science & technology inverters test This copy is licensed to Atonex BV, Wimco Baauw. Straight on a path to success A compact, unadorned 4 kw inverter from China has passed the PHOTON test with a more than respectable result: its efficiency is as impressive as its operation is convenient Highlights Text: Heinz Neuenstein, Jochen Siemer The Goodwe GW4000-SS is a transformerless, single-phase inverter with a nominal AC capacity of 4 kw and DC capacity of 4.15 kw. The device operates without a fan, with a housing featuring protection type IP 65 and a very wide temperature range; it can also be installed at practically any location the user likes. The display and communication options are good. The efficiency develops very consistently and at a very high level: the power dependency is quite low, and the MPP voltage has even less impact on it. To date, no device made by Goodwe Power Supply Technology Co. Ltd. from Suzhou in Jiangsu province, which only started producing inverters in 2010, has undergone testing by PHOTON Lab. As usually the case when newcomers are being tested, especially those from the ar East, a few fundamental questions first needed to be asked: is this another player trying to cut a slice from the solar market, without actually enhancing it with another interesting product? Or does Goodwe actually have a product on offer which is worth taking a closer look at? In this case, it is the latter. or a start, Goodwe is unequivocally making the following claim on its website: the company is avowing its intention to become a serious competitor to market leader SMA, and refers to its plans for a densely-woven supply and service network within Europe. Now, ambitions of this caliber are not exactly a rarity in the solar industry; it remains to be seen if Goodwe can achieve its targets. However, it is now apparent that the Goodwe GW4000-SS, which was sent to PHOTON Lab in July within the scope of the standard test agreement, performed quite remarkably. Construction The Goodwe GW4000-SS features a nominal AC capacity of 4 kw, and feeds into the power grid on a single phase. It is one of a series of 5 devices which come in two different housing sizes. The 3 smaller devices, GW1500-SS, GW2000-SS and GW3000-SS, with their 1.5, 2 or 3 kw nominal AC capacity, are just 35 33 12.5 cm in size, weighing 12 and 13 kg (this being the GW3000- SS) respectively. The test candidate and its cohort, the GW4600-SS with its 4.6 kw capacity, have outside dimensions of 41.7 39 14.2 cm and weigh in at 18 kg. All models in the series are based on circuitry without potential separation, which means none features a transformer. The layout of the GW4000-SS is very clearly arranged, making it a production-friendly device. With the exception of the control unit, the entire circuitry, including the power element, is housed on one large circuit board, which occupies virtually the entire interior of the device. This reduces the total number of connecting wires, some of which are routed under this circuit board. Its three inductors a boost converter and two sinusoidal filter chokes are installed above the board, sharing a space in the housing on the cooling element, which also functions as an assembly platform. The control unit with a display which protrudes through a cutout in the housing cover is located on a smaller circuit board on a second assembly level. Apart from the display, the device status is also communicated by 3 LEDs. The power transistors are fitted from the solder side and have been pressed onto the cooling element with the aid of plastic crosspieces. The device is cooled passively without a fan. At the bottom right of the housing, the automatic circuit breaker is installed just above the grid input; it disconnects the device from the grid as soon as grid voltage, grid frequency, insulation resistance of the solar generator or the grid-side fault current deviate from the specified values. The housing is comprised of three parts: the extruded aluminum cooling element functioning as an assembly platform, a steel plate frame, and the cover made of cast aluminum. It complies to protection type IP 65. The GW4000-SS is therefore suitable for installation both inside and outside buildings. The electrolytic capacitors in the power element, as well as those in the control electronics, belong to the 105 C component temperature class. They are therefore well configured in relation to the permitted ambient temperature of up to 60 C. The device does not feature any varistors for limiting DC and AC side overvoltage. The connections for the solar generator and the grid are guided through plug-in connectors in the housing, using three pairs of MC4 connectors from Multi-Contact on the DC side and one three-pin Wieland connector on the AC side. A mechanical DC disconnect is on the bottom of the device. 158 International December 2012
This copy is licensed to Atonex BV, Wimco Baauw. 90 83 76 68 75.4 60 77.6 52 Verena Körfer / photon-pictures.com (2) 44 36 28 20 Sober and simple: rom the outside, the GW4000-SS makes an impression every bit as unspectacular as its interior. The thermographic imaging reveals how well its fan-free cooling concept functions which of course is also due to its high efficiencies. The inverter features the following communication interfaces: 2 RS485 for communicating with up to 50 other devices along with its in-house made EzLogger monitoring system, and a USB port for data exchange with a PC. Wireless and Bluetooth communication are available as an optional extra. Operation The device is delivered to the user well protected and safely packaged in a cardboard box, in between 2 polystyrene foam shells. It is affi xed to the wall using a bracket. Weighing 18 kg, the GW4000-SS is very light considering its nominal DC power of 4.15 kw. If the solar generator is correctly dimensioned and the internal DC circuit breaker is activated, the inverter still requires 43 seconds to run various tests before connecting to the grid and starting operation. The display, which is easy to read, features white background lighting, which is activated by tapping on the housing. It is capable of displaying graphics and provides a lot of information at the same time. The system status is shown both as a diagram and in writing. A bar diagram shows the quantity of energy fed in over the course of the day, which can also display the total amount of energy fed in, the current output and the time, which are continually displayed in figures as well. The upper part of the display shows status, generator voltage and current, AC voltage and current, frequency, fault log, type designation, fi rmware version and language settings. A but - ton allows the user to scroll through the operating menu. As display language, English, German, rench, Spanish and Italian are available. Instruction manual A clearly presented manual is enclosed with the device. It is available in an English, German and Chinese and comes along with general explanations, which also includes information on assembly, connection, operating characteristics as well as display and fault messages. At the time of testing, it was available in English only as a download from the Goodwe home page. Circuit design The inverter has a two-stage circuit design. The energy from the PV array initially flows through an EMI filter and enters the boost converter, which raises the input voltage to the same level as the intermediate circuit voltage. A capacitor smoothens the input voltage. The H6 output bridge chops the input DC voltage into pulsewidth modulated, rectangular-shaped voltage blocks. Within the output bridge, and before the sine wave chokes, there are two flyback paths which ensure that the blind current from the chokes cannot flow back into the intermediate capacitor. This reduces the switching losses from the output bridge and the hysteresis losses in the output chokes, also increasing the efficiency. The use of a boost converter results in an MPP voltage range of 280 to 500 V. The voltage level in the intermediate circuit has been chosen to ensure that the inverter can always feed into the grid. The sine fi lter, which is comprised of sine wave chokes and other capacitors, smoothens the rectangular-shaped voltage blocks into sinusoidal waveforms with the grid frequency of 50 Hz. Along with the automatic circuit breaker, which is triggered when the grid voltage, grid frequency or a grid-side DC current deviates from specifi ed thresholds, the insulation resistance of the generator is also monitored. Any radio interference which arises is eliminated by an output fi lter located directly before the grid terminals. Measurements All of the following measurements are based on a grid voltage of 230 V. The maximum DC voltage of the GW4000-SS totals 580 The Solar Power Magazine International Goodwe GW4000-SS A 96.9 % for medium irradiation 12/2012 www.photon.info The Solar Power Magazine International Goodwe GW4000-SS A 97.1 % for high irradiation 12/2012 www.photon.info International December 2012 159
science & technology inverters test This copy is licensed to Atonex BV, Wimco Baauw. Conversion efficiency The GW4000-SS reaches efficiencies of 97 percent and more over the entire MPP voltage range. It is just a pity that the upper third cannot necessarily be used due to the insufficient distance to the maximum DC voltage. áå=b VR UR TR s NIMMM M SMM QMM OMM M M O Q S U NM NO TR UR VR áå=b ât B MPPT adjustment efficiency The MPP tracking works flawlessly, with a small weakness at 200 V and lowest power being negligible. VV B = Overall efficiency This is reflected in its result for overall efficiency: there are few differences to the conversion efficiency diagram, with the maximum values being identical. pìã= áå=b VR UR TR s NIMMM M SMM QMM OMM M M O Q S U NM NO TR UR VR pìã= áå=b ât B 160 International December 2012
This copy is licensed to Atonex BV, Wimco Baauw. V, with the nominal DC power amounting to 4,150 W. A maximum PV output of 4,600 W can be connected. When operating in the MPP voltage range of 465 V or more, the open-circuit voltage of the simulator must be limited, as it will exceed the maximum inverter DC voltage at an IV curve fill factor of 75 percent. Locating the MPP: When measurements started, both the DC side and the AC side were switched off. At a predefined IV curve at nominal output and an MPP voltage of 384 V, the inverter requires around 43 seconds before connecting to the grid. Another 4 seconds elapse before it reaches the MPP. The inverter requires 4 seconds to change to the next-lowest MPP voltage range of 373 V, with the change to the next-highest range of 396 V taking around 10 seconds. MPP range: The MPP range of 280 to 500 V is equivalent to that of a normal range inverter. This puts the maximum MPP voltage, at current fill factors, too close to the input voltage of 580 V. Conversion efficiency: The inverter is able to operate at 105 percent of its nominal power in the MPP voltage range spanning 280 to 500 V. The efficiency could therefore be determined for this efficiency range. A hatched area in the diagram stretching from the bottom left to the upper right marks the area where limitations are encountered when crystalline modules are used due to the insufficient distance from the maximum DC voltage; the cross-hatching marks a zone which extends even further down indicating limitations when thinfilm modules are used (due to the lack of potential separation, this device is generally unsuitable for use with thin-film modules). Large areas marked by the same color, indicating uniform efficiency, illustrate the homogeneous development of efficiency. The upper area, at 97 percent or more, forms a large plateau which extends over the entire MPP voltage range. The maximum, which is marked by the intersection of the two blue lines at 40 percent of nominal output and 349 MPP voltage, is 97.8 percent. This corresponds exactly to the manufacturer s specifications. Toward higher MPP voltages, the maximum conversion efficiency drops by around 0.6 percentage points, falling by about 0.7 percentage points toward low voltages. More significant losses only materialized in the low power area, this being less than 15 percent of nominal capacity, with the efficiency falling by 4 to 7 percentage points in this case. The power factor cos φ at nominal output was about 1. Weighted conversion efficiency: The European efficiency reaches its highest measured value at 349 V MPP voltage and totals 97.3 percent at this point with the manufacturer s specification being slightly higher, at»more than 97.4 percent«. The difference between maximum conversion efficiency and maximum European efficiency therefore amounts to 0.5 percentage points. The highest value measured for Californian efficiency was 97.4 percent, which was attained at 361 V MPP voltage. MPPT adjustment efficiency: With an exception at the lowest power level and around 303 to 315 V MPP voltage, the MPPT adjustment efficiency is consistently more than 99 percent over the entire operating range. Overall efficiency: As the product of conversion efficiency and MPPT adjustment efficiency, the overall efficiency also exhibits consistent values, the highest of them being in the middle third of the nominal output and MPP voltage range. Just like the conversion efficiency, the maximum value amounts to 97.8 percent and is reached at 40 percent of nominal power and around 349 V MPP voltage. Overall efficiency curves, average overall efficiency and PHOTON efficiency: At 349 (or to be exact: 349.5) V MPP voltage, the GW4000- SS reaches its highest efficiency of 97.8 (in exact figures: 97.76) percent. The minimum MPP voltage of 280 V is simultaneously the value at which the overall efficiency remains the lowest, although 97.04 percent is still a very respectable value. At the upper end of the MPP voltage range of 500 V, an overall efficiency of up to 97.31 percent is attained. These values illustrate the uniform performance the inverter delivers at a high power level. The averages determined over the entire MPP voltage range and all nominal output levels also result in a very good PHOTON efficiency of 96.9 percent for medium irradiation. Under Californian weighting for high irradiation, it even reaches a value of 97.1 percent. eed-in of nominal power: The GW4000-SS feeds 100 percent of its nominal output in over the input voltage range of 280 to 500 V and at an ambient temperature of 25 C. Displayed output power: In order to verify the power measured by the inverter, the values it displayed at a constant MPP voltage of 384 V, which is the medium range, were compared with those measured by a power analyzer. The values measured and displayed by the GW4000-SS exceeded the real value by 3.3 percent at the lowest powers, after which the measuring inaccuracy decreased significantly (by +0.5 percent), only to increase again to 2 percent at levels above 20 percent of nominal power and to remain there not exactly an impressive level. However, the precision of the display is equivalent to a precision class B meter (previously: precision class 1). Operation at higher ambient temperatures: According to the manufacturer, the GW4000- SS is designed for temperatures of up to 60 C, whereby a reduction in output should be expected at more than 45 C. The test did not verify this statement, with the inverter continuing to feed in, unperturbed, 100 percent of its nominal output into the grid before the test was aborted at 61 C. The efficiency fell over this temperature range by around 0.3 percentage points. The operating point selected was 4,200 W DC power and 384 V MPP voltage. The GW4000-SS, thanks to its wide temperature range of -20 to 60 C and the absence of any reduction in power, can therefore be installed at Manufacturer s response All versions of the SS series are based on identical firmware and the same essential components; the test results delivered by the GW4000-SS should therefore also apply for all the other devices. The precision of the power display can be improved by modifying the software. We are convinced that the deviation can be limited to less than 1 percent. We do not recommend the use of thin-film modules which are grounded on the DC side with this inverter. Other types of thin-film modules can be used. International December 2012 161
science & technology inverters test Weighted conversion efficiency The efficiency curves under European and Californian weighting representing medium and high levels of irradiation exhibit no significant differences. This copy is licensed to Atonex BV, Wimco Baauw. téáöüíéç=åçåîéêëáçå=éññáåáéååó= bìêç I= `b`=áå=b bìêç=ã~åìñ~åíìêéê=ëééåáñáéç =[=KQ=B VU VQ VO UU US UQ UO TU =bìêçéé~å=ïéáöüíéç==== bìêçj~ñ =Z=KP=B =`~äáñçêåá~å=ïéáöüíéç=== `b`j~ñ ==Z=KQ=B TS TQ TO Overall efficiency at different voltages The different MPP voltages have little impact as the corresponding diagram indicates on the efficiency curves. The PHOTON efficiency reflects this. lîéê~ää=éññáåáéååó= pìã =áå=b VU VQ VO UU US UQ UO TU TS TQ TO pìãj~ñ == =Z=KM=s=Eãáå pìãj~ñ =Z=KMQ=B == =Z=KR=s=E pìãj~ñj~ñ pìãj~ñ =Z=KTS=B == =Z=KM=s=Eã~ñ pìãj~ñ =Z=KPN=B ==^îéê~öé=çîéê~ää=éññáåáéååó = ^îöpìãj~ñ =Z=KQO=B mãéç =Z=KV=BI= müáöü =Z=KN=B R NM NR OM OR PM PR QM QR RM RR SM SR TR UR VR NMR NNM NNR NOM Accuracy of inverter display In the lower power range, the figures displayed by the GW4000-SS were rather erratic, but very consistent above 20 percent of nominal power albeit consistently 2 percent too high. aéîá~íáçå=áå=b PM OU OS OQ OO OM NU NS NQ NO NM U S Q O M JO JQ JS JU JNM JNO JNQ JNS JNU JOM JOO JOQ JOS JOU JPM R NM NR OM OR PM PR QM QR RM RR SM SR TR UR VR NMR NNM NNR NOM 162 International December 2012
This copy is licensed to Atonex BV, Wimco Baauw. just about any place in a building or outdoors, something which is also facilitated by its compliance to protection type IP 65 and its fan-free cooling system. Overload behavior: At an MPP voltage of 384 V (and an ambient temperature of 28 C), the inverter was presented with an overload of 1.3 times its input nominal power, which was still specified as 4,200 W at the time of testing. Only later was this corrected to 4,150 W, which is why 5,460 W (and not 5,395 W) were fed in. The inverter limited its DC power to 4,454 W. This is equivalent to a small overload range of 106.1 percent. When limiting power, the device pushed its operating point on the IV curve toward higher input voltages with the DC voltage adjusting itself to around 426.6 V. Own consumption and night consumption: The power consumed by the GW4000-SS in its basic tested state totaled around 0.2 W on the AC side and with severe fluctuations up to 15 W on the DC side. The manufacturer has not provided any information on this. At night, the inverter draws around 0.23 W of real power from the grid. The manufacturer specifies 0 W in this case. Thermography: Thermographic imaging shows the inverter from above while operating at an ambient temperature of 25 C and at nominal power, bearing in mind that due to the partially multi-layer configuration, its entire interior cannot be shown. Temperatures of up to 77.6 C developed on the circuit boards, these being recorded on a resistor on the balancing network circuit within the output bridge. On the right hand side of the power element circuit board, the grid phase cut-off reached a temperature of 75.4 C. The electrolytic capacitors in the power element were in a range between approximately 40 to 45 C. Summary The Goodwe GW4000-SS is a welldesigned inverter with a clear layout, making it production-friendly, with the entire circuitry with the exception of the control unit being housed on a large circuit board. Operation presented absolutely no problems during the test. An easyto-read display provides a large amount of information in a clearly presented form. The power output displayed by the inverter exhibits a relatively good level of accuracy. The device housing complies with protection type IP 65. The temperature range is very broad and can also be fully utilized. The power reduction, which the data sheet specifies as occurring at an ambient temperature of 45 C or more, did not materialize during testing. The dependency of the conversion efficiency on ambient temperature is, at 0.3 percentage points, low. The MPP voltage range specified by the manufacturer is not at a sufficient distance to the maximum DC voltage of the inverter, meaning it cannot be fully utilized by crystalline modules and thin-film modules. When it comes to the efficiency curve, there are, in contrast, virtually no voltage-related limitations: the conversion efficiency attains its highest value of 97.8 percent in the middle third of the MPP voltage range, however the difference of 0.6 percentage points between this and higher voltages is not substantial, just like the difference of around 0.7 percentage points toward lower voltages. The power dependency is somewhat more pronounced, but is effectively also low. Thanks to the very good MPPT adjustment efficiency, the overall efficiency curve is very similar to the conversion efficiency curve. The PHOTON efficiency for medium irradiation totals 96.9 percent, the figure for high irradiation is 97.1 percent. Both earn it an»a«. grade The difference between the PHOTON efficiency and maximum conversion efficiency, which only amounts to 0.9 percentage points, exemplifies the low voltage and power dependency of this inverter. The middle third of the MPP voltage range is optimal when configuring the MPP of a photovoltaic system. Only its unspectacular overload capability of 106.1 percent (albeit widespread among newer inverters) needs to be kept in mind when planning an array. The device completed the test as one of the best 20 inverters tested to date (see table, p. 65), and even scored a place among the top 10 inverters in the power class of up to 5 kw. urther information Contacts page 201 International December 2012 163