
tilted modules mounted lengthwise vertically (Fig. 4-1) clear snow faster than those
mounted lengthwise horizontally. It is suspected that the greater frame width of the
lengthwise horizontally-mounted module catches and holds more sloughing snow,
but no controlled tests have been conducted to verify this.

, i 1
CL s o lule i

Figure 4-1. Tilted modules mounted lengthwise vertically (left) have been observed to clear
snow faster than those mounted lengthwise horizontally (formal controlled tests
have not been conducted to confirm this).
Note the tilt of the PV array has two opposite effects: steeper tilt more wind
loading, but easier snow shedding, and vice versa for a shallower tilt.
4.3 The Battery Box
Batteries are another unique feature of PV lighting systems. If serviceable batteries
are used, easy access for inspecting and filling the batteries is important. Batteries
that are difficult to access will require more time to perform the manufacturer-
recommended inspection and maintenance. The battery box should properly vent
potentially explosive gases that are produced as the batteries are charged. The
battery box should also be designed so that standing water on top of it, left by rain or
melted snow, will not fall onto the electronics mounted inside the enclosure when the
box is opened.

The PV Power System
This chapter provides information on the characteristics and operation of the photo-
voltaic power system: the PV array, the batteries, and the controller.
5.1 The PV Array
The PV array consists of PV modules, environmentally-encapsulated packages of
solar cells that are semiconductor devices that convert sunlight directly into electrical
energy. The short-circuit current (Isc) of a PV array, unlike a rotating machine, is
only about 15% above operating current (Fig. 5-1). The maximum amount of energy
a PV array produces depends primarily on the array size, the amount of sunlight, and
the efficiency of the modules. Commercially available crystalline-silicon and
multicrystalline silicon modules have aperture area (the area inside the frame)
efficiencies ranging typically from 8% to 12%, while the efficiency of amorphous
silicon modules typically range from 4% to 6%.
1000 W/M 2 5C
1000 W/W fa 4 rc -N
500V l/W [a 25

0 2 4 6 8 10 12 14 16 18 20 22
Figure 5-1. Typical current-voltage (I-V) characteristic curve for a crystalline silicon module
5.1.1 Tracking vs. Fixed-Tilt Array
Optimally an array would track, or point directly at, the sun throughout the day and
throughout the year. Such tracking adds expense and complexity to a PV system. In
the case of a PV lighting system, a tracker might not improve system autonomy

enough to be worth its expense. It may be more economical to add modules to a
fixed array.
The optimum tilt for a fixed-tilt array on a PV area lighting system (Fig. 5-2) is one
that maximizes winter battery charging when the hours of sunlight are fewest and the
hours of lamp operation are longest. At locations further from the equator the
optimum tilt becomes steeper (typically, latitude plus 10 to 15).
Figure 5-2. The optimum tilt for a fixed-tilt array on a PV area lighting system is one that
maximizes winter battery charging
5.1.2 PV Array Shading
Shading a small portion of a crystalline-silicon array by a tree branch or power line
can dramatically reduce its output because a PV module is constructed of many small
PV cells connected in series. If any one cell is shaded the modules output, and
hence the whole arrays output, can be reduced.
5.1.3 Solar vs. Magnetic South PV Array Orientation
A fixed PV array should face due solar south to receive maximum sun during the
day. It is important to realize the difference between solar and magnetic south. In
the continental United States, a compass pointing to magnetic south may be off by as

much as 22 from solar south. Installers using a magnetic compass should know the
declination for the system location and compensate for the difference between
magnetic and solar south. In extremely windy locations it may be necessary to
periodically check and adjust the array orientation.
5.1.4 Temperature Effects on PV Array Voltage
A PV array will produce its rated voltage, current, and power at Standard Test
Conditions (STC = 1000 W/m2, with a cell (not ambient) temperature of 25C).
Sunlight striking the solar cells that comprise each module elevates their temperature
above the ambient temperature. At elevated temperatures the operating voltage of a
crystalline-silicon PV array is reduced. For example, the voltage at the maximum
power point for the crystalline-silicon module in Fig. 5-1 operating at 25C is 17.4
V. However, at 47C the voltage drops approximately 2 V, which might be insuffi-
cient to charge 12 V batteries (the array voltage must be higher than the battery
voltage after voltage drops in the system wiring and the charge controller are sub-
tracted). PV area lighting systems in hot climates may require higher-voltage
modules, e.g., modules constructed with more PV cells, in order to properly charge
the batteries.
5.2 The Batteries
The batteries store the electrical energy generated by the PV array during the day and
provide the energy to run the lamp at night. Lead-acid batteries are used in many PV
systems due to their wide availability and low cost. Deep cycle batteries, the batter-
ies best suited for use in PV systems, are different from automobile batteries in that
they are designed to be discharged slowly and more deeply over long periods of time.
The batteries used in the PV lighting systems under test can be seen in Table 2-3.
5.2.1 Serviceable vs. Maintenance-Free Batteries
Lead-acid batteries may be serviceable or maintenance-free. Serviceable batteries
tend to be less expensive and may last longer if they are charged, discharged, and
serviced according to manufacturer specifications. Hydrocaps, which recombine into
water the gases normally given off while the batteries charge, may be installed on
serviceable batteries to reduce battery water loss and corrosion in the battery box, but
they will add to the system cost. Maintenance-free batteries may be desirable over
serviceable batteries in systems where the batteries are not easily accessible or will
not receive regular maintenance. Maintenance-free batteries tend to cost more and

are less tolerant of extreme heat. Maintenance-free batteries may be flooded, gelled,
or captive electrolyte. Gelled and captive electrolyte batteries will not leak or spill
acid electrolyte, making them easier to ship and handle.
5.2.2 Temperature Effects on Batteries
Hot temperatures reduce the life of lead-acid batteries (Fig. 5-3), and cold tempera-
tures reduce the energy that can be extracted from them (Fig. 5-4). An insulated
battery box may improve the performance of batteries operating in temperature
extremes. A battery box may be buried underground in areas that are not prone to
flooding, to better regulate temperatures. Another battery box option [16] consists of
a thermal reservoir to regulate the temperature of the batteries better than a foam-
insulated box, but costs more. A battery box that is exposed to sunlight should be
light colored to reflect the heat of the sun.
Figure 5-3. Hot temperatures reduce the life of lead-acid batteries [17].

Figure 5-4. Cold temperatures reduce the usable capacity of lead-acid batteries [18].
5.2.3 Battery Depth-of-Discharge (DOD)
The open-circuit voltage of a 12-V lead-acid battery when fully charged will be
about 13 V, but as current is withdrawn from the battery its voltage drops. The open-
circuit voltage of a fully discharged battery will be about 11.8 V (not 0 V). Fully
discharging lead-acid batteries may cause irreversible battery damage [19]. Current
PV systems design practice is to design for an 80% depth-of-discharge (DOD).
Some battery manufacturers state their lead-acid batteries, employing new materials,
may be repeatedly fully discharged with no harm. For a given load, a larger battery
bank will have a longer life, but will cost more and will require a larger enclosure.
5.3 System Autonomy
The autonomy, or the number of days the lamp can continue to operate without full
sunshine, of any PV lighting system depends primarily on the size and efficiency of
the array and the size of the battery bank, as well as the power draw of the load. The
array must be large enough to charge the batteries while there is full sun and the
battery bank must be large enough to store energy to run the lamp every night plus
have enough reserve to operate the lamp during several days of cloudy weather. As

more modules and batteries are added to a system its autonomy is increased, but so is
its cost. The ideal PV power system uses the minimum required number of modules
and batteries to maintain the lamp through the longest expected number of days of
bad weather. For this reason it is important that the system designer know the
weather patterns for a particular location and design the PV system for that location.
For example, although it seems that Miami, Florida, being almost 15 (latitude)
closer to the equator than Boulder, Colorado, might receive more solar energy, in
fact the opposite is true. Boulder receives an average of 5.3 kWh/m2 per day annu-
ally compared to 5.1 kWh/m per day in Miami on an annual basis because Boulder
has fewer cloudy days [4].
5.4 The Controller
5.4.1 Battery Charging
The controller extends the life of the batteries by protecting them from being over-
charged or over discharged, and may also turn the lamp on and off automatically.
Different types of batteries have different charging (voltage level) requirements.
Usually the maximum voltage that a PV array produces is greater than the recom-
mended charging voltage for the batteries. The controller limits the charge voltage
for the particular batteries in a system to extend battery life. Battery manufacturers
recommend temperature-compensated charging (cold batteries should be charged at
higher voltages than batteries at hot temperatures). Many charge controllers offer
this feature, but it will add some expense to the controller.

Four of the test systems have the temperature-compensation sensor built into the
controller. Figure 5-5 demonstrates the temperature difference between the batteries
and controller in one of the systems. The controllers average winter temperature
was more than 15C higher than that of the batteries, representing an almost 0.5 V
reduction in potential battery charge voltage during cold weather. Due to the large
mass of the batteries they are not able to heat up or cool down as quickly as the
controller. Had the temperature-compensation sensor been connected directly to the
batteries, instead of built into the controller, the batteries would have been charged to
a higher voltage at cold temperatures as recommended by the battery manufacturer.
Figure 5-5. Measured data indicates how a temperature-compensation sensor built into a
charge controller may not accurately sense battery temperature, which might
cause the batteries to remain undercharged in cold weather.
Typically, when battery manufacturers specify the charging voltage, they are assum-
ing the batteries are being charged after each discharge cycle. In PV systems, the
batteries may be discharged for several cycles (several cloudy days) before being
charged (on a sunny day). One battery manufacturer [20] recommends charging their

lead-acid batteries to 15.5 V (at 27C) when used in PV systems. However, in
applications where they will be charged after each discharge cycle, the recommended
charging voltage drops to 14.5 V.
5.4.2 Low-Voltage Disconnect (LVD)
Many controllers protect the batteries from being over discharged with the low-
voltage disconnect (LVD) feature. The LVD feature disconnects the lamp from the
batteries before they are discharged down to a voltage that might damage them and
reduce battery life. Although some battery manufacturers claim it is acceptable to
completely discharge their batteries, experienced PV system designers typically only
design the for only an 80% DOD. An LVD situation should rarely occur in a well-
designed system.
5.4.3 Lamp Operation
Many controllers govern lamp operation, i.e., detecting sunset and sunrise, by
sensing the array output. Timers may be built into the controller to limit the lamp
run-time, as well as delay how long after sunset the lamp comes on. Timers may be
set by jumpers, switches, or potentiometers (variable resistors). A motion detector
may be used so that the lamp comes on only while people are in the vicinity of the
lamp, allowing for a smaller and less-expensive battery bank.

Other PV Area Lighting System Considerations
This chapter provides information on other important PV area lighting system issues.
6.1 Documentation
PV area lighting systems are more complex than conventional lighting systems and
not as many people are familiar with them. Proper documentation will improve the
chances of successfully purchasing and deploying a PV area lighting system. The
documentation should include:
information on selecting and purchasing a system
an overview of system operation for those not familiar with PV systems
safety precautions for PV systems in general, and lighting systems in particular
a parts list, mechanical drawings, and schematics to aid installation
a maintenance schedule for checking and cleaning the batteries, lamp, modules,
fuses, etc.
a troubleshooting guide to aid in the event problems arise with the system
A good understanding of the PV area lighting system will minimize problems and
system downtime.
6.2 Safety and the National Electric Code (NEC)
As with most electrical systems potential safety hazards do exist with PV area
lighting systems. Care must be taken to design a safe and reliable PV system. The
National Electric Code (NEC) Article 690 covers PV power systems. Issues covered
include fuses, switches and circuit breakers, wiring, connectors, access to junction
boxes, grounding, markings on modules and arrays, and batteries [21]. A PV system
that meets NEC guidelines minimizes the risk of injury to personnel working on the
system and will ensure good electrical design practices. The NEC addresses nearly
all PV power installations, even stand-alone systems with voltages less than 50 V
[22]. In some applications PV systems may be required by local inspection authori-
ties to meet the NEC.
6.3 Installation Requirements
All PV area lighting systems will require a certain amount of assembly, but the
assembly should be relatively easy, requiring tools normally found in the field.

Installation requirements for different manufacturers systems will vary. As with
conventional lighting systems, installation may require several people, an auger for
digging the foundation hole, a bucket-truck or backhoe to lift the system into place,
wire cutters and strippers, fish-tape, assorted screwdrivers and wrenches. Additional
tools may include a magnetic compass, a voltmeter, and silicone caulking. Some
systems require no more work than plugging together several connectors, bolting
together several subassemblies, and then erecting the system on a pole. Other
systems may, in addition, require measuring, cutting, stripping, and fishing all
wiring, as well as drilling holes to mount components and feed wires. Although
connectors may allow a system to plug together quickly, undersized or improperly
installed connectors can lead to premature system failure.
The wiring sequence for some controllers is critical. They may overheat and fail
when connected into the PV system in the wrong sequence. It is important to care-
fully read and follow the system and controller manufacturers information before
wiring a system.
6.4 Specifying and Procuring a PV Area Lighting System
When buying a PV lighting system it is important that the user convey certain
specifications to the manufacturer. Some of these specifications include:
Type of lamp (e.g., fluorescent vs. low-pressure sodium)
Minimum required lighting levels
Type of operation (manual on-off, dusk-to-dawn, 24-hour, etc.)
Days of autonomy (the number of days the system must operate without full
Location (latitude and longitude)
Site specific information (micro-climate, in a canyon or on a flat field, large trees
or power lines immediately south of the system, etc.)
Will the system be required to meet NEC or local codes?
All of the above will effect the size of the PV array, the size of the battery bank, and
the system cost. Some manufacturers leave it up to the purchaser to buy key compo-
nents, especially the batteries or pole, as these items may cost more to ship than
customers would pay to obtain them locally. Therefore, it is important that the
purchaser ask what is included or excluded with a particular PV area lighting system.
If the purchaser will be maintaining the system, spare parts (lamp, fuses, controller)
should be purchased to minimize system down-time. The more inaccessible a

systems location or the more reliable it must be (e.g., an obstruction beacon on top
of a high voltage transmission tower near an airport) the more expensive a system
will be. The system supplier should be willing to help a buyer purchase the right
system for a particular application and site.
Persons buying PV area lighting systems may want to obtain a list of customers from
the manufacturer so they have the opportunity to talk to existing customers about
their satisfaction with that particular lighting system, as well as satisfaction with the
Where a large number of systems are to be purchased it might be possible to arrange
a two-phase procurement. The first phase would involve buying a small number of
the systems, installing them, and monitoring their operation for a short time. Assis-
tance with testing and evaluating the systems might be provided by an independent
laboratory such as NREL, Sandia National Laboratories, or Florida Solar Energy
Center. The second phase would involve buying the remainder of the systems based
upon favorable results from the first phase or working with the manufacturer to
correct any problems found before the final purchase is made.
6.5 System Operation
Before lamp operation, manufacturers may recommend conditioning the batteries
(charging them for several days before their first discharge cycle). The system
supplier or system documentation should provide information detailing how many
days the batteries should be charged by the PV array before the lamp is operated.
Proper battery' conditioning can lead to improved battery capacity and life. The lamp
timer settings should be set for the desired number of hours of lamp operation, but
should not exceed the maximum number of hours specified by the system manufac-
turer as this may decrease system autonomy (the number of days the system will
operate without full sunshine).
In October, 1993, a PV area lighting system (not one of the six test systems involved
in the present study) was installed in front of NRELs new SERF (Solar Energy
Research Facility) building for public display. In early December people passing the
system began to comment that the system was not operating. It turned out that the
light output of the CF lamp had decreased substantially due to cold weather. The CF
lamp was then replaced with an LPS lamp to increase the light output. Soon after
this, people began to comment that the system was not turning on in the evening.
The system was found to be operating as designed, but because the lamp came on
nearly one half-hour after conventional lighting systems in the area, it was perceived

to be malfunctioning. Furthermore, the turn-on threshold for the controller was not
field-adjustable. In general, PV area lighting systems may tend to come on later in
the evening and turn off earlier in the morning to reduce the array and battery size
(along with system cost). If it is important that a PV area lighting system turn on and
off at the same time as conventional lighting systems in the area it may then become
necessary to work with the system manufacturer to have the threshold of the control-
ler be adjustable.
6.6 System Maintenance
In addition to the maintenance required of conventional lighting systems (cleaning
dust build-up from the lamp and lens, replacing the bulb or ballast if they are not
operating), PV area lighting systems will require some additional periodic mainte-
nance. Battery maintenance includes cleaning corrosion from battery posts and
filling serviceable batteries. Hydrocaps, which recombine into water the gases
normally given off while the batteries are charged, installed on serviceable batteries
may reduce water loss and corrosion on the battery posts. The PV array should be
cleaned of dirt and bird dung with water and a squeegee, and its tilt and orientation
should be checked and adjusted if necessary. Fuses should be checked and replaced
if blown. All connections should be tightened periodically as vibration caused by the
wind may loosen them.
6.7 Vandalism and Theft
PV modules and batteries can be the targets of vandals and thieves due to their
expense and the fact that they are often located at remote sites. Several factors can
deter vandalism and theft. Padlocks and theft-resistant hardware can be used on a
battery box located at the bottom of the system. The battery box and modules may
be located at the top of the pole, out of the reach of thieves and vandals. A complaint
of some PV system owners is that the modules are used for target practice. The
modules and lamp lens may be constructed with unbreakable materials other than
glass. However, modules constructed with glass tend to carry longer power-output
warranties because glass does not scratch as easily as other materials.
6.8 Warranties and Manufacturer Support
Most PV area lighting systems will come with warranties to cover the individual
components, ranging from one year to over ten years, depending on the component.
A service contract may be purchased with the system to ensure extended manufac-
turer support of the system.

6.9 Aesthetics
Aesthetics is, of course, highly subjective. Comments made by visitors to NRELs
outdoor PV test site indicate that some of the PV area lighting systems are more
pleasing to look at than others. Some systems were liked by some people, but not by
others. It might be wise to obtain a picture of a system from the manufacturer so that
those concerned can determine if it is something they would find pleasing to look at
on a regular basis. In some situations the acceptance of a PV system may be based
upon its looks.

Results of Testing at NREL
This chapter summarizes the significant results and experiences encountered during
the testing of the six PV area lighting systems at NREL. These results helped the
local utility, Public Service Company of Colorado, gain experience and insights into
PV area lighting systems. At least one PV lighting system manufacturer utilized
NRELs findings to refine his systems design. These refinements led to a reduction
in material costs, manufacturing time, and field-installation time.
7.1 Installation Experiences
All six PV area lighting systems tested at NREL required a certain amount of
assembly. The installation of one system required no more than bolting together four
subassemblies, plugging together three cables that were already cut and strung, and
then erecting the system on a pole. Some of the systems came with ground rods,
eliminating a trip to the local electrical supply shop. The pole-base for one of the
systems was designed to fit a conventional lighting system foundation. One system
(Fig. 2-6) was installed on top of a 1455-kg (3200-lb) concrete block eliminating the
need to dig a hole and pour a concrete foundation (this also allows the system and
foundation to be moved easily in the future).
Some installation problems encountered included modules that did not fit their array
rack, battery-box bolt holes that did not line up with those on the pole, a wrong
controller, short wire length, cable feedthrough holes that were too narrow, wires that
fell out of connectors, and missing documentation. Poor documentation contributed
to at least a couple of installation problems: one controller overheated and failed due
to being connected into the system in the wrong sequence; and one controller was
thought to have failed, but it turned out that it was not jumpered properly.
7.2 Potential System Problems
The following potential problems were found in the six systems under test:
one controller with a mechanical relay is positioned just above the vent of a sealed
flooded battery vent
none of the battery-boxes has vents
three systems lack any overcurrent protection, but those that do use automotive
fuses that are not NEC approved for PV systems
the hinges of a locked battery-box could easily be removed with a wrench

undersized wires and connectors in System #4 overheated while the array charged
the batteries causing plastic and rubber insulation to brown and partially bum off
after several months of system operation
High winds between January and March, 1994, caused the array of System #4 to spin
out of alignment three times, 78 west of south the first time and 16 west of south the
next two times. The steel set screws may be cutting through the softer aluminum pole.
Winds in excess of 29 m/s (65 mph) caused the array of System #3 to spin 68 west of
south. As the array spun and caught more of the force of the wind, this caused the pole
on top of its undersized foundation to permanently lean 0.5 to the east, a noticeable tilt
on a 7.92-m (26-ft) tall pole.
7.3 Indoor Lamp and Ballast Bench Test
A test was conducted to determine the voltage, current, and power output to each
lamp from its ballast using a digital oscilloscope. The test circuit can be seen in Fig.
7-1. An attempt was made to determine what percentage of the input power was
consumed by the lamp, the ballast, and the cathode heaters (where applicable). In the
case of Systems #5 and #6 the scope was unable to accurately compute the power
due to complex waveforms. During testing, the ballast from System #1 became so
hot that it burned its impression into the wooden test bench, pointing out the impor-
tance of having ballasts properly heat-sinked. After this, all ballasts were placed on
an aluminum plate to dissipate any excess heat (as well as protect the work bench).
Table 7-1 contains the results of this test. This section contains some of the meas-
ured waveforms for LPS and CF lamps (Figs. 7-2 through 7-5).
Figure 7-1. Test circuit for the lamp and ballast bench test (CF lamp shown)

Lamp Cathode Heaters Ballast
System Lamp Ballast Input Power (at 12 V) Output Frequency (kHz) Open Circuit Voltage (V) Operating Voltage (Vm,) Power (Wn,) %of Input Power Operating Voltage (V) Power (W) both %of Input Power Power (W) Ballast Efficiency
1 Osram 35 W LPS Bodine Tran-Bal 41 19.2 348 85 34 83% n/a n/a 7.0 83%
2 Osram 18 W LPS Bodine Tran-Bal 24 9.6 329 47 19 79% n/a n/a 5.0 79%
3 Osram Dulux 39WCF Bodine Tran-Bal Rapid Start 41 21.2 356 97 31 76% 4.1 4.3 10% 5.7 86%
4 Osram Dulux 36WCF Gill Instant Start 32 15.2 523 100 29 91% n/a n/a 3.0 91%
5 Philips PL-L 36WCF Thin-Lite Rapid Start 21 18.5 105
6 Philips PL-L 18WCF lota Rapid Start 16 20.7 370 58
Table 7-1. Results of lamp and ballast bench test

Y/Div: Timebase: TRACE
50.0 V 10.0us ch3 AVG(16)
400mA 10.0us ch4 AVG(16)
50.0 W 10.0us m2.1 =ch3*ch4
Figure 7-2. System #1 ballast output voltage, current, and power to 35 W LPS lamp

Y/Div: Timebase:
50.0 V 10.0us
400mA 10.0us
50.0 W 10.0us
ch3 AVG(4096)
ch4 AVG(4096)
m2.1 =ch3*ch4
Figure 7-3. System #3 ballast output voltage, current, and power to 39 W rapid start CF lamp
(cathode heaters disabled)

Y/Div: Timebase: TRACE
5.00 V 10.0us ch3 AVG(4096)
400mA 10.0us ch4 AVG(4096)
2.00 W 10.0us m2.1 =ch3*ch4
Figure 7-4. System #3 ballast output voltage, current, and power to cathode heater of 39 W
rapid start CF lamp (lamp disabled)
The following calculations were performed to verify the scopes ability to calculate
rms readings in Fig. 7-4. These waveforms were chosen because they closely
approximate true sine waves.
Assuming the voltage waveform (Fig. 7-4) to be a perfect sine wave and assuming
Vpk = 6 V then
Vrms = vpk 0.7071 = 6 0.7071 = 4.243 V

Because the actual waveform (Fig. 7-4) is not a perfect sine wave its rms value will
be less, so 4.07 V is a reasonable reading.
Assuming the power waveform (Fig. 7-4) to be a perfect sine wave riding on a dc
offset power of 2 W and assuming Ppk = 4 W, and using the definition of root-mean-
square then
]{Pk +pPk sinxfdx
4 + 4 sinx)2 dx
= 2.45 W
Because the actual waveform (Fig. 7-4) is not a perfect sine wave its rms value will
be less, so 2.24 W is a reasonable reading.
In the case of Systems #5 and #6 (Fig. 7-5), complex waveforms led the scope to
inaccurately compute the output power as being greater than the measured dc input

Y/Div: Timebase: TRACE
100 V 10.0us ch3 AVG(4096)
200mA 10.0us ch4 AVG(4096)
Figure 7-5. System #6 ballast output voltage and current to an 18 W CF lamp
7.4 Outdoor I-V Curve Traces of the PV Arrays
The following data (Fig. 7-6 and Tables 7-2 through 7-7) was measured using a
Daystar portable I-V curve tracer connected to each array in situ. A module yields it
rated output when solar irradiance is 1000 W/m2 and the cell (not ambient) tempera-
ture is 25C. It is rare that both conditions occur simultaneously. It is important for
PV system designers to realize the difference between what a module is rated to
produce and what it actually produces under real-world operating conditions. The
measured voltage and power output of the six arrays under test at NREL did not yield
their rated outputs mainly due to either elevated cell temperatures or reduced irradi-

Figure 7-6. System #3 actual array I-V trace (44C, 894 W/m2) along with two normalized
3 m-Si modules E i Tilt = 55
Irrad. (W/m2) Temp, (deg C) Voc (V) i l*c FF (A) | (%) i ^max j ^max Fmax (V) i (A) (W) i Area Aper. tj (m2) (%)
Rated 10001 25.0 21.1 11.41 17.1 10.51 180.0 1.665!

Dec-14-94 1033! 26.8 20.2 10.7 70.7 15.6 9.8 153.1 1.665! 8.9%
12:23 PM (MST) j I '
i |
Mar-21-94 1025: 43.0 19.2 11.0 68.6 14.8 9.7 144.5 1.665 8.5%
11:35 AM (MST)
I i i
Jun-28-94 856 46.4 18.8 9.0 69.2 14.1 i 8.31 116.7 1.665! 8.2%
12:01 PM (MST) I i I ;
Table 7-2. System #1 summary of array I-V traces taken outdoors with Daystar portable I-V
curve tracer

2 x-Si modules i Tilt = 48
Irrad. Temp. (W/m2) | (deg C) Voc l.c FF (V) (A) I (%) V II P max 'max rmax (V) I (A) (W) Area i Aper. r| (m2) (%)
Rated 1000 25 21.7 6.8 17.4 6.1 106 0.802
Nov-19-93 1009 25.5 20.9 6.7 67.0 15.9 5.9 93.8 0.802 11.6%
11:07 PM (MST) i ! i I i
! ; ;
Mar-21-94 1045 33.5 20.2! 7.0 65.4 14.9 6.1 91.9 0.802 11.0%
11:23 AM (MST) ! i ; I
Jun-28-94 904! 44.0 19.3; 6.0^ 65.1 14.3 5.3! 74.9 0.802! 10.3%
11:45 AM (MST) ] ; i
Table 7-3. System #2 summary of array I-V traces taken outdoors with Daystar portable I-V
curve tracer
5 m-Si modules ; ; Tilt = 52
Irrad. (W/m2) Temp, (deg C) voc (V) l.c FF (A) | (%) I Vmax | l max (V) | (A) P max (W) Area (m2) Aper. r| (%)
Rated 1000i 25 20.7 15.25 16.56! 14.8 245 2.022
j ; ' i !
Nov-08-93 936.0 29.9 20.2 13.6! 70.7 15.4 12.5! 193.1 2.022 10.2%
1:21 PM (MST) ! i I
: i ; i
Mar-21-94 1068.0! 35.0 *19.71 15.1 68.8 14.9| 13.8 204.5 2.022! 9.5%
11:08 AM (MST) i i I !

Jun-28-94 894.0 43.9 19.0 12.9: 68.3 14.1: 11.9! 167.1 2.022 9.2%
11:55 AM (MST) I i i
Table 7-4. System #3 summary of array I-V traces taken outdoors with Daystar portable I-V
curve tracer

2 x-Si modules O Irrad. Temp. Voc l,c FF V l I P vmax I 'max i 'max Area Aper. t|
(W/m2) i (deg C) (V) (A) (%) (V) | (A) j (W) (m2) (%)
Rated 1000 25 21.3 9.62 16.5 8.5 140 1.272
Nov-19-93 1042 23.9 20.5 9.3 61.5 14.3: 8.2 117.5 1.272 8.9%
11:21 PM (MST) |
Mar-21-94 1056! 37.4 19.6 9.5j 57.1 13.3! 8.0 105.8 1.272 7.9%
11:15AM (MST) I ! I i i

Jun-28-94 926 44.5 18.8 8.2 56.0 12.6 6.9: 86.4 1.272 7.3%
11:30 AM (MST) i j I I I
Table 7-5. System #4 summary of array I-V traces taken outdoors with Daystar portable I-V
curve tracer

Due to its horizontal tilt, the array of System #5 is not rinsed by rain and moisture as
the tilted arrays are. As a result, dirt tends to build-up on the array. On one day, an
I-V trace was taken just before and just after the array had been cleaned (Table 7-6).
There was a 4.5% increase in the power output of the array after being cleaned.
3 x-Si modules - Tilt = D
Irrad. Temp. Voc LL. LL u (4 vmax I P 'max 'max ! Area Aper. ti
(W/m2) (deg C) (V) (A); (%) (V) (A) | (W) (m2) i (%)
Rated 1000 25 19.8 10.2 15.9 9.06! 144 1.208!
Dec-14-93 488 13.9 19.0 4.4! 66.2 15.2 3.7! 55.8 1.208 9.5%
12:03 PM (MST) | | 1
! I
Dec-14-93 483 12.9 19.0 4.5 68.6 14.9 3.9 58.3 1.208 10.0%
12:14 PM (MST) J ; i
Clean : I j
j ! I i i ; I I I
Mar-21-94 842 30.3 17.7 8.21 64.8 13.4 7.T 94.6 1.208 9.3%
11:51 AM (MST) i i
Jun-28-94 979 47.3 16.6 9.7 61.3 11.9 8.3! 99.3 1.208 8.4%
11:38 AM (MST) i i : l I
Table 7-6. System #5 summary of array I-V traces taken outdoors with Daystar portable I-V
curve tracer

2 x-Si modules i i Tilt = 50
Irrad. Temp. (W/m2) (deg C) Voc (V) l.o FF (A)! (%) V 1 P max | 'max i max (V) (A) i (W) Area (m2) Aper. r| (%)
Rated 1000| 25 21.7 6.8 17.4 6.1! 106 0.802!
Nov-21-93 1050! 35.3 20.4 7.01 67.8 15.8 6.1 96.4 0.802! 11.4%
11:40 PM (MST) i I i i
! i ! j i j
Mar-21-94 1050 43.0 20.1 7.0? 67.3 15.4 6.21 95.2 0.802! 11.3%
11:29 AM (MST) ! i ; i i
j ! !
Jun-28-94 900 51.0 19.4 5.9 67.5 15.0 5.1|77.1 0.802! 10.7%
11:50 AM (MST) ! ! I ; i j
Table 7-7. System #6 summary of array I-V traces taken outdoors with Daystar portable I-V
curve tracer
7.5 Lamp Output and Run-time
People buying any area lighting system are primarily interested with the light it
produces and its reliable operation. Although the amount of light the LPS lamps
produce is the highest, their monochromatic (pure yellow) light limits their use to
applications where distinguishing colors is not important. On the other hand, the CF
lamps tested produce a light that facilitates distinguishing colors, but their light
output (Figs. 7-10 through 7-13) is highly dependent upon the ambient temperature
(Fig. 7-7). The illuminance of the LPS lamps changed relatively little (Figs. 7-8 and
7-9) and did not trend with the temperature.
The reliability of the systems was varied:
System #1 (Fig. 7-14) experienced random short run-times due to a faulty
controller that was replaced in May 1994
System #2 (Fig. 7-15) ran almost all night, every night, missing only one nights
operation during the winter
System #3 (Fig. 7-16) (the most expensive) ran almost all night, every night

System #4 (Fig. 7-17) may have performed better had the manufacturer used
larger wire and better connectors (a 2-V drop was measured between where the
array voltage enters the battery-box and at the batteries (Table 7-8)). It missed
only one nights operation during the winter months, and in May, 1994, the bal-
last failed.
System #5 (Fig. 7-18) experienced shortened run-times and missed 3 nights
during the winter months due to the horizontal array
System #6 (Fig. 7-19) in theory should have performed better, but from the
results of the battery capacity test (Table 7-10) it seems as though the usable ca-
pacity of the batteries is low. In late October, operator error caused its control-
ler to fail
The system availability was defined to be the actual lamp run-time divided by the
estimated all-night lamp run-time. The estimated run-time was found by shifting the
computed curve for the hours between sunset and sunrise [3] to approximately fit the
actual run-time and then summing the hours of this shifted curve. Gaps in the actual
lamp run-times due to controller or ballast failures were disregarded since spare parts
could have eliminated such system down-times. Lamp run-times were determined
by accumulating the time the battery current was negative (current flowing out of the
batteries) while the solar irradiance was nearly zero.
It is important to look at a particular application before passing judgment on a
systems performance. For example, the operation of System #5 (Fig. 7-18) might be
acceptable at a summer campground where the lamp would not be needed in the
winter months, thus the need for a more reliable and more expensive system may be
avoided [23].

o _X ro O M O _k IO o _k tO o M o ro o NJ o M o V ro o V ro o
o> O) 05 o> cn O) cn cn cn cn cn cn .t* CO CO CO cn cn cn -t- CO
i> > > CO CO CO 6 6 6 z z z 6 6 6 cl_ c_ C TI TI T| 2 2 2 i> > > 2 2 2 c_
c c c (D <0 (D o o O o o o 0) 0) ti) (D CD CD 0) 0) 0) TJ T) *o £=
CO cu CO "D JD T5 < < < O O O O' O' O' 7 7 *< *< *< 3
CO CO CO CO cb cb CO CO CO cb cb cb cb cb cb CO CO CO cb cb cb cb cb cb CO CO CO l 1 cb
Figure 7-7. Measured average daily temperature at NREL's outdoor test site

Figure 7-8. System #1 35 W LPS lamp average nightly illuminance

Figure 7-9. System #2 18 W LPS lamp average nightly illuminance

Figure 7-10. System #3 39 W CF lamp average nightly illuminance

Figure 7-11. System #4 36 W CF lamp average nightly illuminance

Figure 7-12. System #5 24 W CF lamp (with domed diffused lens) average nightly illuminance

Figure 7-13. System #6 18 W CF lamp average nightly illuminance

Hours between sunset and sunrise = 4019 ^
Estimated all-night lamp run-time = 2762 hours
-H N> IH o h-H IH N> 111 o H-H 1 N) -M o H-H _k M o N> o ro O N)
00 00 -vl -sj -nI O) CO O) O) CO O) cn cn cn cn cn cn &. -0.
c!_ c_ > > > CO cn CO 6 6 6 z z z n n n cl_ cl_ cL
c c c CD CD 0) Q) 0) 0)
t CO CO 1 CO CO CQ CO CO "O O O f t r* < < < O o O 3 3
CO CO CO CO CO CO cp cp cp CO CO CO CO CO CO CO cb cb

ho o N> o ro o _k ru
CO cn cn cn u -0.
n k k k > i> > k s k
CD 0) 0) 0) T3 "O T3 0) 0) 0)
O 7 7 i i i *< *< *<
CO CO &. to CO A CO CO CO -fr. t CO 1 CO i CO
Figure 7-14. System#! lamp run-time

_fc. N) o k hO o hO o ho o _x ho o N) o all ro O hO o ho o V hO o k N) o
00 00 -nJ O) cn 03 O) 03 03 cn cn cn cn cn cn CO CO CO cn cn cn a -fc. b. -fc. CO
c_ c_ > > > in in in 6 6 6 z z z 6 6 n cL cL C T| T| Tl i: s > > > c_
c c c (I) CD (D o o o o o o n> CD (13 Q) 0) CD CD CD 0) Q) 0) T3 O T3 Q) c
1 1 CO CO CO "O u TO 1 1 < < < o o O O" O' CT i \ -n ~i i < 3
00 CO cb cb cb cb cb cb CO CO CO cb cb cb cb cb CO CO CO CO cb cb cb cb cb cb CO CO CO 1 1 1 CO
CO CO CO CO CO CO CO CO CO CO CO CO 4^. -fc. 4^ -U
Figure 7-15. System #2 lamp run-time

-X to o ->
00 00 S Nj
£r <=r > >
E. EL c c
CO CD i i
o M C) N) o _A N) o ro o IO o X to o _x to O to O ro
CD cd O) CD CD CD cn cn cn cn cn cn b. -b b CO CO CO cn cn cn 4k 4^ 4k 4k 4k 4k
(n GO clo 6 6 6 z z Z n n n cL cL cL n Tl *n k k k i > > k k k
(T> CD o> o o o O o o n> CD CD CD CD CD 0) 0) 0) o "O "O Q) 0) 0)
T3 "O T3 < < < O O O O' O O 7 i i V 1 *< *< *<
Figure 7-16. System #3 lamp run-time

Figure 7-17. System #4 lamp run-time

UlflUfj ufiuyi
O M o M o K) O N> o _k,
vl *^l CJ) CD CD CD O) CD cn cn cn cn cn
> > :> CO CO CO 6 6 6 z 2 z a 6
C c c CD o O n O o o (D (D
CQ CQ (O D u KJ f 1 r 1 1 < < < O o
r*o o M O M o N> o _L N> O V K> O
cn u CO CO CO cn cn cn -fc. CO
6 (D O c_ 0) 3 C_ 0) 3 c_ 0) 3 T1 CD O" Tl a> cr Tl CD cr k 0) T k Q} k 03 :> TJ t > "a i > T3 -i k 0) *< k 0) k 0) c_ c 3
cb CO CD CD CD fc. CD cb cb 4^ CD A cb cb CD CD CD 4* 1 CD 4w 1 CD 4*. 1 CD -U CD 4*
Figure 7-18. System #5 lamp run-time

ity > 85%
_k M o _k N) O M
00 00 -vj -vj O) o> O)
c_ c_ > > > CO CO CO
c c c c C CD (D CD
1 CD CO 1 CO CO CO CO CO *o ~o O
o o o
|" '|' "i" ^ ^ O O o -j- y y w w
o fvj o M O N> o V M O
CO CO CO cn Ol cn *. -fck CO
c_ 0) 3 n CD cr n (T> cr n CD cr k 0) 7 £ 0) 1 k 0) -s i> O S i > T3 i > TJ 7 0) *< k 0) *< k 0) cl_ c 3
CD CD CD CO CO 4!k CD CD CD 4* CD CD -U 1 CD fc. 1 CD 1 CD 4* CD 4*>
Figure 7-19. System #6 lamp run-time

7.6 System Voltage Drops
Tests were conducted to determine the voltage drops through the wiring, connectors,
and electronics of each system while the batteries were being charged during the day
(Table 7-8) and discharged during the night (Table 7-9). Voltages were measured
only inside the enclosures because of the difficulty in trying to reach and measure
voltages from the arrays and lamps at the tops of the poles. System #4 had the
highest voltage drops due to undersized wire and connectors, greatly reducing its
ability to charge its batteries.
System Irradiance (W/m2) Vpv at enclosure entrance Vpv at controller Vbatt at controller Vbatt at batteries Total Voltage Drop
1 900 13.08 13.03 12.56 12.45 -5%
2 1000 12.73 12.62 12.58 12.40 -3%
3 900 13.60 13.33 12.55 12.34 -9%
4 1000 14.90 13.72 13.25 12.85 -14%
5 1000 13.55 13.37 13.06 12.85 -5%
Table 7-8. Daytime system voltage drops during battery charging (June 19,1994)
System Vbatt a* Batteries ^batt 3t | V|amp at Controller! Controller V|amp 3t Enclosure Exit Total Voltage Drop
1 12.60 12.55 12.49 12.48 -1%
2 12.74 12.70 12.68 12.65 -1%
3 12.65 12.61 12.54 12.50 -1%
4 12.35 12.23! 12.12 12.03 -3%
5 12.88 12.831 12.76 12.73 -1%
6 12.59 12.581 12.55 12.51 -1%
Table 7-9. Nighttime system voltage drops during lamp operation (April 20,1994)

7.7 Daily Systems Data
Included in this section is the available solar energy, battery voltage, and battery
current balance for the six systems on a daily basis.
The available solar energy is shown for two systems (Figs. 7-20 and 7-21), System
#3 at an array tilt of 52 and for System #5 with a horizontal array. Note the differ-
ence in the available energy between the two tilts. In the winter, the horizontal tilt
receives less energy and in the summer it receives more. Generally, the more avail-
able solar energy an array has, the better able it is to charge the batteries.
The six battery voltage graphs (Figs. 7-22 through 7-27) represent on a daily basis
the maximum, average (the dark line), and minimum voltages. The maximum
voltage represents the highest array charging voltage, the minimum represents the
lowest battery voltage while the lamp was running, and the average voltage can be
thought of as an indicator of overall battery charge. Generally, the voltage of a 100%
charged lead acid 12 V battery is around 13 V, whereas when the voltage drops down
to 11.8 V, it is near a 0% state-of-charge.
The six battery current balance graphs (Figs. 7-28 through 7-33) represent, on a daily
basis, the net difference between the amp-hours delivered to the batteries by the array
and the amp-hours withdrawn by the lamp. Generally, more reliable PV systems will
keep the batteries at a positive balance on more days.

00 OO o -s| CD CD o> CD
<_ c_ > > > CO CO CO 6
c c c c 1 CO CO CO CD CD CD CD CD
_k ro O -Jt N) o _x ro
6 6 Z 2 Z 6 6 6
o o o o O CD CD a>
1 1 < < < O O o
o ro O _k ro o
4* -fe. CO CO CO CD CD
cL 0) c_ 0) cL 0) T1 CD *T1 CD *T| CD k 0) 1 k CD i
=3 3 3 CT cr o*
4^ 4*. 4^ 4*
ro o _v ro O V ro o
CD -U 4* 4*> -fc. 4^ CO
k > > > k k k cL
0) 1 T3 s i T3 -i i a t t 0) *< 0) 03 < c 3
CD 4*> CD 4^ CD 4* CO 4> i CD 1 CD 4^ 1 CD 4^ CD -U
Figure 7-20. System #3 daily available plane-of-array solar energy (array at a 52 degree tilt)

tO o fO O to
00 00 -O cL cL > > > CO CO CO
c c c c C CD CD CD
1 CD CO 1 CO CO (O CO (O o TD "O
o to o to O
o> G) O) Ol cn cn cn
6 6 6 z z z 6 6
o o o o o o CD CD
1 1 1 < < < o o
CD CD CD 1 CD 1 CD 1 CD t CD 1 CD
to O to O to O
cn 4s 4s 4s CO CO CO cn
6 CD cL 0) cL 0) c_ CD T1 (D "Tl CD Tl CD k 0)
O 3 i 3 i 3 i O' O O
CO A- 4s 4s. 4s 4s
to o _k to O _k to O
cn cn 4s 4s 4s 4s. 4s. 4s CO
k k > > > k k k cL
0) 0) *o Q "D CD 0) 0) c
7 7 i t *< 3
CD 4s CD 4s CD 4s CD 4s CD 4s. i CD 4s. 1 CO 4s 1 CD 4s CD 4s
Figure 7-21. System #5 daily available plane-of-array solar energy (horizontal array)

a ro o M o _k ro o _k N) o M o _k N> o N> o NJ o _k M o M o N) o
00 00 "vj -vl o> cn o> O) O) o> cn cn cn cn cn cn 4. 4^ CO CO CO cn cn cn 4>- 4>. 4* 4^ 4^ 4^ CO
c_ c_ > > CO CO CO 6 6 6 z 2 2 n 6 6 c!_ c_ c_ "Tl Tl "Tl s s s > > > 2 2 c_
c c c CD 0) a) O o O o o O (!) CD CD (D CD a> 0) 0) Q> T3 D T3 0) 0) c
l_ CO CO CO O O a < < < O o o cr O cr i *7 -i -i *< *< *< 3
A CO cb CO CO cb cb CO CO CO cb cb cb cb cb cb CO CO CO cb cb cb cb CO cb CO CO CO 1 1 1 CO
4*. 4. 4^ -b. -k 4^ co 00 00 CO CO CO 4*. k 4* 4^. -N 4.
Figure 7-22. System #1 daily battery voltage (maximum, average, and minimum)

1 -
1 -
1 _
J . J L 1

o hJ o N) o N> o N) o
-*4 **4 cn cn cn O) cn cn Ul cn cn cn cn
oo oo
T T"
c c c a> a> CD o o o o o o a> CO CO CO "O TJ T3 1 1 rt 1 1 < < < o o o
cb cb cb cb cb cb CO CO CO cb cb cb cb cb cb

1 i t - - -t - -1- i - - t- 1 - t" - t
o M o N) o ro o x M o I 1 1 K> r~~t o
CO CO CO cn cn cn -fc>. CO
cL 0) 3 cL 0) 3 cL 0) 3 T1 0) CT T1 CD O' "T| CD O' k 0) k 0) 1 k 0) i :> "O -a V > ~a 7 k 0) *< k 0) k 0) *< cL c 3
CD -U CO CO cb cb cb cb cb cb -b* CO 4^ CO CO CO CO CO CO
Figure 7-23. System #2 daily battery voltage (maximum, average, and minimum)

CO CO i i i i
^^^cbcocbcbcbcbtP Ca) OJ OJ ~ N K fv k *. > rvrwPv
cu vl/ w vl vi/ cko jr w w vl/ cl; \*/ w sr
Figure 7-24. System #3 daily battery voltage (maximum, average, and minimum)

I r * I I 1 T i i i 1 1
I " r T r T T
NO O NO O no O NO o NO o NO O N> o NO o 9 NO o NO o NO o
00 00 -vj -vl cn cn 05 o> O) cn cn cn cn cn cn cn 4k 4k co CO co cn cn cn 4k -b. 4k 4k 4k- CO
c_ c_ > > > in CO CO 6 6 6 z z z 6 6 n cL cL cL T| Tl "Tl k s k > > > k S cL
c c c (C (n CD O o O o o o CD CD (D CD CD CD Q) QO Q) TO TO TO 0) Q) Q) c
CO CO CO cb cb cb cb CO CO CO cb cb CO CO CO CD CO CD CO cb cb cb CO cb cb CO CO CO CO
CO CO co CO CO CO CO co CO co co CO CO co CO N. 4k .n. 4k 4k 4k -N -N 4k. 4k.
Figure 7-25. System #4 daily battery voltage (maximum, average, and minimum)

i i i _ r - r
- i ~
-i - t-
a fO o ro O ro O NO O ro o ro o IO O IO O ro O ro O * 1 1 ro r~ O
00 00 -o O) CD o> O) C0 O) cn cn cn cn cn cn .u CO CO CO cn cn cn a t*. CO
c_ cL > :> > CO CO CO 6 6 6 z z z 6 n n cl_ cL c_ "TI "TI TI k 2 s > :> > k c_
c c c CD (D CD O O O O O o n> CD (D 0) CD CD CD CD 0) 0) 0) T> "O O n) Q) c
1 1 (U CO (O *0 "O T3 i i < < < o O O O O O" ? 1 1 7 -1 *< < *< o
Figure 7-26. System #5 daily battery voltage (maximum, average, and minimum)

Hi i-t- ro H- o M- H- ro M o M M-H N> M- o
00 00 o> CO CO CO
c_ c_ > > > CO co Co 6
c c c c c CD (D CD o

I i I I i I i I i I i I i I <-1 i 1 t-H I
* t-t-f-t-H
o N) o M o hJ O ro O ro O N>
Ol Ol Oi cn cn cn CO CO CO cn cn cn
z 2 Z 6 6 6 cL cL cL Vi "T1 "T1 k S 2 > > >
o o o CD CD CD 0) 0) 0) CD CD CD 0) 0) 0) "O "O *o
< < < O o o o 0 1 o O" O" O' "jl 1 x i i i 1 i
U -U -tk
Figure 7-27. System #6 daily battery voltage (maximum, average, and minimum)