Identification of a 7-phase claw-pole - for a micro-hybrid automotive application Antoine Bruyère, Thomas Henneron, Eric Semail, Fabrice Locment, Alain Bouscayrol, J.M Dubus„ Jean-Claude Mipo

To cite this version:

Antoine Bruyère, Thomas Henneron, Eric Semail, Fabrice Locment, Alain Bouscayrol, et al.. Iden- tification of a 7-phase claw-pole starter-alternator for a micro-hybrid automotive application. Inter- national Congress on Electrical Machines, Sep 2008, Vilamoura, Portugal. pp.1-6, ￿10.1109/ICEL- MACH.2008.4800046￿. ￿hal-01107781￿

HAL Id: hal-01107781 https://hal.archives-ouvertes.fr/hal-01107781 Submitted on 3 Feb 2015

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Science Arts & Métiers (SAM) is an open access repository that collects the work of Arts et Métiers ParisTech researchers and makes it freely available over the web where possible.

This is an author-deposited version published in: http://sam.ensam.eu Handle ID: .http://hdl.handle.net/10985/9255

To cite this version : Antoine BRUYÈRE, Thomas HENNERON, Fabrice LOCMENT, Eric SEMAIL, Alain BOUSCAYROL, Jean-Marc DUBUS, Jean-Claude MIPO - Identification of a 7-phase claw-pole starter-alternator for a micro-hybrid automotive application - In: International Congress on Electrical Machines, Portugal, 2008-09-06 - International Congress on Electrical Machines - 2008

Any correspondence concerning this service should be sent to the repository Administrator : [email protected] Identification of a 7-phase claw-pole starter-alternator for a micro-hybrid automotive application

A. Bruyère1,2, T. Henneron1, E. Semail1, F. Locment1 A. Bouscayrol1, J.M. Dubus2, J.C. Mipo2

1Laboratoire d’Electrotechnique et d’Electronique de Puissance de Lille (L2EP), 2Valeo Electrical System, Arts et Métiers ParisTech, 8, boulevard Louis XIV, 2, rue André Boulle, BP 150, 59046 Lille, FRANCE 94017 Créteil Cedex, FRANCE Tel. +33 (0)3-20-62-15-61, Tel. +33 (0)1-48-98-84-37 Fax. +33 (0)3-20-62-27-50 Email: [email protected] Email: [email protected]

Abstract- This paper deals with the identification of a new high to develop the during the short duration of the start. In power starter-alternator system, using both: a Finite Element section II, the predeterminations obtained by a Finite Element Method (FEM) modeling and an experimental vector control. The drive is composed of a synchronous 7-phase claw-pole machine Method (FEM) modeling are compared with experimental supplied with a low voltage / high current Voltage Source Inverter results for the electromotive forces and for the torque per (VSI). This structure needs specific approaches to plan its Ampere. electrical and mechanical behaviors and to identify the In section III, resistive and inductive parameters of the parameters needed for control purpose. At first, a Finite Element whole drive are determined. These parameters are needed for Method (FEM) modeling of the machine is presented. It is used for the predetermination of the electromotive forces and of the controlling the starter-alternator, in its both functioning modes: torque. Experimental results are in good accordance with in motor mode (for the ICE start) and in alternator mode. As numerical results. In a second part, resistive and inductive the voltage is low (12V), the resistive and inductive electrical parameters of the drive are determined by an original parameter values of the battery and VSI are not negligible in experimental approach that takes into account each component of comparison with those of the machine. At first, results obtained the drive: the battery, the VSI and the machine. with classical determination approach are given and discussed. Secondly, an experimental approach based on an elementary I. INTRODUCTION vector control of the machine is developed to find directly the Claw-pole synchronous machines with separate six characteristic time constants needed for controlling the are commonly used to achieve the alternator function in seven-phase drive. Automotive. This is due to their low cost and their ability to II. DETERMINATION OF EMF AND TORQUE BY FINITE work in a very large speed range. With the studied belt driven ELEMENT METHOD starter-alternator system [1], starting the car Internal Combustion Engine (ICE) is made with this claw-pole This kind of claw-pole structure is known to be sensitive to machine. This kind of system is used to bring the micro-hybrid modeling with a numerical method as FEM [4] because of their Stop-Start function, without major modifications of the car 3D characteristics, a thin airgap (about 0.3% of the external powertrain. The new needed starter function adds a new diameter) and highly saturated magnetic materials. Moreover, constraint on the electrical machine: the ability to develop a the studied machine is a synchronous 7-phase claw pole large torque during the start. Starter- already equip machine with permanent between the claws. Sixteen small cars, with a small 1.4 ICE. The major interest of this poles and a number of slots per pole and per phase equal to simple system concerns the limited extra cost for the final car 0.25 imply a modelling of a quarter of the structure at least [1], when the hybridization operation is achieved. (Fig. 2). The aim of the modeling is not only to plan the To take up this challenge for powerful Internal Combustion voltage output, in generator mode, but also to determine the Engine (ICE) without changing the classical (12 Volts) DC- maximum available torque in motor mode, during the start of Bus voltage level [1], a new claw-pole starter-alternator with the car ICE. The first step in order to validate the 3D-modeling seven phases and permanent magnets between the claws has is to compare the experimental electromotive forces (emf) with been developed [2]-[3] (Fig. 1). The obtained torque density the calculated ones. The second step is to obtain, for a defined (Nm/m3) allows keeping the economical benefits of using a 12- repartition of the currents among the phases, the maximum Volts battery [1]. possible torque. In order to carry out the starter function, a modeling of the The thin airgap leads to a mesh composed of 380000 machine is at first necessary to plan the ability of the machine tetrahedral elements, the air-gap being meshed with two layers

 of elements. The resolution is made using the magnetostatic (3/4), assumption and the scalar potential formulation, with with concentrated windings CARMEL, a software developed at the L2EP. Fig.2 gives front claw-pole wheel results of the numerical simulation, related to the magnetic flux permanent magnets density B under no load condition and a 5A current IF (between the claws) supplying the excitation coil. The highest flux density values (in yellow) are obtained under the stator teeth. Fig. 3 compares the electromotive force waveform named e(t), calculated under no load condition, at a rotation speed N = 1800rpm and IF = 5A, with an experimental measurement in the same conditions. These results validate the numerical model good accuracy, at the first order, for the determination of the voltage outputs. Fig. 4 gives the calculated torque T as function of the rotor excitation coil position θ, when the machine currents are imposed to be the rotor rear claw-pole wheel same as in an experimental test (Square Wave Voltage supply), for which the torque is maximized. The maximal numerical Fig.1. Exploded view of the multiphase claw pole starter-alternator torque value is Tcalculated = 62.4 Nm, to be compared with the experimental maximum measured torque: Tmeasured = 66.5 Nm, that shows a difference of about 6 %. For the determination of the inductive parameters, numerical FEM leads to too large durations of computation in the present case. It is due to the 3D characteristics of the machine and its thin air-gap (which implies a high number of 380000 elements). Moreover, the magnetic materials being highly saturated for this application, the simulation has to take into account the non-linear characteristics of the materials. In Fig. 5, the magnetic flux density is represented for a 2D slice of the whole structure (rotor and stator); z-coordinate of the slice is at the middle of the structure. This figure points out the magnetic saturation areas under no load conditions, with iF = 5 A. These Permanent saturated areas evolve with the rotor position and the currents

(excitation coil current iF, and stator windings currents). They Fig. 2. Rotor magnetic flux density B under no-load condition with iF=5A are at the origin of variable magnetic air-gap effects. As consequence, the inductances depend on the rotor position and 15 on the currents. Finite Element modeling to calculate the 10 inductive parameters for every rotor position, and for every 5 supplying situation, would lead to huge computation and analysis times. 0 0,002 0,0025 0,003 0,0035 0,004 0,0045 0,005 0,0055 0,006 0,0065

As consequence, experimental methods have been developed -5 for the determination of the machine inductive parameters. (V) voltage EMF -10 experimental measurement Numerical calculation -15

Fig. 3. Electromotive force under no load condition; N=1800rpm, IF = 5A and comparison with an experimental measurement 80 60 40 20 0 0 50 100 150 200 250 300 350 -20

Torque (Nm) -40 Numercal calculated torque -60 Experimental maximal torque reference value -80 Fig. 4. Calculated torque as function of the position θ and comparison with an experimental maximal torque reference value with [Flux] the seven-dimensional vector of flux linked by the seven stator phases, [Is] the associated current vector and [LSS] the 7X7 stator inductance matrix. Several assumptions are made. At first, the magnetic characteristic of the materials is considered to be linear. Secondly, whereas for a seven-phase machine the [LSS] stator inductance matrix contains 49 values, only 7 different values, one self inductance and six mutual inductances, are considered taking into account standard assumptions of regularity, symmetry and circularity properties for this matrix. These seven inductances lead to six cyclic inductances and one homopolar inductance (or zero-sequence inductance). This last one is omitted since a wye-coupling of the machine implies that the homopolar inductance has no impact on the control. Fig. 5. Magnetic flux density repartition at the middle of the machine The procedure of measurement is the following for different under no load condition and iF=5A values of the rotor position, which is imposed and kept constant by a motion controller: III. INDUCTIVE AND RESISTIVE PARAMETERS IDENTIFICATION  a magnetic state is imposed by a choice of a value BY EXPERIMENTAL VECTOR CONTROL for the excitation current iF;  sinusoidal current at chosen frequency is imposed In order to control the drive during transient operations, such in phase n°1 by using a VSI with a control of the as starts, the electrical time constants are relevant parameters currents; since they imply the choice of the carrier frequency of the  voltage is measured across each phase successively; Pulse Width Modulation (PWM) for the Voltage Source  measured voltage is integrated. Flux and Inverter (VSI). For a wye-connected three-phase machine with corresponding inductance are then deduced using reluctance effects, only two time constants are sufficient to (1). characterize the drive. As the resistance is the same in the two The following measurements are made for several fixed axis, the various values of the time constants are directly positions. Then, Table 1 gives average values of inductances. associated with those of the cyclic inductances. In comparison Considering average values is in accordance with the with conventional methods [5], more recent methods have been circularity assumption used for this first method. presented using current controls, either in the stator frame [6], Practically, it has been chosen a 120 Hz frequency with a 17 or in the dq-reference frame [7]-[9]: determination of the A RMS sinusoidal current, around a constant 18 A current. cyclic inductances L and L can be thus obtained either d q Table I gives the average values (in µH) versus  for two indirectly or directly. values of excitation currents. For a 7-phase machine, three different dq-reference frames, or “subspaces S1, S2 and S3”, can be determined. TABLE I INDUCTANCES VALUES (µH) IN THE STATOR FRAME Consequently, six time constants can be then evaluated, depending on the importance of the reluctance effects. These time constants are needed for the control. iF=0A 45 -3 -7 -3 -3 -7 -3 Two methods, each one based on current control, are iF=5A 39 -3 -6.5 -3 -3 -6.5 -3 presented and compared in the next paragraphs. The experimental set-up is described with Appendix A2. It shows With standard assumptions of symmetry and circularity for the starter-alternator directly mechanically connected to a the [LSS(θ)] stator inductance matrix, the cyclic inductances in brushless machine, used to simulate the ICE behavior. The the three d-q reference frames of the seven-phase machine are seven-leg VSI is supplied by an electronic programmable obtained by (2). In (2), L0 is the zero-sequence inductance. The source. A programmable electronic load is used in parallel with six other terms are considered by pairs, in two dimensional the voltage source in order to simulate the on-board electrical subspaces S1, S2 and S3 defined with the generalized network energy consumptions. The whole system is managed Concordia transformation (denoted with the matrix [C] in (2), using a modular dSPACE DS1006 real time control board. cf. Appendix A1). Table 2 gives the inductive parameters once the generalized Concordia transformation has been achieved. A. Determination of resistance and inductance in the stator frame Considering separately these three subspaces, there is no The first approach is based on a simplified model. Starting difference between the inductances along the d- and q-axes. from the inductances definition (1): This is due to the circularity assumption of the inductance [Flux] = [LSS] [Is] (1) matrix (variable airgap effects are not taken into account for this first method). Table II gives average cyclic inductance values versus . inductive parameters and time constants. With the next

 L0  0 0 0 0 0 0  approach it is possible to determine directly the cyclic     0 L 1dS 0 0 0 0 0  inductances in conditions closed to the working ones.   00 L 1qS  0 0 0 0 (2)   B. Direct determination of time constants in dq-frames  00 0 L 2dS  0 0 0   00 0 0 L  0 0  In the studied case, since the values of resistances and  2qS   00 0 0 0 L 3dS  0  inductances are very low, it is then preferable to consider a    00 0 0 0 0 L 3qS   direct method of identification that takes into account 1 C LSS  )(  C  implicitly the complete drive and gives directly the time constants [10]. The experimental method is based on an TABLE II CYCLIC INDUCTANCE VALUES (µH) elementary vector control for a seven-phase machine. The control is a generalization of classical vector control in dq- I F 0 S1d S1q S2d S2q S3d S3q reference frame developed for three-phase machines. In this 0A 19 50 50 55 55 43 43 case, two Ld and Lq parameters are obtained for a machine with 5A 14 44 44 48 48 38 38 reluctance effect (leading to 2 time constant values, on d- and q-axes). For a 7-phase machine, three different dq-reference Finally, with the knowledge of these cyclic inductances it is frames, or “subspaces S1, S2 and S3”, can be determined. easy to determine the time constants, dividing the inductance Consequently, six time constants can be then evaluated. The by the resistance. However, using the resistance value to principle of the control is given Fig. 6. In each subspace, there calculate the time constants values introduces a new is a control of the dq-currents. The currents controllers are uncertainty as it is detailed below. Proportional Integral controllers with compensation of the A first approximation of the phase resistances value (RS), electromotive force. Constant current references are used using the stator windings geometrical characteristics, gives RS except in the channel where the cyclic inductance must be  = 7.6 m Measurement with a digital RLC meter bridge using identified. In this channel a simple Proportional controller is a low current level (0.5 A) gives a confirmation for this used with a square reference added to the constant reference.  winding resistance value: 8 m at low frequency. However, The PI controllers in the different channels allow ensuring a these values concern only the machine windings. Such low decoupling with the identified channel. The error and the resistance value implies the necessity to take into account all closed loop time constant relative to the response to the square the parasitic resistances usually neglected: resistance of the reference allow obtaining resistance and cyclic inductance. VSI MOS transistors and resistance at the electrical Fig. 7, gives experimental results of the time constant connections. This uncertainty will be confirmed by the final measurements. These parameters depend on the excitation results of this paper, using the second approach, for which the current. For one of the 3 dq-reference frames, the S1-subspace,  total resistance value is about 21,7 m . two different values of time constants, on d- and q- axes, can If this first method is theoretically easy to understand and to be clearly identified: the difference can be explained by a use, it must be also noticed that the values of inductances are variable reluctance effect in the machine. For the two others  weak (a few H, especially the mutual terms of [LSS]). So, the dq-reference frames, only one inductive parameter appears. induced voltages are only a few Volts. Consequently, parasitic These results are compared in table III and table IV with those voltage drops induce uncertainties on the measurements. obtained in table II for two values of IF. Moreover, the use of Concordia [C] transformation matrix For S2 and S3 subspaces the error is satisfying (max 26 %). implies that each determined cyclic inductance is obtained by For S1 subspace, it appears a difference between Ld and Lq the sum of six ones: as consequence there is an accumulation values. Indeed, the variable reluctance effect was not taken into of errors on the measurements with this method. account by the previous method because only average If it will be observed in the following paragraph that the inductances values versus  angle were considered. The new results are acceptable for average values, it must be pointed out approach confirms that a reluctance effect is quite noticeable that with this approach it is not possible to take into account for the S1 subspace. If the average value obtained by the the effect of all the stator currents on the values of the first method is compared to the average value (LS1-d+ LS1-q)/2 inductances. Fundamentally the method is based on an obtained by the last approach, the relative errors are 12% for assumption of magnetic linearity. Basically it is thus not IF=0A and 27% for IF=0A. The results are coherent. possible to impose to the machine a magnetic state representative of real working conditions of the drive. Nevertheless, obtained values allow a first estimation of ,9 &21&/86,21 Y6GT UHI L6GT UHI 6FRQWUROOHUZLWK   5 ș 6 Y6Įȕ UHI  FRPSHQVDWLRQRIIHP L $PL[HGDSSURDFKKDVEHHQSUHVHQWHGIRUWKHLGHQWLILFDWLRQ ,QYHUVH YUHI L6GT UHI 3:0 6FRQWUROOHUZLWK  &RQFRUGLD  5 ș 6 RIDVHYHQSKDVHVWDUWHUJHQHUDWRU7KH)(0PRGHOLQJDOORZV  FRPSHQVDWLRQRIIHP WUDQVIRUPDWLRQ 96, 60 & YUHI L6GT UHI L 6FRQWUROOHUZLWK  Y  ș ȍ SODQQLQJ WKH PDLQ RXWSXW YDOXHV RI WKH PDFKLQH DV  5 ș 6 6Įȕ UHI  FRPSHQVDWLRQRIIHP Y6GT UHI HOHFWURPRWLYH IRUFH DQG WRUTXH 7KLV UHVXOW LV SDUWLFXODUO\ H6GT H 6GT )(0 LQWHUHVWLQJ IRU WKH GHVLJQ VWHS $ FRPSDULVRQ ZLWK HVWLPDWRU H6GT H[SHULPHQWDO UHVXOWV VKRZV D JRRG DFFXUDF\ RI WKH QXPHULFDO PRGHO IRU WKH GHWHUPLQDWLRQ RI WKHVH YDOXHV )RU SDUDPHWHUV

5 ș L6GT 6 L6Įȕ ZKLFKDUHPRUHFKDUDFWHULVWLFRIWKHZKROHGULYHVXFKDVWLPH &RQFRUGLD FRQVWDQWV LW KDV EHHQ VKRZQ WKDW WKH GHWHUPLQDWLRQ RI 5 ș L6GT 6 L6Įȕ WUDQVIRUPDWLRQ & LQGXFWDQFHVLQWKHVWDWRUIUDPHOHDGWRXQFHUWDLQWLHVLQWKHFDVH

5 ș 6 L L6GT 6Įȕ RI D YHU\ ORZ YROWDJH VDWXUDWHG PDFKLQH $Q H[SHULPHQWDO )LJ6FKHPHRIWKHYHFWRUFRQWUROXVHGIRUWKHVHYHQSKDVHPDFKLQH DSSURDFK XVLQJ D YHFWRU FRQWURO KDV EHHQ FRQVHTXHQWO\ SURSRVHGLQRUGHUWRREWDLQGLUHFWO\WKHUHTXLUHGWLPHFRQVWDQWV 7$%/(,,, IRUWKHGHVLJQRIWKHGULYH &203$5,6212)$9(5$*(&<&/,&,1'8&7$1&(6)25,) $ —+ 9 5()(5(1&(6 /6G! /6T! /6G! /6T! /6G! /6T! )LUVW >@ ' 5LFKDUG < 'XEHO ³9DOHR 6W$56 7HFKQRORJ\ $ FRPSHWLWLYH       6ROXWLRQ IRU +\EULGL]DWLRQ´ 3RZHU &RQYHUVLRQ &RQIHUHQFH 3&&¶ PHWKRG 1DJR\D-DSDQSS$SULO,6%1; 6HFRQG       >@ -0'XEXV$'H9ULHV'(YHQDQG-&0LSR3RO\SKDVHVWDWRURI PHWKRG D URWDWLQJ HOHFWULFDO PDFKLQH ZLWK FODZSROH URWRU DQG DOWHUQDWRU RU 5HODWLYH DOWHUQDWRUVWDUWHUFRPSULVLQJVDPH)UHQFK3DWHQW:2$       0DUFK HUURU >@ )/RFPHQW$%UX\HUH(6HPDLO;.HVWHO\Q$%RXVFD\URODQG-0 7$%/(,9 'XEXV³&RPSDULVRQRIDQGOHJ9ROWDJH6RXUFH,QYHUWHUVIRUORZ &203$5,6212)$9(5$*(&<&/,&,1'8&7$1&(6)25,) $ —+ YROWDJH DSSOLFDWLRQV´ ,((( ,QWHUQDWLRQDO (OHFWULF0DFKLQHV DQG'ULYHV &RQIHUHQFH,(0'&$QWDO\D7XUNH\SS0D\ /6G! /6T! /6G! /6T! /6G! /6T! ,6%1 )LUVW >@ 6 6FKXOWH & .DHKOHU & 6FKOHQVRN DQG * +HQQHEHUJHU ³&RPELQHG       PHWKRG DQO\WLFDO DQG QXPHULFDO FRPSXWDWLRQ DSSURDFK IRU GHVLJQ DQG RSWLPLVDWLRQRIVL[SKDVHFODZSROH DOWHUQDWRUV´6FLHQFH0HDVXUHPHQW 6HFRQG DQG7HFKQRORJ\,((3URFHHGLQJV±9RO,VVXH1RYSS       PHWKRG ± 5HODWLYH >@ ³*XLGH 7HVW 3URFHGXUHV IRU 6\QFKURQRXV 0DFKLQHV 3DUW , $FFHSWDQFH       DQG 3HUIRUPDQFH 7HVWLQJ 3DUW ,, 7HVW 3URFHGXUHV DQG 3DUDPHWHU HUURU   'HWHUPLQDWLRQIRU'\QDPLF$QDO\VLV´,(((6WG >@ . 0 5DKPDQ 6 +LWL ³,GHQWLILFDWLRQ RI PDFKLQH SDUDPHWHUV RID  V\QFKURQRXVPRWRU´,(((7UDQVDFWLRQVRQ,QGXVWU\$SSOLFDWLRQV9RO ,VVXHSS,6610DUFK$SULO /6G  >@ %  1DKLG0REDUDNHK % ) 0HLERG\7DEDU ) )0 6DUJRV ³2QOLQH LGHQWLILFDWLRQ RI 3060 HOHFWULFDO SDUDPHWHUV EDVHG RQ GHFRXSOLQJ /6T FRQWURO´ &RQI 5HF ,((( ,QGXVWULDO $SSOLFDWLRQV 6RFLHW\ $QQXDO  0HHWLQJ&KLFDJR,/86$YROSS± /6G >@ '$ .KDEXUL 0 6KDKQD]DUL ³3DUDPHWHUV LGHQWLILFDWLRQ RI SHUPDQHQW  PDJQHW V\QFKURQRXV PDFKLQH LQ YHFWRU FRQWURO¶ 3URF WK (XURSHDQ /6T &RQI3RZHU(OHFWURQLFVDQG$SSOLFDWLRQV (3( 7RXORXVH)UDQFH  ±6HSWHPEHU LQGXFWDQFH —+ /6G >@ /LX / &DUWHV '$ ©6\QFKURQLVDWLRQ EDVHG DGDSWLYH SDUDPHWHU  LGHQWLILFDWLRQ IRU SHUPDQHQW PDJQHW V\QFKURQRXV PRWRUV´ &RQWURO /6T 7KHRU\ $SSOLFDWLRQV,(7-XO\9ROSS  >@ $ %UX\HUH ( 6HPDLO ) /RFPHQW $ %RXVFD\URO -0 'XEXV -& 0LSR³,GHQWLILFDWLRQRIVHQVLWLYH5/SDUDPHWHUVRIDPXOWLSKDVHGULYH  H[FLWDWLRQILHOGFXUUHQWL) $ E\ D YHFWRU FRQWURO´ ,((( 3RZHU (OHFWURQLFV 6SHFLDOLVWV &RQIHUHQFH 3(6&  )LJ([SHULPHQWDOLQGXFWDQFHVRQGDQGTD[LVLQWKHVXEVSDFHVDVD IXQFWLRQRIWKHH[FLWDWLRQFXUUHQW $&.12:/('*0(17 7KLV ZRUN ZDV VXSSRUWHGE\ WKH)UHQFKFDU VXSSOLHU9DOHR DQG WKH UHJLRQDO FRXQFLO RI )UDQFH 5HJLRQ1RUG3DV'H &DODLV APPENDIX

A1: SEVEN-DIMENSIONAL CONCORDIA TRANSFORMATION MATRIX:

 11 1 1 1 1 1    2   4   6   8  10  12   cos22   cos2   cos2   cos2   cos2   cos2     7   7   7   7   7   7    2   4   6   8  10  12   sin20   sin2   sin2   sin2   sin2   sin2     7   7   7   7   7   7      4   8  12  16   20   24   cos22   cos2   cos2   cos2   cos2   cos2   C  7/1   7   7   7   7   7   7    4   8  12  16   20   24   sin20   sin2   sin2   sin2   sin2   sin2     7   7   7   7   7   7    6  12  18   24   30   36   cos22   cos2   cos2   cos2   cos2   cos2     7   7   7   7   7   7    6  12  18   24   30   36   sin20   sin2   sin2   sin2   sin2   sin2     7   7   7   7   7   7 

A2: EXPERIMENTAL SET-UP DESCRIPTION:

(a) (b) (c) (a): Seven-phase starter-alternator (rear of the picture) mechanically connected to a brushless synchronous machine (front of the picture) (b): Seven-leg Voltage Source Inverter (c): Programmable Electrical source and load