22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.

STATUS AND RELEVANCE OF THE DC VOLTAGE DEPENDENCY OF THE INVERTER EFFICIENCY

F. P. Baumgartner, H. Schmidt*, B. Burger*, R. Bründlinger**, H. Häberlin***, M. Zehner****

University of Applied Sciences Buchs, NTB; Switzerland, www.ntb.ch/pv; [email protected]

*Fraunhofer Institute for Solar Energy Systems ISE, Freiburg, www.ise.fraunhofer.de; [email protected]
**arsenal research; Vienna, Austria; www.arsenal.ac.at/; [email protected]
***Berne University of Applied Sciences (BFH), Burgdorf, Switzerland; www.pvtest.ch; [email protected]
**** University of Applied Sciences Munich; Fb04 Elektro-/Informationstech.; www.solem.de; [email protected]

ABSTRACT: The efficiency characteristics and the “European Efficiency” of a grid connected inverter is dependent
on the hardware topology and changes also significantly with varying input (DC) voltage. Therefore the energy yield
of a PV-system based on an inverter with outstanding peak efficiency might be lower compared to a standard
product, if an inappropriate voltage is chosen. Today, the PV system designer still faces a lack of data on this topic
within most inverter data sheets provided by the manufacturer. For the same reason, most PV-simulation programs
do not take this dependency into account. This paper summarizes previous work related to this topic and presents
approaches on how to effectively model the voltage dependent efficiency. The resulting model parameters are
presented for different inverter products available on the market and the individual average inverter efficiency is
calculated for different DC voltage levels. Finally the authors strongly recommend to measure inverter efficiencies at
independent test laboratories und publish the data. Round robin test for PV inverters have to be realized with partners
form independent laboratories as well as partners from industry.
Keywords: inverter, efficiency, yield, economy

1 INTRODUCTION Table 1 Relation between PV inverter EURO efficiency

and specific costs. In the typical private house PV
The efficiency of a grid connected inverter depends installation the inverter accounts for 10% of the overall
not only on the actual power but also significantly on the system price [6,10].
level of the input (DC) voltage [1-4]. Today, most of the efficiency difference relative Inverter
customers compare the performance of inverters only by between efficiency price
one figure of merit, the weighted inverter efficiency i.e. change €/W
the EURO efficiency ηEU. This may be a weak point in average efficiency value to
4.1% 41%
the design of the PV plant, because most of the highest efficient product
manufacturer’s data sheets don’t give the DC voltage
level used to determine the EURO efficiency value. Thus lower limit of efficiency
and upper limit of efficiency 6.0% 60%
in most cases the information is not available, which is
the optimum DC voltage level to select, in order to of 95% of the market share
achieve the highest weighted efficiency. This lack of data
concerning the DC voltage dependency of the efficiency measurement uncertainty +/- 0.5% +/- 5.0%
directly influences the yield of the whole PV installation
and thus the price of the inverter itself [6]. maximum allowed deviation
The published and recently measured inverters to the guaranteed efficiency -1.3% -13%
conversion efficiencies will be discussed together with according to IEC 61683
the estimated measurement uncertainties. In a critical
review it will be shown that published uncertainty values
of efficiency below 0.1% are not feasible in today’s 0.0%
inverter test laboratories [7]. A model is described which
-0.2%
can be used to calculate the efficiency not only as a
function of the power but also as a function of DC -0.4%
guaranteed efficiency

voltage. This is one of the preconditions to implement the -0.6%

lower deviation of

inverter efficiency characteristics in the widely used PV -0.8%

-1.3%
design software tools [8].
-1.0%

2 INVERTER EFFICIENCY AND PRICE -1.2%

A recent market survey of all PV inverters available -1.4%
in Germany resulted in an average efficiency of 93.1% -1.6% only
when looking upon the weighted EURO efficiency value 93%-1.3%= 91.7%
-1.8%
[9]. New products launched in 2007 offer an average guaranteed
Euro efficiency of 94.4%. The maximum EURO -2.0%
efficiency value is 97.7%. About 95% of all inverter 89% 91% 93% 95% 97% 99%
products provide efficiencies within a range of +/- 3% guaranteed efficiency
around the average of 93.1% [9]. In a simplified manner
the relation between efficiency and price is shown in Fig. 1: Maximum allowed deviation to the guaranteed
table 1. In the used model the inverter accounts for 10% efficiency according to draft IEC 61683:1999 (according
of the overall PV system costs. to the formula – 0.2 (1- η) η% ) [10]

F. Baumgartner et. al 1
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.6

In table 1 it is shown that the best inverter in terms of According to GUM the influence of the measurement
efficiency, featuring 4.1% higher efficiency than an statistics concerning the actual reading, known as type A
average product, may be 41% more expensive in terms of uncertainty contribution, has to be combined with the
€/W when assuming the same generation costs of PV type B contribution as given in table 2. The type A
electricity. uncertainty contribution which corresponds to the
If a manufacturer guarantees for example an inverter standard deviation of the means can also be in the range
efficiency of 93% the customer can expect acutal of some tens of a percent. It strongly depends on the
efficiency values which may not be lower than 91.7% stability of the working point, the active MPP tracking,
according to the IEC 61683 [10]. An efficiency 1.3% the DC source of the inverter and the update interval
lower than its nominal value corresponds to a decrease in during the measurement procedure.
inverter prices of 13%. (Fig. 1) Impacts on the uncertainty of the AC measurements
may occur due to the active inverters impedance
measurement (ENS). Thereby the ENS control system
3 UNCERTAINTY OF MEASURED EFFICIENCY has to generate periodically significant current peaks as
shown in Fig. 2. This may cause an automatic change of
Always, measured values have to be discussed with the current measurement range during a typical inverter
regard to their measurement uncertainties. The state of test sequence. This leads to a deterioration of the
the art in measurement uncertainty of inverter efficiency uncertainty values because then the actual reading is
has to be highlighted before drawing conclusions based much smaller compared to full scale (see Tab. 2)
on fractions of a percent in inverter efficiency. Since When comparing the measurement results of
1993 the international “Guide to the Expression of different laboratories on the same DUT the whole set of
Uncertainty in Measurement” (GUM) is an international uncertainty contributions must precisely be taken into
standard and must also be applied to efficiency account. Relative changes of efficiency values during an
measurements of PV inverters [11]. Based on GUM, individual test sequence typically show a significant
table 2 reports the comparison of the type B uncertainty resolution between 0.1% and 0.05% when performed by
budgets of two different inverter efficiency qualified laboratories.
measurements setups, respectively power analyzers. The In conclusion, the overall measurement uncertainty
uncertainty of the inverter efficiency u(η) is given by the of the inverter efficiency will be in the range between
combined uncertainty of the DC power measurement 0.2% and 0.6% of reading based on a 95% confidence
u(PDC) and the AC power measurement u(PAC) applying (GUM k=2). The uncertainty of the AC power should be
a good estimate when focussing only on the significance
u(η)2 = u(PDC ) 2 + u(PAC ) 2 . Eq. 1 of the relative change of efficiency, for example as it
occurs when the inverters DC voltage is altered. Thus
relative changes of the inverter efficiencies with a
Type B uncertainties in table 2 and 3 were estimated significant resolution between 0.05% and 0.15% may be
by considering the combined uncertainty of the accuracy expected when the test are carried out by personnel in
values relative to the reading and relative to the full scale qualified laboratories using state of the art advanced
value, as given in the data sheets. standard equipment.
The values in table 2 show that the measurement
uncertainty of each efficiency value strongly depends on
the actual power reading relative to full scale. Values
between 0.2% and 0.5% are good estimates. Thereby the
DC power measurement dominates the whole type B
uncertainty budget.
Uncertainty contributions of external current sensors
must also be taken into account. Both systems compared
in table 2 are equipped with shunts. In table 3 the
uncertainty budget of DC and AC power is reported for
the D6000 power analyzer. For this particular case it is
evident, that the main contribution is not caused by the
shunt but by the uncertainty of the DC voltage
measurement from the external current sensor as well as
by the inverters DC voltage reading.

Fig. 2:
Time response of AC voltage and current of an SB3800
inverter while a current test impulse of the impedance
measurement occurs every second at the zero-crossing of
the voltage operating at relatively low power (PAC ≈
450 W) [18].

F. Baumgartner et. al 2
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.6

Table 2 Uncertainty budget of inverter efficiency

measurement setups given for an expanded uncertainty
level of k=2 (95% confidence, see type B contribution)
based on two different power analyzers. The
manufacturer’s data sheet accuracy specification ax was
used to calculate the GUM uncertainty by the well
known formula 2 ax / √3 [11, 12]. The uncertainty values
are given at 3 different actual reading values relative to
full scale.
Relative uncertainty System I System II
power analyzer D6000* WT3000**
30A range LEM Norma Yokogama
actual power readings at full scale
u(PDC) 0.26% 0.13% Fig. 3: Measured conversion efficiency of an inverter
u(PAC) 0.13% 0.05% shown as a function of AC power at four different DC-
u(η) for k=2 0.29% 0.14% voltages. In the upper curve a measurement uncertainty
of +/-0.3% is indicated by the error bars. The DUT was a
actual power readings at 1/2 of full scale
NT6000 new version since June 2007 an inverter without
u(PDC) 0.28% 0.24% transformer (www.sunways.de). The maximum
u(PAC) 0.14% 0.10% efficiency of 97.7% was measured at a MPP voltage of
u(η) for k=2 0.31% 0.26% 349Vdc. The measurement was performed at the ISE
actual power readings at a quarter of full scale Test Laboratory using a D6000 power analyzer using a
DC power supply in the constant voltage mode (see also
u(PDC) 0.37% 0.47%
Table 2, 3 and Fig. 10).
u(PAC) 0.16% 0.19%
u(η) for k=2 0.40% 0.50%
D6000* accuracy values guaranteed for 24 months
WT3000** accuracy values guaranteed for 6 months

Table 3 Uncertainty budget of inverter efficiency

measurement setups given for an expanded uncertainty
level of k=2 (95% confidence, see typ B contribution
[11], based on power analyzers data sheet accuracy
specification, only internal current sensors used). Here,
the actual measured values should be at a quarter of the
full scale reading.
power analyzer actual readings at
D6000 a quarter of
shunt-triax 0.1 to 30A full scale full scale
Uncertainty budget of the DC power measurement
u(voltage signal DC) 0.18% 0.22%
u(current signal DC) 0.18% 0.29%
u(shunt-triax) 0.03% 0.03% Fig. 4: Measured conversion efficiency of an inverter
u(PDC) for k=2 0.26% 0.37% shown as mapping plot in the AC power versus DC
Uncertainty budget of the AC power measurement voltage plane. The DUT was a standard inverter without
transformer. The plot is based on 440 individual
u(voltage signal 50Hz) 0.05% 0.07% efficiency measurements at 20 different DC voltage
u(current signal 50Hz) 0.05% 0.10% levels (350V – 600V) and AC power values (plotted
u(shunt-triax) 0.03% 0.03% relative to PN=2.6kW). The measurement was performed
u(real power calc.) 0.10% 0.10% at the arsenal research [12] using a D6000 power
analyzer (see also Table 3 and Fig. 8)
u(PAC) for k=2 0.13% 0.16%
The maximum efficiency of 95.5% was measured at
the lowest DC voltage value of 350V at about 40% of
4 EFFICIENCY VERSUS DC VOLTAGE LEVEL
nominal power. The efficiency drops to 93% at the upper
limit of the DC voltage range.
The conversion efficiency of an individual grid
Similar efficiency characteristics of other
connected inverter significantly depends on AC power
transformer-less products (Sunways NT6000, SMA
and DC voltage. Fig. 3 and 4 show the two different plots
SB850) were published at previous conferences in the
of this dependency of two different products on the
same AC power /DC voltage plot [1]. However
market. The typical dependence of the widely used
differences in maximum efficiency of up to 3% were
transformer-less inverter topology is shown in Fig. 4 as a
found. The dependency of efficiency with DC voltage is
mapping plot of the efficiency versus power and voltage.

F. Baumgartner et. al 3
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.6

found to be less than 1% for inverters with record

maximum efficiency of ≥ 97%. Transformer less 4.1 CUSTOMERS NEED RELIABLE INFORMATION
inverters with maximum efficiency values ≤ 95% ABOUT INVERTER EFFICIENCY!
exhibited a significantly higher voltage dependency of
about 2.5% [1] similar to recent results shown in Fig. 4. In the last years the state of California started the
There is no general rule about the dependency of well recognized solar initiative promoted by governor
inverter conversion efficiency on DC input voltage. Arnold Schwarzenegger. Since 2005 lists are published
There are inverters that have their maximum efficiency at of eligible PV products, like inverters, PV modules by
low voltages like those shown in fig 3, 4, 5 and 6, but the California Customer Energy Center [15]. There the
there are also inverters that have their top efficiencies at customer will find detailed information for example on
high voltages (e.g. Solarmax 6000C, SB5000TL the inverter efficiency of up to 170 PV inverters available
Multistring) and also inverters that have their maximum on the US market. Among that list some products with a
efficiencies at a medium DC voltage (e.g. Fronius IG30 nominal power of 5kW are listed in table 4. In Dec 2006
and IG40 [2], Magnetek power one). Due to the the California Energy Commission published the final
commonly used topologies of transformer-less inverters guidelines, describing the criteria to be listed as an
they will reach their maximum of efficiency at a DC eligible product [16]. One of the criteria is the
voltage level close to the peak value of the AC voltage. publication of measured inverter efficiency values at
Inverters setup with a transformer tends to increase least for 3 different DC voltages. Furthermore the tests
efficiency at lower DC voltage levels. must be performed by an independent laboratory.
The compilation of this list since 2005 was necessary
Sunny Mini Central 8000TL: Efficiencies η (PDC/PDCn) to provide the customer with transparent and independent
99 information about the inverter efficiency beyond the
98 efficiency values stated in the marketing documents from
97 the manufactures.
E ffic ie n c y η [% ]

96 The establishing of such an independent list would be

95 also highly desirable for the European market. First steps
94 are done by the magazine Photon [7] published the
93 Conversion Efficiency; Vmpp = 350 V results of the author’s publication [1], a mapping plot of
92 the efficiency versus power and voltage, together with
Conversion Efficiency; Vmpp = 400 V
91 their yearly inverter market survey in 2005 [19]. This
90 Conversion Efficiency; Vmpp = 490 V
was very helpful to bring that topic to a broader
89 audience. The magazine Photon started in January 2007
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 to publish in each monthly issue measurement results of
PV-Simulator 20kW,
Kennlinie: FF = 75%, RP = 1MΩ
Normalised DC-power PDC / PDCn (P DCn = 8250W) one individual inverter product with focus on the DC
Fig. 5: Highest measured conversion efficiency at a voltage dependency.
prototype of an inverter without transformer at three
different DC-voltages (SMC 8000TL, PDCn = 8.25 kW)
[2]. At low power the efficiency shows a considerable Table 4 Weighted inverter efficiency values measured at
voltage dependency. With increasing power this independent US test laboratories according to the
dependency decreases and at the highest power levels Guidelines given by the California Customer Energy
efficiency is virtually independent on the DC voltage. Center [16]. Only inverters suitable for the US market are
listed with nominal power 5kW and DC voltage above
100V. The measurement uncertainty is not given. They
SunnyBoy 3800: Conversion efficiency η = f(PDC/PDCn)
may not be directly compared to products of the same
97
96 company available on the European market due to other
95 grid frequency, voltage level and other design features.
94 The weighted efficiency values are shown for 3 different
E ffic ie n c y η [% ]

93 DC voltage values. (The US weighting factors to

92 calculate the average inverter efficiency are given at the
91 end of the this table)
90
89 Product Company VDC,min VDC,nom VDC,max
88 Efficiency; Vmpp = 200 V SB5000US-240V SMA 96.4% 95.9% 94.6%
87 Efficiency; Vmpp = 280 V GT 5.0 – POS Xantrex 95.6% 95.5% 95.2%
86 Efficiency; Vmpp = 350 V
85 PVP 5200 PV Power 96.4% 96.2% 95.5%
84 SUNString 5000 Sunset 93.8% 93.7% 93.3%
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 Power
PV-Simulator 25kW, Kennlinie: FF = 75%, RP = 1MΩ Normalised DC-power PDC / PDCn (P DCn = 4000 W) PVI 5000, US240V One 96.1% 96.7% 96.1%
Fig. 6: Measured conversion efficiency of an inverter IG 5100 Fronius 94.2% 94.9% 93.7%
with transformer at three different DC-voltages (SB According to the CA guidelines [16] the weighting factors to
3800, PDCn = 4 kW) [2]. At all power levels, conversion calculate average efficiency are: 10% 0.04; 20% 0.05; 30% 0.12;
efficiency depends somewhat on the DC voltage. At low 50% 0.21; 75% 0.53; 100% 0.05
power, this difference is up to 4%, whereas at high
power, it reduces to about 1%.

F. Baumgartner et. al 4
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.6

Fig. 8: An excellent example of detailed information

available on the manufactures datasheet shown here for
the NT6000 inverter version 06/2007 from Sunways.
Also the numeric values of the measurement results are
given (www.sunways.de). (see Fig. 3.)

Fig. 7: Weighted efficiency of inverters available in the

US as a function of DC voltage according to the
measurement results published by the California energy
commission see table 4.

4.2 MANUFACTURERS INFORMATION ON DC

VOLTAGE DEPENDANCY

Seven years ago it was suggested in the IEC 61683

standard to measure the inverter efficiency values at least
at three DC voltage levels. [10] Today in most of the
manufactures datasheets this information is lacking.
Typically the manufacturers only give one single Fig. 9: In the last years only the dependency of the
weighted efficiency value, typically the European maximum efficiency on DC voltage was given from the
efficiency ηEU without specifying the corresponding DC manufacturer SMA to their customers. [14]
voltage. A few manufactures like Sputnik
(www.solarmax.com) list the ηEU at two different DC
voltage values in their datasheets [1]. 5 MODELLING CONVERSION EFFICIENCY
This is particularly interesting to the fact, that for VERSUS VOLTAGE AND POWER
example a 2% lower actual efficiency on the same DC Inverter efficiency is mainly determined by the
voltage level lead to a 20% lower specific price of the inverter topology, the power transistors, modulation
inverter (Table 1). Maybe the PV plant designers did not method, switching frequency and filters. Moreover, the
request that information and create their purchase losses in the power switches are a function of the DC
decision only on the simple average Euro- efficiency input voltage and the current in the semiconductors and
values. filters. The following simple model [1, 5] of inverter
Up to now the most comprehensive information losses will be used. It has to be noted that this model is
about ηEU(P,V) are available from Sunways about their not appropriate to be applied to inverters which uses
inverters [7]. On their datasheet the plots of ηEU(P) at internally switching of general parts of the power
three DC voltage levels are shown. Additionally electronic circuits (Fronius IG, SMC8000, Solarmax E
Sunways present a numerical listing of their typical series, Danfoss…) In the following P represents the
measurement results on the same datasheet. They are in output AC power of the inverter and c0, c1, c2 are
comparison with the measurement at the ISE test constant coefficients for a certain DC input voltage VDC.
laboratories as shown in Fig. 3. Together with equation 1 the inverter efficiency can be
In Fig. 8 the characteristics of the maximum expressed by
efficiency at different DC values are given, according to
PAC
a detailed technical documentation of the market leader η(PAC ) = , (Eq. 2)
SMA which was requested [14]. This was for the last PAC + Ploss (PAC , V )
years one of the only information’s on that topic.
Recently SMA also gives the efficiency characteristics as Ploss (P,V) = (c0,0 + c0,1 V + c0,2 V2 + c0,3 V3 ) +
a function of part load at 3 DC voltage levels to their + (c1,0 + c1,1 V + c1,2 V2 + c1,3 V3)P +
customers. + (c2,0 + c2,1 V + c2,2 V2 + c2,3 V3)P2 .
(Eq.3)

F. Baumgartner et. al 5
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.6

and a0,2 given in table 6. The second part of the losses is

Table 5 Fitted coefficients ci,j of Sunways inverter proportional to the current leading to the coefficient U0.
NT6000 measured in 2003 and a the new NT6000 In that model U0 will describe the sum of all the voltage
version 2007 (Fig.2) measured recently. The fit drops over the power semiconductors in the inverter. The
performed according to Eq. 2, 3 and describes the voltage last group of losses are attributed to the over all ohmic
and power dependency of the inverter (data P>5% PN); losses represented by the coefficient R0 which is also
units of power values PAC in [W] and VDC in [V]. The fit fitted by three coefficients (Eq. 4 and table 6).
of the Sunways NT6000 version 2007 coefficients based Together with the nine coefficients of table 6 and the
on measurement values at the fixed DC voltages 349 V, AC current, which may by calculated as a function of the
417V, 499V and 599V at 13 power values each. AC power at a fixed AC voltage the efficiency is defined
Sunways Sunways as a function of DC voltage and AC power.
NT6000 NT6000 Both equation 4 and 5 uses the same type of
version 06/2007 Ref. [1] polynomial descriptions of the DC voltage dependency
(meas. in 2007) (meas. in 2003)
but the coefficients are different. The advantage of the
η max 97.7% 97.0% model based on Eq. 5 is to check the quality of the fitted
c0,0 -1.195E+00 4.848E+00 results of P0, V0 and R0 individually as can be seen in
c0,1 4.508E-02 1.504E-02 Fig. 11a-c. Thus bigger deviations between fit and
measurement are noticed and detailed analyses for
c0,2 -3.251E-05 2.368E-05
example of the measurement uncertainties at lower AC
c1,0 8.060E-03 6.740E-03 power values may be performed.
c1,1 -4.161E-06 1.037E-05
c1,2 2.859E-08 2.716E-08 Table 6 Fitted coefficients to calculate the efficiency
c2,0 3.530E-06 8.934E-06 based on the measurement results shown in Fig. 4
according to Eq. 4 (details see also Fig. 11)
c2,1 5.667E-09 -1.254E-08
a0,0 a0,1 a0,2
c2,2 -8.161E-12 6.378E-12
19.259 -0.065 1.737·10-4
a1,0 a1,1 a1,2
0.846 0.015 -1.104·10-5
a2,0 a2,,1 a2,2
0.612 -7.75·10-4 1.519·10-6

Fig. 11a: Extracted coefficient of the power losses at

IAC=0 as a function of DC voltage shown together with
the fitted curve according to Eq. 4 and Tab. 6. Data are
based on the measurement results at 20 different DC
voltage values shown in the efficiency mapping in Fig. 4.
Fig. 10 Measured conversion efficiency values of the
(Fit performed by ISE, measured at arsenal research)
transformer-less inverter Sunways NT6000 advanced
version 06/2007. The mapping plot is based on the fitted
values given in table 5.

The following expression represents the power looses

model proposed by ISE as a function of AC current IAC
and DC voltage VDC.[17]
Ploss (I,V) = P0(VDC) + U0(VDC) IAC + R0(VDC) IAC2 =
= (a0,0 + a0,1 VDC + a0,2 VDC2) +
+ (a1,0 + a1,1 VDC + a1,2 VDC2) IAC +
+ (a2,0 + a2,1 VDC + a2,2 VDC2) IAC2
(Eq.4)
The first term P0 represent the losses at vanishing
output power at IAC=0. Fig 8a shows the dependency of
P0 as a function of DC voltage. In Eq. 4 this Fig. 11b: Extracted coefficient of the constant voltage U0
characteristics is described by a second order polynomial as a function of DC voltage shown together with the
term with the corresponding 3 fitted coefficients a0,0, a0,1 fitted curve according to Eq. 5. (see Fig. 11a)

F. Baumgartner et. al 6
22nd European Photovoltaic Solar Energy Conference and Exhibition, 3-7 September 2007, Fiera Milano, Session 4DO.4.6

REFERENCES

[1] F. Baumgartner; 20th European PV Solar Energy

Conference; 6-10 June 2005; Barcelona, Spain
[2] H. Häberlin, L. Borgna, M. Kämpfer, U. Zwahlen;
"New Tests at Grid-Connected PV Inverters:
Overview over Test Results and Measured Values of
Total Efficiency". 21th EU PV Conf., Dresden,
Germany, September 2006
[3] H. Häberlin: "Optimum DC Operating Voltage for
Grid-Connected PV-Plants". 20th EU PV Conf.,
Barcelona, Spain, June 2005
[4] B. Bletterie, R. Bründlinger, H. Fechner; “Sensitivity
of photovoltaic inverters to voltage sags - test results
Fig. 11c: Extracted coefficient of the constant loss for a set of commercial products”; 18. International
resistance R0 as a function of DC voltage shown together Conference on Electricity Distriution,06.-09.06.2005,
with the fitted curve according to Eq. 5. (see fig. 11a) Turin, Italien
[5] H. Schmidt et al.: “Wechselrichter-Wirkungsgrade”
Sonnenenergie 4/1996, pp 43…47
[6] F. Baumgartner, H. Scholz et. al, 19th European PV
Solar Energy Conference; Paris 2004; p681;
[7] H. Neuenstein et. al; Photon Jan 2007, p70 .. 74
[8] M. Zehner M., Professionelle Anlagenkonfiguration
– der Status Quo der PV-Programme, Staffelstein 21.
PV-Symposium 2006, Germany
[9] Inverter market survey Germany, Photon 2007 März
[10] IEC61683:1999; Photovoltaic systems – Power
conditioners – Procedure for measuring efficiency;
German version: EN61683:2000 (C.1)
[11] Guide to the Expression of Uncertainty in
Fig. 11d: Calculated weighted efficiency of the Measurement GUM; first edition, 1993, ISBN 92-67-
transformer-less inverter according to the efficiency 10188-9 corrected and reprinted 1995, International
model described also by Eq. 5 and the EURO efficiency Organization for Standardization ISO; Geneva,
weighting coefficients [6]. (Measurement shown in Fig. Switzerland.
4; details see Fig. 11a-c and Tab. 6) [12] LEM Norma www.lem.com manufactures the
D6000, while Yokogama www.yokogama.com
manufactures the WT3000
6 CONCLUSION and OUTLOOK [13] R. Bründlinger, personal communication;
measurement performed at Arsenal Vienna;
It is strongly recommended to establish and publish a www.arsenal.ac.at/; combined measurement
list of weighted inverter efficiency values at different DC uncertainty of about 0.4% (k=2)
voltages for the European market(s). Similar to the very [14] J. Laschinski, “Die optimale PV-
successful California list the inverters have to be Anlagenauslegung”; SMA technical documentation;
examined by independent laboratories. page 148-151; 2004; www.sma.de
Based on this list and further measurements the [15] Lists of eligible Renewable Energy Equipment
dependence of efficiency on power and DC voltage supporting the California’s Solar initiative;
should be modelled and implemented in the common PV http://www.consumerenergycenter.org/erprebate/equi
System design software products. pment.html
These software products could also offer the function [16] Final Guidebook, Emerging Renewables Program,
of a simple economic estimate comparing the inverter 8th ed., December 2006; published by the California
price in €/W to overall PV system price taking into Energy Commission;CEC-3002006-001-ED8F-CMF
account the influence of the resulting weighted efficiency [17] B. Burger, H. Schmidt, ISE Freiburg, personal
of the final design. communication; www.ise.fraunhofer.de
A round robin test of inverters at different [18] H. Haeberlin: “Photovoltaik – Strom aus
independent laboratories should be performed in the near Sonnenlicht für Verbundnetz und Inselanlagen”.
future. AZ-Fachbuchverlag, CH-5001 Aarau, 2007,
Also the deviation in inverter efficiency of a given ISBN 978-3-905214-53-6, und VDE-Verlag,
production line is of high relevance especially for the ISBN 978-3-8007-3003-2 (in German).
inverter manufacturer and the guaranteed efficiency [19] Inverter market survey Germany, Photon 2005 März
values on the market.

F. Baumgartner et. al 7