Energy and Power Engineering, 2013, 5, 182-188
doi:10.4236/epe.2013.54B035 Published Online July 2013 (http://www.scirp.org/journal/epe)
Cost-effective Energy Monitoring of Domestic
Off-grid PV Systems
Arthur James Swart, Ruaan Schoeman, Christo Pienaar
Institute of Applied Electronics, Vaal University of Technology, Private Bag, Vanderbijlpark, South Africa
Received September, 2012
ABSTRACT
Domestic off-grid renewable energy systems have become common place in many areas of the world, as humanity
seeks to keep abreast with global technological changes and advancements. This paper aims to present a cost-effective
energy monitoring system which may be used to analyze and evaluate the operation of a domestic off-grid PV system.
Parameters which are monitored include the output voltage and current from a 55 W polycrystalline PV panel. The
output voltage and current from a power regulation circuit (which could be a DC-DC converter, solar charger or MPPT)
is also monitored with this singular system which incorporates a data logging interface circuit, a data logger and a per-
sonal computer.
Keywords: Off-grid PV System; Data Logging; Current Sensor; Differential Amplifier
1. Introduction
“Most people spend more time and energy going around
problems than in trying to solve them” [1]. These words,
by Henry Ford, well illustrate the inefficiency of trying
to solve existing problems by finding ways to circumvent
them. A current major environmental challenge which
exists is climate change [2], a challenge that cannot be
circumvented as a possible solution. The challenge needs
to be tackled head-on.
Fortunately, many nations have started to enforce
policies and regulations surrounding the reduction of the
global carbon footprint, which could, in the long run,
mitigate climate change. For example, in December
1997 the Kyoto Protocol was developed in which 160
nations signed an agreement in order to cut or reduce
carbon emissions through carbon taxes and the Clean
Development Mechanism (CDM) [3].
Many governments are also investing huge sums of
money into searching for alternative energy sources as a
means to replace current fossil fuel and nuclear energy
power plants. The worldwide venture investment in
“green technology” companies totalled US$ 2.57 billion
for the first quarter of 2011[4].
Subsequently some suggest that renewable energy
could supply as much as 77 percent of the world’s energy
demand by 2050, which would equate to a reduction of
one third of greenhouse gas emissions compared to busi-
ness-as usual projections [5]. It is predicted that solar
power will become the dominant energy source in the
future (see Figure 1). A branch of solar energy research
that has received worldwide attention is the photovoltaic
(PV) renaissance [6].
Solar power generation through PV arrays (strings), is
arguably the most eco-friendly, emission free and sus-
tainable source of energy known to man [7]. Many com-
mercial systems for domestic off-grid applications cur-
rently exist which could be purchased and installed within
a matter of hours. However, these systems rarely incor-
porate some or other type of monitoring, whereby the
user could analyze and evaluate not only the energy
consumption, but also the energy supplied [8].
Figure 1. Energy forecast according to the Scientific Advi-
sory Board [9].

A. J. SWART ET AL. 183
Numerous reasons for energy monitoring have been
documented [10-13], and include verification of energy
usage, determining where energy can be saved, avoiding
maximum demand loads, increasing reliability of energy
supplied, verifying mathematical calculations of energy
consumption, detecting faults and enabling effective in-
tervention to minimize possible energy losses.
The purpose of this paper is to present a cost-effective
energy monitoring system for domestic off-grid PV sys-
tems. It will firstly review some commercially available
data loggers and their required interfaces. The practical
setup used to validate the performance of the data logger
equipment is then provided. Results and discussions fol-
low with some succinct conclusions.
2. Data Loggers
A variety of commercial data loggers can be found via
the Internet today. Just one search on Google for “data
loggers” revealed over two million hits, while Google
Scholar presented 27 700 hits. Noteworthy too is the
number of research articles on IEEE Xplore which fea-
ture the words “data loggers”, being 354 in total [14].
However, only six (6) articles from the IEEE Xplore
consider data loggers in the PV environment, with one of
these six articles referring to a PICO based logger [8].
A data logger, or energy meter, should be cost-effec-
tive, user-friendly, easily and quickly accessible and re-
liable. Three commercially available data loggers in
South Africa were reviewed as possible candidates for
the cost-effective energy monitoring system. These three
data loggers are given in Table 1 with selected advan-
tages and disadvantages.
The PICOLOG 1012 is manufactured by PICO Tech-
nology in the United Kingdom [15] and is available from
two local approved distributors in South Africa [16,17].
They are currently the most cost-effective loggers avail-
able (12 single-ended input channels with a maximum
input voltage of 2.5 V), but need to be permanently con-
nected to a personal computer (PC) to enable data storage
and analysis. A general purpose notebook (price of
around Є400 in South Africa) is the more logical choice,
as it is portable, compact and provides battery backup in
case of power disruptions.
The DAQPro 5300 is manufactured in the United
States (USA) by Fourtec Fourier Technologies [18], and
is available from a number of local suppliers in South
Africa. They are modestly priced and do not require a
permanent connection to a PC for logging purposes.
They are limited to eight analogue input channels with a
maximum input voltage of 10 V.
The CAMPBELL Scientific CR800 is the Rolls Royce
of data loggers and is manufactured in the USA [19]. It
has six single-ended analogue inputs with a maximum
input voltage of 5 V. It can store data over an extended
period of time and be remotely activated for data
downloads.
Due to the advantageous listed in Table 1, the PI-
COLOG 1012 was selected as the preferred choice for
the cost-effective monitoring system.
3. PICOLOG 1012 Setup
The PICOLOG 1012 is relatively easy to install with the
supplied software. The user interface comprises the PI-
COLOG PLW Recorder which needs to be setup for data
to be recorded directly to the hard drive of a PC. The
main window of this software program is shown in Fig-
ure 2, while the parameter window is shown in Figure 3.
Real time graph analysis of the results may be viewed
instantaneously, while comments or notes may be added
to each individual log file. The spread sheet button may
be used for data transfer to other software packages such
as MS EXCEL or WORD.
Table 1. Data logger summary.
Data loggerAdvantages Disadvantages
PICOLOG
1012
 Relatively cheep (Є200)
 Very user friendly inter-
face
 Repetitive unlimited sam-
ples with unique file
names
 Real time mathematical
calculations
 12 input analogue / digital
channels
 2 output digital channels
 Remote download
 Time stamp per sample
missing
 Must be permanently
connected to a PC
DAQPro
5300
 User friendly interface
 Onboard memory for
medium data storage
 Modestly prices (Є800)
 8 input analogue
channels
 Limited number of
samples
 Mathematical
calculations after data
collection
 One alarm output
reduces input channels
by 1
 No remote download
CAMPBELL
Scientific
CR800
 User friendly interface
 Onboard memory for
extensive data storage
 Remote download
 4 output digital channels
 Resistance measurements
 Pulse counter
 Very expensive
(Є1000)
 6 single-ended
analogue input
channels
 Mathematical
calculations after data
collection

A. J. SWART ET AL.
184
Add note or comm ent
Real time gr aph analysis
Spread sheet for data transfer
Unique channel / unit labels
Samples per time / count
Personal directory / log file name
Figure 2. PICO 1012 software main window.
Scaling data to true values
Alarm parameters, activation and outputs
Unit values and display
Figure 3. PICO 1012 software parameter window.
The log file can be saved to a personal directory with a
specified file name. This file name should have a digit
value at the end which will automatically be increased by
1 for each repetitive log saved to the PC. Sampling can
be either per time (maximum time for measuring and
recording) or per count (set a maximum number of
counts). The sampling time interval must also be stipu-
lated, and must never cover more than a few days in or-
der to avoid data file corruption.
Up to 10 channels may be used for recording (units are
stipulated in the channel setting screen), while an unlim-
ited number of channels may be used for basic mathe-
matical calculations using the recorded data. A bonus of
this software is its alarm capability (see red light next to
DC voltage) which can be used to alert the user to possi-
ble faults or undesirable conditions. It may also provide a
digital output signal via the PICOLOG 1012 to an exter-
nal alarm or switching circuit.
The PICOLOG 1012 has the disadvantage of not in-
cluding a time stamp per sample. However, a time stamp
is included for the final sample within a given log file.
The process of determining the individual time stamps
for previous samples within the log file proves tedious
and laborious.
4. Data Logging Interface
The average energy consumption for domestic homes can
vary between 2 kW/h and 16 kW/h per day [20-24] de-
pending on appliance usage, household size, occupancy
profile and seasonal variation. The maximum load pro-
file would require a 16 kW DC-AC inverter, such as the
VICTRON QUATRO 24/8000/200 available from Cur-
rent Automation in South Africa [25]. This inverter has a
maximum input DC voltage of 66 V, which is inline with
other research reporting on output PV arrays or string
voltages for domestic usage [20]. However, it is possible
that these PV voltages could be as high as 400 V, but this
would rarely be found in general domestic applications
where the roof structure will not allow for large numbers
of PV panels. Considering the 66 V parameter then re-
veals that the data loggers mentioned above will not be
able to accommodate these high voltages, as most of
them have a maximum input voltage of 2.5 V. A data
logging interface (DLI) circuit is subsequently required
to condition the voltage to make it less than the maxi-
mum input voltage required by the data loggers. The DLI
must also be able to provide DC current monitoring using
hall-effect current sensors. The advantages of the DLI
used in this research include the following:
 Internal battery backup for current sensors;
 Generic use for solar chargers, MPPT’s and DC-DC
converters;
 Dual channel monitoring on one-board;
 Current sensor interchange-ability according to current
demands;
 Easy calibration using a multimeter; and
 Protection for data loggers that have a maximum input
voltage of 2.5 V.
The circuit diagram for two channels is shown below
in Figure 4. Channel B is used to monitor the voltage
and current from the PV panel to the MPPT, while chan-
nel A monitors the same parameters from the MPPT to
the battery. Points 1-9 and A-H are the connection points

A. J. SWART ET AL.
185
for the hall-effect current sensors, which could be either
the LTS 6-NP (LEM product) or the ACS756 (AL-
LEGRO product) (see Table 2 for selection). The inline
fuse ensures safety in the case of a short circuit and is
highly recommended for PV system installations on ei-
ther side of the power regulation circuit (being the MPPT).
The additional on-board DC-DC converter provides an
additional power source for circuits requiring isolation
from the main power regulator (certain data loggers re-
quire an isolated power source). The battery serves a
two-fold purpose, being a power source to the load dur-
ing the night and a constant power source to the current
sensors.
It is important to note that the negative rails on the
circuit board may not be connected to a common ground
point. This is due to the fact that most MPPT’s have a
common positive rail. A differential amplifier is there-
fore required to provide a common ground point required
for data logging (achieved using the operational amplifi-
ers TL072).
Voltage calibration was performed using a 55 W poly-
crystalline PV panel, a 12 V 12 Ah battery and a digital
multimeter. The battery was initially connected to the
output of channel A on the DLI (X8). The battery termi-
nal on the MPPT was connected to the input of channel
A (X5), with no load being attached. The differential
amplifier produces an output voltage which is reduced by
R3/R10 to give a voltage below 2.5 V. The output of the
DLI circuit was then measured at point X2-3 (0.970 V)
which becomes the input to channel 1 in the PICOLOG
1012. This corresponds to a true battery voltage of 12.81
V (measured with the multimeter). The output of the DLI
circuit for channel B was also measured (point X3-3
giving 0.5 V representing a true value of 0 V; these val-
ues were assumed to be the same for channel A).
Figure 4. Circuit diagram of the DLI.
Table 2. Hall-effect current sensors.
Sensor number LTS 6-NP ACS756
IMAX ±19 A ±50 A
Cost Є10 Є6
Sensitivity 104 mV/A 40 mV/A
Advantage/Disadvantage Finer resolutions based on wire taps
Greater accuracy for smaller currents
Rectangular mounting bars make PCB mounting difficult
Less accurate for smaller currents

A. J. SWART ET AL.
186
The PV panel was subsequently connected to the input
of channel B, with the output being left open. This results
in the open circuit voltage of the PV panel being meas-
ured (being 20.51 V in this case). The output of the DLI
for this open circuit voltage was then measured at point
X3-3 (1.737 V).
Current calibration was then done by measuring the
output voltage of channel B (not yet connected) of the
DLI circuit (X3-2). This output voltage (1.245 V) repre-
sents a 0 A flow of current, and is assumed to be the
same for channel A. The maximum current which can be
detected by the current sensors is 19 A, which is then
represented by the maximum input voltage to the data
logger, being 2.5 V.
Next, the output of channel B was connected to the PV
panel point on the MPPT which results in current flow,
as the MPPT starts charging the battery. This step can
only occur around midday when a clear sky condition (no
clouds) exists. An additional measurement is taken for
channel A and B of the DLI, as the output of the differ-
ential amplifier is non-linear. These additional measure-
ments were taken after 30 minutes of charging, and are
shown in Table 3. These parameters were then entered
under the “Table Lookup” option in the Scaling window
of the PICOLOG 1012. Maximum input voltage to this
DLI is limited to 250 V, in order to prevent an over volt-
age condition on the data logger.
The reason for not using a power supply for calibration
is due to the fact that most home owners would not have
a power supply. However they could still make use of the
voltages available from the PV panel and battery by us-
ing a commercially available digital multimeter.
5. Practical Setup
Figure 5 shows a block diagram of the practical setup
which was used to verify the operation of the DLI and
PICOLOG 1012.
The practical setup comprises a 55 W polycrystalline
PV panel, a 12 V 20 A MPPT, a 12 V 12 Ah lead-acid
battery and a 12 V 3 W LED load. A basic solar charger
or DC-DC converter may also function as the power
regulation circuit [26]. The PV panel was set to a tilt an-
gle of 36° at a fixed orientation angle of 0°(true north).
This tilt angle was justified by another research project
currently underway at VUT, and was recently presented
at an international conference in China [27]. The PI-
COLOG 2012 recorded the input and output voltages and
currents for each channel. Calibration was done using a
digital voltmeter, and was discussed in the previous sec-
tion.
6. Results and Discussion
The battery and PV voltage over a 24 h time period is
shown in Figure 6, while Figure 7 shows the current
distribution graph of the battery and PV panel for the
same time period. Figure 8 presents the input and output
power of the MPPT used in the practical setup. These
graphs are copied straight from the PICO 1216 software,
and have not been edited in any way.
Table 3. Scaling parameters.
PICOLOG 1012 Channel A Channel B
DLI out 0.50.970 1.188 0.5 1.5721.737
Scale
True value0 12.81 14.07 0 17.1220.51
DLI out 1.245 2.5 1.245 2.5
Scale True value0 19 0 19
Battery
Notebook12 Ah
12 V
Load
LED 3 W
Orientation =0°12 V
Tilt =36°
MPPT
20 A
12 V
PICOLOG 1012
PV panel
p-Si 55 W
DLI DLI
Figure 5. Practical setup.
V
10
15
20
B a tte ry vo ltagePV v oltage
06:00 12:00 18:00
AB
Figure 6. Battery and PV voltage for a 24 h day.
A
0
0.5
1
1.5
Battery currentPV current
06:00 12:00 18:00
C
D
Figure 7. Battery and PV current for a 24 h day.

A. J. SWART ET AL. 187
W
0
10
20
30
Pow er inputPow er output
06:00 12:00 18:00
E
F
Figure 8. Input power (from the PV panel to the MPPT)
and output power (from the MPPT to the battery).
The PV panel’s voltage rises at 08:00 and falls at
18:00, giving rise to 12 hours of possible sunlight (see
Figure 6). However, this does not mean that optimum
output power is available for 12 h, as is indicated by the
gradual rise of the current graph shown in Figure 7.
Point A, in Figure 6, illustrates the PV panel’s output
voltage rise from approximately 11 V (diffused radiation
received) to 22 V (its maximum power point with direct
radiation received) in accord with the movement of the
sun. Point B shows the effect of cloud movement result-
ing in the PV panel’s output voltage varying considerable.
Point C (in Figure 7) highlights a concern in that the PV
panel’s output current is not precisely 0 A during the
night. This suggests that current calibration still requires
attention. Point D verifies the fact that the PV current
will always be higher than the battery current during the
day, as the MPPT also consumes energy.
Point E, in Figure 8, reveals the change-over in power
consumption. During the night the battery provides
power to the MPPT for driving the load, while the PV
panel takes over during the day to fulfil this role. The
negative power simply indicates that current is flowing in
the reverse direction (from battery to the MPPT). Point F
denotes the maximum power drawn for the day, being
approximately 32 W, at approximately 10:15 the morning.
7. Conclusions
A cost-effective energy monitoring system for domestic
off-grid PV system installations was presented. The PI-
COLOG 1012 was identified as an ideal real time data
logger and a suitable DLI circuit was designed and con-
structed. The results prove that the input and output pa-
rameters of the power regulation circuit can be success-
fully recorded and used to analyse the PV system’s per-
formance.
REFERENCES
[1] At Task, “Six Common Mistakes that Plague IT Projects
and Three Keys to Avoiding Them,” 2011.
[2] S. V. Larsen, L. Kørnøv, and A. Wejs, “Mind the Gap
in SEA: An Institutional Perspective on Why Assessment
of Synergies amongst Climate Change Mitigation, Adap-
tation and Other Policy Areas are Missing,” Environ-
mental Impact Assessment Review, Vol. 33, No. 1, 2011,
pp. 32-40. doi:0.1016/j.eiar.2011.09.003
[3] S. M. Shaahid and M. A. Elhadidy, “Economic Analysis
of Hybrid Photovoltaic-diesel-battery Power Systems for
Residential Loads in Hot Regions-A Step to Clean Fu-
ture,” Renewable and Sustainable Energy Reviews, Vol.
12, No. 2, 2008, pp. 488-503.
doi:0.1016/j.rser.2006.07.013
[4] IPCC, Special Report on Renewable Energy Sources and
Climate Change Mitigation, [Online], Available at:
http://www.ipcc.ch/. Accessed: 20 November 2011
[5] R. Pichs, “Statement by a Colleague of IPCC Working
Group Co-chair Prof Ottomar Edenhofer,” 2011.
[6] T. P. Chang, “Optimal Tilt Angles for Photovoltaic Mod-
ules Using the PSO Method with Nonlinear Time-varying
Evolution,” Energy, Vol. 35, No. 5, pp. 1954-1963.
doi:0.1016/j.energy.2010.01.010
[7] E. Dursun and O. Kilic, “Comparative Evaluation of Dif-
ferent Power Management Strategies of a Stand-alone
PV/Wind/PEMFC Hybrid Power System,” International
Journal of Electrical Power and Energy Systems, Vol. 34,
No. 1, 2012, pp. 81-89. doi:10.1016/j.ijepes.2011.08.025
[8] L. K. Biswas, M. H. Rahman and S. Haque, “A Low Cost
Caracteristics Analyzer of PV Module,” Developments in
Renewable Energy Technology (ICDRET), 2012 2nd In-
ternational Conference, 2012, pp. 1-5.
[9] Energy Revolution-YUDU, [Online], Available at:
http://content.yudu.com/Library/A1tlhk/Rexelrenewables
Sharp/resources/12.htm. Accessed: 2012-04-12,
[10] N. Lemmer, “Using Energy Meters as a Tool to Improve
Energy-efficiency,” 25 Degrees in Africa, 2012, pp.
28-29.
[11] S. C. Brofferio, A. Antonini, G. Galimberti and D. Galeri,
“A Method for Estimating and Monitoring the Power
Generated by a Photovoltaic Module Based on Super-
vised Adaptive Neural networks,” Smart Measurements
for Future Grids (SMFG), 2011 IEEE International Con-
ference, 2011, pp. 148-153.
[12] D. Sera, R. Teodorescu and P. Rodriguez, “Photovoltaic
Module Diagnostics by Series Resistance Monitoring and
Temperature and Rated Power Estimation,” Industrial
Electronics, IECON 2008, 34th Annual Conference of
IEEE, 2008, pp. 2195-2199.
[13] N. Pearsall and B. Atanasiu, “The European PV System
Monitoring Guidelines - Modernisation under the PER-
FORMANCE Project,” Photovoltaic Specialists Confer-
ence (PVSC), 2009 34th IEEE, 2009, pp. 000256-000261.
[14] IEEE Xplore, Data Loggers, [Online], Available at:
http://ieeexplore.ieee.org. Accessed: 1 July, 2012
[15] PICO Technology, Homepage, [Online], Available at:
http://www.picotech.com/index.html. Accessed: 11 July,
2012

A. J. SWART ET AL.
188
[16] AP Electronics, Homepage, [Online], Available at:
http://www.apelectronics.co.za/. Accessed: 11 July, 2012
[17] RS Components, Homepage, [Online], Available at:
http://za.rs-online.com/web/. Accessed: 11 July, 2012
[18] Fourtec Fourier Technologies, Homepage, [Online],
Available at: http://www.fouriersystems.com/. Accessed:
17 July, 2012
[19] Campbell Scientific, Homepage, [Online], Available at:
http://www.campbellsci.com/. Accessed: 20 July 2012
[20] M. Abdel-Salam, A. Ahmed, M. Amery, M. Swify, A.
El-kousy and K. Sayed, “Design and Implementation of
Stand-alone Residential PV System,” Applied Electrical
Engineering and Computing Technologies (AEECT),
2011 IEEE Jordan Conference, 2011, pp. 1-6.
[21] R. A. R. Kilpatrick, P. F. G. Banfill and D. P. Jenkins,
“Methodology for Characterising Domestic Electrical
Demand by Usage Categories,” Applied Energy, Vol. 88,
No. 3, 2011, pp. 612-621.
doi:10.1016/j.apenergy.2010.08.002
[22] U. Deichmann, C. Meisner, S. Murray and D. Wheeler,
“The Economics of Renewable Energy Expansion in Ru-
ral Sub-Saharan Africa,” Energy Policy, Vol. 39, No. 1,
2011, pp. 215-227. doi:10.1016/j.enpol.2010.09.034
[23] S. K. Firth, R. A. Buswell, and K. J. Lomas, “A Simple
Model of Domestic pv Systems and Their Integration
with Building Loads,” Eleventh International IBPSA
Conference, Glasgow, Scotland, 2009.
[24] Z. Li, F. Boyle, and A. Reynolds, “Domestic Application
of Solar PV Systems in Ireland: The Reality of Their Eco-
nomic Viability,” Energy, Vol. 36, No. 10, 2011, pp.
5865-5876. doi:10.1016/j.energy.2011.08.036
[25] Current Automation, Homepage, [Online], Available at:
http://www.rectifier.co.za/index.html. Accessed: 20 July,
2012
[26] O. Asowata, J. Swart and C. Pienaar, “Optimum Tilt An-
gles for Photovoltaic Panels during Winter Months in the
Vaal Triangle, South Africa,” Smart Grid and Renewable
Energy, Vol. 3, No. 2, 2012, pp. 119-125.
doi:10.4236/sgre.2012.32017
[27] O. Asowata, A. J. Swart and H. C. Pienaar, “Optimum
Tilt and Orientation Angles for Photovoltaic Panels in the
Vaal Triangle,” Asia-Pacific Power and Energy Engi-
neering Conference, Grand Mercure Baolong Hotel,
Shanghai, China, 27-29 March, 2012.