Inter national J our nal of P o wer Electr onics and Dri v e System (IJPEDS) V ol. 11, No. 3, Septemeber 2020, pp. 1388 1397 ISSN: 2088-8694, DOI: 10.11591/ijpeds.v11.i3.pp1388-1397 r 1388 Design and implementation of an optimized multile v el po wer in v erter structur e based on C MEX and PSPICE Mohammed Setti, Mohamed Cherkaoui Engineering for Smart and Sustainable Systems Research Center , Mohammed V Uni v ersity in Rabat, Morocco Article Inf o Article history: Recei v ed Feb 9, 2020 Re vised Mar 6, 2020 Accepted Mar 22, 2020 K eyw ords: C MEX S-function Multile v el in v erter PSPICE Switching frequenc y Symmetric/asymmetric ABSTRA CT In this paper , both symmetric and asymmetric operations for an optimized cascaded multile v el po wer in v erter (MLI) are thoroughly e xamined. While symmetric configurations are more suitable for A C dri v es due to their equal po wer sharing among the v arious le v els, the asymmetric topologies fulfill the higher number of po wer source combinations under the same po wer semiconductor switch count. Additionally , particular attention w as put on the design of this optimized topology in terms of both reliability and po wer ef ficienc y by managing redundant states and minimizing the number of po wer switch commutations. Furthermore, a fundamental switching frequenc y modulat ion (FSFM) is thoroughly described in C MEX programming language and then the resulting g ating signals are fed into the po wer circuit designed with PSPICE. By applying this co-simulation approach, the control design task is greatly simplified while achie ving adv anced analyses with more realistic electronic de vices. This is an open access article under the CC BY -SA license . Corresponding A uthor: Mohammed Setti, Mohammadia School of Engineers, Ibn-Sina A v enue, 765, Rabat, Morocco. Email: mohammedsetti@research.emi.ac.ma 1. INTR ODUCTION Multile v el po wer in v erters (MLIs) are considered as one of the most ef fecti v e w ay to generate high quality A C po wer . Besides a good harmonic profile and lo w electromagnetic interferences that a staircase w a v eform can pro vide, some partial adv antages related to design and reliability are directly obtained [1]. The po wer semiconductor switching de vices in use can w ork in syner gy to deli v er the desirable po wer le v el, which let each one of them participate only in a fraction of the total po wer magnitude. Moreo v er , some multile v el po wer in v erter structures can pro vide dif ferent switching combinations referred to as r edundant states which can be useful to program f ault-tolerant operations [2–5] monitored by switch f ault diagnosis algorithms [6]. Despite of all their e x c iting adv antages, MLI topologies come with a double-edged sw ord. The y require a lar ge number of po wer s emiconductor v alv es which penalizes both the o v erall po wer system ef ficienc y and reliability while meeting b udget constraints [7]. The situation tends to be more gloomy with the introduction of dedicated ICs for dri ving po wer switches, so additional losses should be e xpected and the call for sophisticated controllers becomes ine vitable. F or a ll these reasons and more, se v eral MLI topologies ha v e been de v eloped to address these shortcomings by limiting the number of their po wer switching de vices [8–10]. In practice, the cascaded topologies wit h multiple DC sources are considered as one of the most authoritati v e multile v el topologies currently a v ail able, the y ha v e been successfully tested in some rene w able J ournal homepage: http://ijpeds.iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 r 1389 ener gy systems for more than a decade. The y can be found either in symmetric or in asymmetric configuration depending on the magnitude of their v oltage sources. Asymmetric nature of DC sources w ould pro vide e xtra le v els to the output signals so as to enhance their harmonic profile. Ho we v er , in the most asymmetric structures, po wers deli v ered by the v arious le v els are quite di f ferent. This is a serious limitation of such configurations for applications in A C dri v es [11, 12]. Due to their nature, asymmetrical configurations ha v e unequal po wer sharing; this doesn’ t allo w the bridges to be easily replaced (as the high and lo w side bridges most lik ely to be made by dif ferent po wer de vices) and hence lose the adv an- tage of modularity [11]. Also, it goes without saying that the peak in v erse v oltage (PIV) is noticeably greater in asymmetric operation. The ne xt section deals with an optimized po wer in v erter topology under asymmetric op e ration. It has already been the subject of a pre vious w ork [13], b ut ne v ertheless, did neither stress an y aspect of po wer ef ficienc y or reliability of such po wer con v erter nor try to address the issue of the PIV withstanding. F or these reasons and more, this article w as fully re written, better arranged and thoroughly upgraded all the simulation results. Also, a ne w symmetric configuration is proposed on the basi s of what w as fulfilled, where redundant states in dif ferent operating modes required by f ault-tolerant applications are highlighted and commented. Afterw ards, the po wer ef ficienc y of the proposed MLI topology is e v aluated and compared to some state-of-the-art topologies. Furthermore, a guide to co-simulating po wer control systems using C programming language via Simulink MEX S-function under PSPICE en vironment is ne wly pro vided. This will bring more adv antages o v er con v entional approach adopted in such pre vious w ork and a v oid to manage t he dizzy number of Simulink function blocks. The generated digital signals can be transferred to the tar get embedded system or fed into the po wer circuit made from realistic electrical components under PSPICE. Results and discussion are carried out while rele v ant conclusions are highlighted at the end of this paper . 2. PR OPOSED PO WER INVER TER T OPOLOGY In contrast to some structures suggested in the literature that call for e xtra capacitors, diodes or transformers to clamp/bring the v oltage up to a suitable le v el, the proposed MLI topology depicted in Figure 1(a) (top) mak es use only of v oltage sources and switches (the Schottk y diode D S only tak e part of the dri ving circuitry of Figure 1(a) (base) to allo w asymmetric turn on and of f). Namely , for the symmetric scheme (where k = 1 ), there are four unidirectional ( S 1 ;j to S 3 ;j ) and tw o bidirectional ( S 2 ;j ) po wer switches supplied by tw o identical DC v oltage sources. The sam e frame w ork can be e xtended to pro vide asymmetric operation by ha ving the v oltage of the second DC source equals twice that of the first one ( k = 2 ). Notice that S 1 ; 1 and S 1 ; 2 share a common drain while S 3 ; 1 and S 3 ; 2 share a common source electrode. On the other hand, asymmetric arrangement with k = 3 is also possible b ut require that the location between the pair of switches S 1 ;j and S 2 ;j then between S 2 ;j and S 3 ;j are sw apped. This arrangement will not be considered here because it e xhibits the same performance as that with k = 2 . The current path of dif ferent operat ing modes of the proposed topology is depicted by Figure 1(b) while a li st of switch states is tab ulated in T able 1. In order to ensure that all the desirable v oltage le v els are produced and e v ery v oltage source is protected ag ainst short-circuits. The general rule can be then defined as follo ws: S i;j = f 0 ; 1 g 2 Y j =1 3 X i =1 S i;j = 1 : (1) Since the proposed structure are b uilt from unidirectional and bidirectional po wer swi tches, and because each topology comes with a number of redundant switching states, dif ferent switching scenarios are a v ailable. When one or more po wer switches within a gi v en po wer in v erter circuit f ail, either permanently or temporarily , the po wer deli v ered to the load will therefore be altered and causing a decrease in the o v erall system reliability . T o pre v ent that from happening, a f ault-tolerant application should be programmed by making use of a v ailable redundant switch states. F or ea ch configuration, the T able 2 summarizes these redundant states and pro vides the switching patterns accordingly . Also, some hints are gi v en to address the po wer consumption issue. Design and implementation of an optimized multile vel ... (Mohammed Setti) Evaluation Warning : The document was created with Spire.PDF for Python.
1390 r ISSN: 2088-8694 ( a ) ( b ) (a) ( a ) ( b ) (b) Figure 1. (a) Proposed po wer in v erter structure (top) and optocoupler and totem-pole g ate dri v er (base). (b) Current path of dif ferent operating modes under symmetric ( k = 1 ) and asymmetric ( k = 2 ) setups. k is the v oltage ratio T able 1. Switching combinations of the proposed MLI topology for symmetric and asymmetric operations le v el in p.u. symmetric asymmetric S 1 ; 1 S 1 ; 2 S 2 ; 1 S 2 ; 2 S 3 ; 1 S 3 ; 2 S 1 ; 1 S 1 ; 2 S 2 ; 1 S 2 ; 2 S 3 ; 1 S 3 ; 2 3 n/a n/a n/a n/a n/a n/a 0 1 0 0 1 0 2 0 1 0 0 1 0 0 0 0 1 1 0 1 0 1 1 0 0 0 0 1 1 0 0 0 1 ( r edundant state ) 0 0 0 1 1 0 n/a n/a n/a n/a n/a n/a 0 1 1 0 0 0 0 1 1 0 0 0 0 0 ( r edundant state ) 0 0 0 0 1 1 0 0 0 0 1 1 0 ( r edundant state ) 0 0 1 1 0 0 0 0 1 1 0 0 1 1 0 0 1 0 0 1 0 0 1 0 0 1 ( r edundant state ) 0 0 1 0 0 1 n/a n/a n/a n/a n/a n/a 2 1 0 0 0 0 1 0 0 1 0 0 1 3 n/a n/a n/a n/a n/a n/a 1 0 0 0 0 1 3. PO WER EFFICIENCY COMP ARISON WITH O THER T OPOLOGIES Mainly , there are tw o types of po wer losses dissipated within a semiconductor switching de vice, namely the conduction and the switching losses. The former loss es are associated with the drain-to-source O N resistance, while the latter are link ed up with the input/output parasitic capacitances C iss and C oss . Also and since all FETs are symmetrical de vices, the a v erage po wer conduction loss of the MOSFET is calculated as follo ws: P m = R ds ( ON ) 1 2 Z 2 0 i 2 ( t ) d t = R ds ( ON ) I 2 RMS ; (2) Where R ds ( ON ) is the drain-to-source O N -state resistance of the po wer MOSFET . The a v erage po wer conduction loss P m can be e v aluated to as the product of R ds ( ON ) and the square of the RMS drain current. Considering that at instant t , there are n MOSFETs in the current path, the a v erage v alue of the conduction po wer loss P c of the DC-to-A C con v erter can be formulated as belo w: P c = P m 2 Z 2 0 n ( t ) d t = P m N A V G : (3) Int J Po w Elec & Dri Syst, V ol. 11, No. 3, Septemeber 2020 : 1388 1397 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 r 1391 T able 2. Redundant switch states under symmetric and asymmetric operations of the proposed MLI Configuration Redundanc y State P attern Note Asymmetric 2 0 S 3 ; 1 + S 3 ; 2 Preferred (lo w po wer losses) ( k = 2) 0 S 2 ; 1 + S 2 ; 2 Not recommended (high po wer losses) Symmetric 4 V 1 S 2 ; 2 + S 3 ; 1 Identical performance ( k = 1) 0 S 3 ; 1 + S 3 ; 2 Suitable when either V 1 or + V 1 redundant state is used (lo w switching po wer losses) 0 S 2 ; 1 + S 2 ; 2 Not recommended (high po wer losses) + V 1 S 2 ; 1 + S 3 ; 2 Identical performance The switching po wer loss P s is equal to the sum of all turn-on and turn-of f ener gy losses in a fundamental c ycle of the output v oltage [14]. This can be written as follo ws: P s =   n X k =1 n on ;k X i =1 E on ; ki + n X k =1 n o ;k X i =1 E o ; ki ! f sw ; (4) where f sw is the switching frequenc y and n on ;k and n o ;k are the number of turn-on and turn-of f of the k th MOSFET during a fundamental c ycle. Also, E on ; ki is the ener gy loss of the k th MOSFET during the i th turn-on and E o ; ki is the ener gy loss of the k th MOSFET during the i th turn-of f. Since most MLI structures in v olv e high number of po wer switches, the reason to dri v e them by fundamental frequenc y modulation (FSFM) becomes ine vitable so as to alle viate the po wer switching losses, which mak es the conduction losses ( R ds ( ON ) ) dominate o v er those incurred by the switching transitions ( Q g and C oss ): 2 R ds ( ON ) I 2 RMS ( Q g V gs + C oss V 2 ds ) f sw ; (5) where Q g is the g at e char ge of the MOSFET , V gs is the RMS g ate-to-source v oltage, C oss is the MOSFET output parasiti c capacitance ( C gd + C ds ) and V ds is the RMS drain-to-source v oltage. The total l o s s P t is then the sum of the conduction and switching losses: P t = P c + P s P c : (6) T able 3. T otal blocking v oltage and acti v e po wer switches in dif ferent cascaded MLI topologies under FSFM T opology (a) (b) (c) (d) (e) (f) (g) (h) Proposed No. of DC sources 3 2 2 2 5 2 2 4 2 No. of bidirectional switches 0 0 1 6 1 2 0 2 2 No. of unidirectional switches 6 6 6 0 10 6 6 5 4 symmetric (5-le v el) No. of acti v e po wer switches n/a 24 28 32 n/a 28 n/a 26 20 T otal blocking v oltage in V 1 n/a 8 11 10 n/a 14 n/a 11 10 asymmetric (7-le v el) No. of acti v e po wer switches 36 36 n/a 48 72 44 36 40 32 T otal blocking v oltage in V 1 18 12 n/a 16 18 22 12 16 16 Most po wer ef ficient topologies cannot be e v aluated just in terms of ho w man y electronic parts the y w ould require, b ut in addition, the y should be assessed with re g ard to the total number of their acti v e po wer switches o v er one full-time period and their v oltage blocking capability . As it w as sho wn in T ables 3 and 4, the proposed MLI e xcels all the 8 structures depicted by Figure 2 both in terms of po wer ef ficienc y and total number of acti v e po wer switches. The total blocking v oltage comparison is also considered since switching de vices’ ratings are al w ays correlated to their maximum blocking v oltage. Moreo v er , the total harmonic distor - tion (THD) and indi vidual po wer dissipation in e v ery po wer swi tch are tab ulated in T able 4. Finally , it should be pointed out that all the bidirectional switches are assumed to be arranged by tw o back-to-back N-channel po wer MOSFETs. Design and implementation of an optimized multile vel ... (Mohammed Setti) Evaluation Warning : The document was created with Spire.PDF for Python.
1392 r ISSN: 2088-8694 T able 4. Comparison of THD, switch po wer dissipation (in W atts) and po wer ef ficienc y of selected highly ef ficient cascaded MLI topologies under FSFM symmetric asymmetric T opology 7.2% thd i , 17.9% thd v 4.0% thd i , 12.5% thd v S 1 ; 1 S 1 ; 2 S 2 ; 1 S 2 ; 2 S 3 ; 1 S 3 ; 2 P (%) S 1 ; 1 S 1 ; 2 S 2 ; 1 S 2 ; 2 S 3 ; 1 S 3 ; 2 P (%) (a) n/a n/a n/a n/a n/a n/a .599 .590 .606 .588 .599 .593 3.57 99.66 (b) .636 .651 .659 .663 .688 .684 3.98 99.63 .590 .601 .590 .598 .588 .604 3.57 99.66 (g) n/a n/a n/a n/a n/a n/a .588 .601 .590 .604 .591 .601 3.58 99.66 Pr oposed .630 .627 .258 .260 .485 .486 2.74 99.75 .448 .444 .396 .396 .554 .556 2.79 99.73 (a) (b) c Z ` (c) Z ` (d) (e) (f) Z ` (g) (h) Figure 2. Cascaded MLI topologies with reduced po wer switch count presented in literature: (a) from [16], (b) from [17], (c) from [18], (d) from [19], (e) from [20], (f) from [21], (g) from [7] and (h) from [22] 4. C MEX AND PSPICE CO-SIMULA TION This section is de v oted to co-simulate the proposed MLI using C MEX S-function and PSPICE program. Instead of using mathematical function blocks and graphically connecting them together to b uild system models, the user can rely on the system-functions called C MEX S-functions , which are defined as computer language description written in C/C++ programming language, to benefit from their capabilities to interact with the Simulink engine. Henceforth, this approach has the adv antage to not be cumbersome as the former does and requires less hardw are resources and time during e x ecution and maintenance. Ho we v er , only the switching control can be achie v ed that w ay , since Simulink uses only ideal models and non technology- specific de vices. On the contrary to other simulation programs lik e Simscape Electrical, PLECS and PSIM, PSPICE is an industry standard SPICE-based simulator and is primarily intended to model technology-specific electronic components so that to pro vide in-depth det ails about some realistic circuit beha viors during simulation. F or these reasons and more, the proposed MLI circuit is described by netlist. C MEX S-functions are based on the C language. The y must be compiled as MEX files using mex util- ity , so the resulting binary file is platform-dependent and dynam ically loadable e x ecutable utilized by Simulink program [15]. Figure 3(a) depicts the typical w orkflo w for C MEX compiling and inte grating into Simulink models. During the simulation, the Simulink engine must retrie v e information about the function used by the Int J Po w Elec & Dri Syst, V ol. 11, No. 3, Septemeber 2020 : 1388 1397 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 r 1393 C MEX S-function. As the simulation proceeds, the C MEX S-function, the Ordinary Dif ferential Equations (ODE) Solv er and the Simulink engine interact together to achie v e specific tasks. The interaction between C MEX S-function and Simulink engine is realized by in v oking callback methods that the S-function implements. In order to model the g ate switching patterns required to dri v e the proposed po wer in v erter , the engine should in v ok e sequentially the C MEX S-function callback methods as illustrated in the flo wchart of the Figure 3(b). The methods dra wn in black line are mandatory , whereas those sho wn in gray color are optional. Also, in the flo wchart gi v en by the Figure 3b, mainly three steps are in v olv ed: ( I ) initialization, ( I I ) calculation of switching outputs in the major time st ep, and ( I I I ) termination of the simulation. The mdlInitializeSizes() callback method aims to identify the width and data type of dif ferent input and output ports. The mdlInitializeSampleTimes() is then in v ok ed to specify both the of fset time and sample rate at which the S-function should operate. Afterw ards, at e v ery simulation time step, the mdlOutputs() method is called to compute the S-function’ s switching outputs by means of the Simulink engine and then dispatches the results as array elements. The block’ s core logic has to be implemented inside the body of this callback method. The FSFM algorithm follo ws this requirement, which in f act corresponds to one of the switching pattern set tab ulated in T able 1. As its name implies, the mdlTerminate() method is called to perform an y tasks required at the end of the simulation process lik e closing deb ugging files or freeing dynamic memory allocations. The generated FSFM g ating signals for dri ving the proposed MLI are based on the NLC scheme [8] and gi v en by Figure 4 under symmetric and asymmetri c operations. The rounding functions defined inside mdlOutputs() must handle the misleading number representation of half-inte gers by defining a proper tie-breaking rule such as round-half-to-e v en. These signals can either be fed into PSPICE or directly sent to the tar get hardw are via the Embedded Coder , which allo ws the code to be compiled and e x ecuted on an y processor . C-MEX sour ce file mex utility C compiler GC C ,MSV C ,... Binary MEX-file S-function b lock R efer ence Gating Signals Œ S i ; j  n m ODE solv er Sim ulink engine *.MEXW32/64 (WIN OS) T o SPICE via FSTIM MOSFET ga te dri v ers On tar get har d w ar e (Embed ded Coder) (a) mdlInitializeSizes() mdlSetInputPortFrameData() mdlInitializeSampleTimes() mdlSetInputPortDataType() mdlSetOutputPortDataType() mdlSetDefaultPortDataTypes() mdlOutputs() mdlTerminate() ( i ) ( ii ) ( iii ) (b) Figure 3. (a) T ypical w orkflo w for C MEX compiling and inte grating into Simulink models. The results can be either link ed to PSPICE or run on the tar get hardw are via the Embedded Coder . (b) C MEX S-function callback methods flo wchart used for modeling the switching controller S 1 ; 1 S 1 ; 2 S 2 ; 1 S 2 ; 2 S 3 ; 1 0 4 8 12 16 20 24 28 32 time in milliseconds S 3 ; 2 0 4 8 12 16 20 24 28 32 time in milliseconds sy mmetric (5- le vel ) asy mmetric (7- le vel ) Figure 4. Gating signals for the proposed MLI for both symme tric and asymmetric configurations under FSFM Design and implementation of an optimized multile vel ... (Mohammed Setti) Evaluation Warning : The document was created with Spire.PDF for Python.
1394 r ISSN: 2088-8694 The PSPICE file stimulus (FSTIM) de vices are aimed to pro vide digital signals to a gi v en circuit node through e xternal data file [23], so as to pro vide a link to the FSFM g ating signals. Another approach called PSPICE SLPS interf ace has already been suggested to pro vide co-simulation under Simulink [24]. This tool comes with good features, b ut still remain some dra wbacks that should be addressed: (a) The actual electronic block is substi tuted with an ideal model since the simulation of the po wer circuit is set up under Simulink en vironment; (b) The internal properties of PSPICE that ha v e smaller time steps than those of Simulink cannot be check ed. Also, some Simulink minor steps are skipped, and thus data is not fully e xchanged with PSPICE [25]; (c) The SLPS data e xchange is synchronized with Simulink steps. T o pre v ent o v erlooking phenomena from happening, Simulink maximum step size should be chosen carefully , so to ensure the best trade-of f between accurac y and simulation time; (d) Only one SLPS block per Simulink model is allo wed [25]. The number of interf aces an SLPS block can handle is limited. As reported by man y users, Simulink will crash once the limit is e xceeded; (e) Some l imitations are inherited from the f act that Simulink are generic simulation programs not tar geted for mix ed-signal electronic designs. In contrast to ho w the SLPS interf ace tool operates, the suggested co-si mulation approach is no w fulfilled thoroughly under PSPICE en vironment, so as to pre v ent the idealization of the used electronic de vices. Consequently , the o v erlooking and data e xchange issue are no longer a problem as soon as the g ating signals retrie v ed from Simulink are correctly processed and adapted to the requirements of PSPICE stimulus de vices. The dri ving of the dif ferent po wer MOSFETs of Figure 1(a) (top) is done by the A v ago HCPL-3140 optocoupler based dri v ers sho wn by Figure 1(a) (base). Ho we v er , each non-in v erting optical ly isolated totem- pole g ate dri v er requires one floating po wer supply of 12 v olts with respect to the ground node V EE , so as to be able to dri v e high-side N-channel po wer MOSFETs. Ad di tionally , a 100 nF bypass capacitor C D must be added to stabilize the operation of the high g ain linear amplifier , otherwise, the switching functionality may be compromised. Finally , PSPICE simulator needs at least one ground or node zer o to reference all the calculated v oltages, so in order to isolate the control side of the circuit from the po wer side and distinguish between the tw o side grounds, a resistance R 1 with high v alue is used to pre v ent floating node error from happening. Moreo v er , the load Z ` represents an inducti v e load with a DPF = 98 % lagging formed by an R ` = 50 in series with an L ` = 32 mH . All the po wer MOSFETs operate at fundamental frequenc y (50 Hz) and are under h a rd switching condition. Lo w switching po wer losses are al w ays correlated with f ast commutations which are the major reason for electromagnetic interference and v oltage spik e a cross the drain and source terminals due to the e xistence of stray inductances. Additionally , the switching speed v aries with respect to the v alue of the g ate resistor R G which w ould mak e the selection more dif ficult in order to a v oid the o v erlap time that will cause lar ge current sur ge. Both Figures 5(d) and 5(e) present some sur ges in the drain current of the MOSFETs on each column when the e xternal g ate resistors R 2 and the Schottk y diodes D S are omitted from the dri ving circuit of Fig- ure 1(a)(base). As it can be seen, the maximum rating allo wed for the pulsed drain current is e xceeded in case of a 600V CoolM OS CFD7 N-channel po wer MOSFET , so in order to get rid from this situation, each po wer switch should be dri v en through dif ferent g ate resistors to allo w asymmetric turn on and turn of f. Figure 5(f) and 5(g) highlight, respecti v ely , the di f ference by plotting the g ate-to-source v oltage V GS for both S 2 ; 2 / S 3 ; 2 S 1 ; 1 / S 2 ; 1 that ha v e been equi pped with g ate resistors on top of those which ha v e no additional g ate resistors. As the Figures 5(h) and 5(i) suggest, the drain current spik es through the six po wer switches are no w mostly v an- ished. The A C output v oltage and current of the proposed 5- and 7-le v el po wer in v erters are illustrated together through the Figure 5(a) under an inducti v e load with a DPF 0 : 98 lagging. Additionally , the v oltage/current harmonic profiles for both c o nfi gu r ations are also included, respecti v ely , by Figures 5(b) and by Figures 5(c), wherein their THD are calculated separately within one complete time period ( 20 ms ). 0 5 10 15 20 320 V 0 C 320 V l o ad v ol t a ge V ` TIME IN MILLISECONDS (a) 8 A 0 C 8 A l o ad current I ` (cos 0 : 98 ) 0 5 10 15 20 0 l o ad current I ` (cos 0 : 98 ) TIME IN MILLISECONDS 320 V 0 C 320 V l o ad v ol t a ge V ` (a) Figure 5 Int J Po w Elec & Dri Syst, V ol. 11, No. 3, Septemeber 2020 : 1388 1397 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 r 1395 0 0 : 5 1 1 : 5 2 2 : 5 3 3 : 5 4 4 : 5 5 0 336 V 37 V thd 17 : 9 % y - scale not linear 0 0 : 5 1 1 : 5 2 2 : 5 3 3 : 5 4 4 : 5 5 0 6 : 6 A 0 : 3 A thd 7 : 2 % y - scale not linear FREQ IN KIL OHER TZ (b) (b) 0 0 : 5 1 1 : 5 2 2 : 5 3 3 : 5 4 4 : 5 5 0 330 V 20 V thd 12 : 5 % y - scale not linear 0 0 : 5 1 1 : 5 2 2 : 5 3 3 : 5 4 4 : 5 5 0 6 : 5 A 0 : 2 A thd 4 % y - scale not linear FREQ IN KIL OHER TZ (c) (c) 0 5 10 15 20 150 A 0 C 150 A S 1 ; 1 S 1 ; 1 S 1 ; 1 S 2 ; 1 S 2 ; 1 S 2 ; 1 = S 3 ; 1 S 2 ; 1 (d) (d) 20 0 5 10 15 20 150 A 0 C 150 A S 1 ; 2 S 1 ; 2 S 1 ; 2 S 2 ; 2 = S 3 ; 2 S 2 ; 2 S 2 ; 2 S 2 ; 2 (e) (e) 1 ; 666 1 ; 667 1 ; 668 0 C 15 V V T H (f) C (f) 668 3 ; 135 3 ; 136 3 ; 137 0 C 15 V V T H (g) (g) 0 5 10 15 20 10 A 0 C 10 A TIME IN MILLISECONDS (h) (h) 0 5 10 15 20 10 A 0 C 10 A TIME IN MILLISECONDS (i) (i) Figure 5. Output w a v eforms/spectra and switching characteristics of the proposed MLI with an inducti v e load ( DPF 0 : 98 ). (a) Load v oltage V ` and current I ` w a v eforms in symmetric/asymmetric configura tion, respecti v ely . (b) Load v oltage/current spectrum under symmetric setup. (c) Load v oltage/current spectrum under asymmetric setup (The y-axis scale of fundamental spectrum and harmonic spectra is nonlinear). Drain current through the po wer MOSFETs where no g ate resistor ( R 2 = 0 ): (d) S 1 ; 1 ( ), S 2 ; 1 ( ) and S 3 ; 1 ( ). (e) S 1 ; 2 ( ), S 2 ; 2 ( ) and S 3 ; 2 ( ). V oltage across the g ate and source terminals of the po wer MOSFETs: (f) S 2 ; 2 ( / ) and S 3 ; 2 ( / ) with/without g ate resistors. (g) S 1 ; 1 ( / ) and S 2 ; 1 ( / ) with/without g ate resistors ( V T H is the g ate threshold v oltage of the 600V CoolMOS CFD7 po wer MOSFET). Drain current when g ate resistors and Schottk y diodes are added: (h) S 1 ; 1 ( ), S 2 ; 1 ( ) and S 3 ; 1 ( ). (i) S 1 ; 2 ( ), S 2 ; 2 ( ) and S 3 ; 2 ( ) 5. CONCLUSION In this article, the symmetric and asymmetric configurations of the proposed MLI were compared to some well-established topologies in the literature. It has sho wn that the proposed structure accomplishes the highest po wer ef ficienc y scor e ( 99 : 73 99 : 75% ) by restraining the number of acti v e po wer switches per c ycle. Moreo v er , the proposed symme tric topology e xhibits lo w total blocking v oltage and se v eral redundant switch states, which w ould bring additional enhancement to the o v erall system reliability . Alternati v ely , asym- metric structure pro vides a good harmonic profile (about 4% THD i and 12.5% THD v ) which positi v ely impact the A C side lo w-pass filter in order to comply with the international standards in terms of po wer quality . Design and implementation of an optimized multile vel ... (Mohammed Setti) Evaluation Warning : The document was created with Spire.PDF for Python.
1396 r ISSN: 2088-8694 Afterw ards, the FSFM w as described i n C MEX and run thr o ugh a single Simulink S-function block. This w ould bring significant enhancement in de v elopment and simulation speed and mak e maintenance ef fort- less when compared to the con v entional approach based on graphical modeling. The generated g ating signals can be either sent to the embedded hardw are or inte grated as digital stimuli through the PSPICE file stimulus de vice. This technique has some adv antages o v er the SLPS interf ace t o ol especially when simulating open- loop SMPS topologies and mak e use of m o r e realistic electronic de vices pro vided by almost e v ery electronics manuf acturers. REFERENCES [1] M. V ijeh, et al., A general re vie w of multile v el in v erters based on main submodules: structura l point of vie w , IEEE T r ans P ower Electr on , v ol. 34, no. 10, pp. 9479-9502, 2019. [2] H. Khounjahan, et al., “Reconfigurable multile v el in v erter with f ault-tolerant ability , IEEE T r ans P ower Electr on , v ol. 33, no. 9, pp. 7880-7893, 2018. [3] M. Jalhotra, et al., “De v elopment of f ault-tolerant MLI topology , IET P ower Electr on , v ol. 11, no. 8, pp. 1416-1424, 2018. [4] N.K. De w ang an, et al., Approach to synthesis of f ault tolerant reduced de vice count m u l tile v el in v erters (FT RDC MLIs), IET P ower Electr on , v ol. 12, no. 3, pp. 476-482, 2019. [5] K.K. Gupta, et al., “Multile v el in v erter topologies with reduced de vice count: a re vie w , IEEE T r ans P ower Electr on , v ol. 31, no. 1, pp. 135-151, 2016. [6] T .G. Manjunath and A. K usagur , Analysis of dif ferent meta heuristics method in intelligent f ault detec- tion of multile v el in v erter with photo v oltaic po wer generation source, Int J P ow Elec & Dri Syst , v ol. 9, no. 3, pp. 1214-1222, 2018. [7] E. Babaei, et al., A ne w general topology for cascaded multile v el in v erters with reduced number of components based on de v eloped H-bridge, IEEE T r ans Ind Electr on , v ol. 61, no. 8, pp. 3932-3939, 2014. [8] N. Prabaharan and K. P alanisamy , A comprehensi v e re vie w on reduced switch multile v el in v erter topolo- gies, modulation techniques and applications, Rene w Sust Ener g Re v , v ol. 76, pp. 1248-1282, 2017. [9] E. Samadaei, et al., “Ne w h ybrid structure for multile v el in v erter with fe wer number of components for high-v oltage, IEEE T r ans P ower Electr on , v ol. 33, no. 2, pp. 987-996, 2018. [10] J. Singh, et al., “Buck con v erter -based cascaded asymmetrical multile v el in v erter with reduced compo- nents, Int T r ans Electr Ener g Syst. , v ol. 28, no. 3, 2018. [11] S.K. Chattopadh yay and C. Chakrabort y , A ne w multile v el in v erter topology with self-balancing le v el doubling netw ork, IEEE T r ans Ind Electr on , v ol. 61, no. 9, pp. 4622-4631, 2014. [12] V .R. Nair , et al., A V ery High Resolution Stack ed Multile v el In v erter T opology for Adjustable Speed Dri v es, IEEE T r ans Ind Electr on , v ol. 65, no. 3, pp. 2049-2056, 2018. [13] M. Setti, et al., “Ne w po wer ef ficient se v en-le v el in v erter topology controlled by fle xible nearest le v el modulation, 3r d International Confer ence on Electrical and Information T ec hnolo gies , 2017. [14] S.T . Meraj, et al., “Single phase 21 le v el h ybrid multile v el in v erter with reduced po wer components emplo ying lo w frequenc y modulation technique, Int J P ow Elec & Dri Syst , v ol. 11, no. 2, pp. 810-822, 2020. [15] L. Zamboni, “Getting started with Simulink, P ac kt Publishing Ltd. , ISBN 978-1-78217-138-6, pp. 91- 109, 2013. [16] M. Ahmed, et al., Asymmetric cascaded half-bridge multile v el in v erter without polarity changer , J Ale xandria Eng , v ol. 57, no. 4, pp. 2415-2426, 2018. [17] M.F . Kang arlu and E. Babaei, A generalized cascaded multile v el in v erter using series connection of sub-multile v el in v erters, IEEE T r ans P ower Electr on , v ol. 28, no. 2, pp. 625-636, 2013. [18] S.P . Gautam, et al., A single-phase v e-le v el in v erter topology with switch f ault-tolerance capabilities, IEEE T r ans Ind Electr on , v ol. 64, no. 3, pp. 2004-2014, 2017. [19] E. Babaei E, et al., “Reduction of dc v oltage sources and switches in asymmetrical multile v el con v erters using a no v el topology , Electric P ower Systems Resear c h , v ol. 77, no. 8, pp. 1073-1085, 2007. [20] A.K. Y ada v , et al., A h ybrid se v en le v el in v erter topology formed by cas cading t-type and acti v e neutral point clamped in v erter for induction motor dri v es, 44th Annual Confer ence of IEEE Industrial Electr on- ics Society , pp.4423-4428, 2018. Int J Po w Elec & Dri Syst, V ol. 11, No. 3, Septemeber 2020 : 1388 1397 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 r 1397 [21] M.T . Khosroshahi, “Crisscross cascade multile v el in v erter with reduction in number of components, IET P ower Electr on , v ol. 7, no. 12, pp. 2914-2924, 2014. [22] E. Babaei, et al. A ne w topology for cascaded multile v el in v erters with reduced number of po wer elec- tronic switches, 7th P ower Electr onics, Drive Systems and T ec hnolo gies Confer ence , pp. 165-170, 2016. [23] Cadence Design Systems Inc., PSPICE Refer ence Guide , 2010. [24] O.A. Ahmed and J.A.M. Bleijs, PSPICE and Simulink co-simul ation for high ef ficienc y DC-DC con- v erter using SLPS interf ace softw are, 5th IET International Confer ence on P ower Electr onics, Mac hines and Drives, 2010. [25] Cadence Design Systems Inc., PSPICE SLPS Interface User Guide , 2016. Design and implementation of an optimized multile vel ... (Mohammed Setti) Evaluation Warning : The document was created with Spire.PDF for Python.