Power Electronics And Microelectronics Domains English Language Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Originally, DC drives were very attractive to engineers due to their simplicity of use and control. For the last 30 years, synchronous machines have been replacing the DC drives thanks to enhancements achieved in power electronics and microelectronics domains that introduced faster DSPs and FPGAs. AC drives were preferred to DC drives in most four-quadrant applications, since the former compare favorably with the latter in terms of cost and performance. Moreover, an AC machine can be brushless whereas a brush/commutator system is essential to a DC drive, which requires maintenance that is expensive

The speed and torque in DC drives can be easily controlled by varying the stator and armature current while phase angle between the current and the induced voltage is mechanically fixed by the commutator and brushes. In AC drives, not only the currents amplitudes must be controlled but also the phase angle can be adjusted by the electronic hardware and control loop. In other words, the amplitude and angle of the current must be controlled with an algorithm, i.e. the current vector. This constitutes the principle of the vector control.

When the current vector is properly controlled in AC drives, optimal performances and a good tolerance to load torque disturbances are obtained. Using vector control, AC drives (whether synchronous or asynchronous machines) become equivalent to DC drives. The control of flux becomes independent from torque control through a coordinated change in the supply voltage and frequency <cite>boldea</cite>.

Many types of vector control for AC drives were developed since the introduction of vector control theory in 1970's by Blaschke <cite>blachke</cite>. The most common one for a synchronous machine is the rotor oriented vector control with a voltage source inverter. This is the method used in this work.

In this chapter, the principle of the vector control used for MEGEVE application is considered and simulation results are presented. However, the FPGA implementation of the vector control was tested with ALLISON machine as stated in Chapter 3 and experimental results are shown.

Principle of vector control used in the brushless synchronous machine

In motor mode, the machine must convert electrical power provided by the electronic devices to mechanical power. The exciter stator is fed with AC voltage that induces, via a transformer effect, an electromagnetic field in the exciter armature, whether the exciter rotor is rotating or at standstill. This electromagnetic field induces AC voltages in the stator armature of the exciter, which in turn are rectified by the rotating rectifier and supplied to the rotor of the main starter/generator. Variable frequency AC power is supplied from the voltage source inverter to the main starter/generator stator. This AC power produces a rotating magnetic field in the main stator leading the main rotor to rotate and supply mechanical power.

Vector control scheme <label>vector</label>

The aim of the control in MEGEVE application is to create an electromagnetic torque significantly higher than the load torque in order to start the main engine of the aircraft from 0 up to 8000 rpm. So a torque vector control was chosen for this application.

The developed control scheme of the brushless synchronous machine MEGEVE using PI controllers for the regulation of the stator currents in the rotating frame (d, q) is presented in Figure 4.1.

The d-q current references I_{d}^{*} and I_{q}^{*} are deduced from the electromagnetic reference torque G_{e}^{*}. Three phases currents I_{a},I_{b},I_{c} are then measured and a Park transformation is applied using the electrical rotor position ? in order to obtain the d-q current components. Each current component is then regulated with a PI controller that delivers the d-q reference voltage V_{d1} and V_{q1}. Then a decoupling feed forward compensation of the induced back-emf E_{d} and E_{q} is applied to ensure a total independent control of the torque and the flux. The compensation terms are expressed in equation (<ref>fem</ref>)

<K1.1/> <label>fem</label>

<K1.1 ilk="MATRIX" >




After compensation, the resultant voltages V_{d}^{*} and V_{q}^{*} are transformed with an inverse Park transformation that generates the three reference phase voltages V_{a}^{*}, V_{b,}^{*} V_{c}^{*}. These voltages are then compared with a Pulse Width Modulation PWM carrier having a frequency of 20 kHz and the comparison result is nothing but the command signal of the used IGBTs.

As for the excitation voltage of the exciter, it is delivered by an external AC source with a fixed frequency of 633 Hz . This frequency is equivalent to the maximum reached mechanical speed seen from the exciter (i.e. 8000 rpm).

[TeX field]<LaTeX>\landscape</LaTeX>

[vector scheme]

[Figure 4.1: Vector control scheme]



Electromagnetic reference torque

The electromagnetic reference torque for the MEGEVE application is represented in Figure 4.2. The control algorithm must develop a maximum transient torque up to the base speed (5000 rpm in this case) <cite>Krishnan</cite>. At that point, the motor reaches its limit speed, the so-called rated speed (also called base speed when talking about flux-weakening). The maximum voltage of the inverter is also reached for the maximum transient torque.

Beyond that limit, the motor torque decreases rapidly toward its minimum value, which depends on the load torque profile. It is the mechanical power that is now kept constant up to the maximum speed <cite>boldea</cite>. A common method in the control of synchronous motors that allows to keep the maximum mechanical power above the base speed is to reduce the magnetizing flux. This method is known as field-weakening. The air gap flux is weakened by producing a negative d-axis current component I_{d} while the maximum transient power is ensured at the rated voltage of the inverter. In case of MEGEVE, the maximum torque is 260 N.m and the maximum electrical power is 130 kW .


[Figure 4.2: Electromagnetic reference torque]

Load torque

In this application, the mechanical load is the main engine of the aircraft. An example of main engine profile is represented in Figure 4.3. It is important to note that the starting process uses the brushless exciter starter/generator to spin the main turbine shaft (see Figure 4.3, Part I). The motor is bolted to the outside of the engine and uses a shaft and gears to connect to the main shaft. The electric motor spins the main shaft until there is enough air blowing through the compressor and the combustion chamber to light the engine (see Figure 4.3, Part II). Fuel starts flowing and an igniter similar to a spark plug ignites the fuel. Then fuel flow is increased to spin the engine up to its operating speed (see Figure 4.3, Part III). At the end of this third step, the motor mode of the brushless exciter motor ends up and the engine starts to lead the main shaft. Therefore the brushless exciter machine operates in the generator mode above 8000 rpm. Since this study is limited to the motor mode only, the control of the machine stops at 8000 rpm.


[Figure 4.3: Load torque of the main engine of the aircraft.]

Current references

The current references are calculated from the reference load torque as mentioned in section <ref>vector</ref>.

� Control in constant torque region

In this region, the control strategy is chosen so that the machine produces the maximum torque. Although the main starter/generator is a salient synchronous machine, it was preferred not to develop the reluctance torque (second term of equation (1.14)), but to provide the needed maximum torque by controlling I_{f} and I_{q}. Consequently, the d-current reference value is set to zero in this region.


As explained in Chapter 1, the excitation current of the main machine I_{f} is provided by the rectified rotor voltages of the exciter. Figure 4.4 shows the experimental results of both excitation currents. This is obtained by exciting the exciter with AC voltage as required in the motor mode while it supplies resistive loads values (corresponding to the rotor of the main SG at ambient and high temperatures). It can be easily deduced that a quasi-constant gain K_{I} exists between the excitation currents of the exciter and the main starter generator. Therefore, in equation (<ref>current</ref>), I_{ex} represents the peak value of the AC excitation current and I_{f} represents the DC value of the MSG excitation current.

I_{f}=K_{I}.I_{ex} <label>current</label>


[Figure 4.4: MSG excitation current vs. exciter excitation current]

As a result, the reference excitation current of the exciter is chosen in order to create the maximum flux linkage in the rotor of the main machine, as per equation (<ref>flux</ref>):

I_{ex}^{*}=((I_{f}^{*})/(K_{I}))=((?_{fmax})/(K_{I}.L_{fsat})) <label>flux</label>

As far as the reference q-current is concerned, it is deduced from the electromagnetic reference torque (see Chapter 1, section knowing that I_{d}^{*} is zero by :


The speed runs from 0 up to the base speed where the maximum voltage and maximum torque are reached.

� Control in constant power region

In this region, the maximum voltage is reached and the speed is still running up. A flux-weakening strategy must be adopted. Therefore a negative d-current is injected in order to decrease the flux linkage whereas the excitation current is kept constant to produce the maximum torque. Many flux weakening strategies exist in the literature nowadays thanks to enhancements in microelectronics domain <cite>Jang Hee</cite>- <cite>Jahns</cite>- <cite>Krishnan1</cite>.

In this work, the d-current reference is calculated in order to prevent the voltage amplitude exceeding the maximum voltage that can be delivered by the voltage source inverter VSI. When applying the space vector modulator SVM, the maximum peak voltage of the VSI is defined as <cite>xiao</cite>:


As a result, the voltage amplitude applied to the stator of the main machine must respect equation <ref>stator voltage</ref> as stated in <cite>morimoto</cite>

V_{s}=v(V_{d}�+V_{q}�)?((V_{dc})/(v3)) <label>stator voltage</label>

By referring to the electrical equations of the machine (see Chapter 1, section neglecting the stator resistor plus the transient state, the d-current reference is calculated using:


The q-current reference is deduced from the electromagnetic torque and the d-current reference values by:




Therefore, in steady state, the current vector is controlled to keep the resulting voltage V_{s} (including the resistance drop voltage) within the maximum voltage V_{max}.

PWM strategy

Many Pulse Width Modulation strategies have been developed throughout the years <cite>Pou</cite>- <cite>kezimierkowski</cite> such as:

� CB-PWM (carrier based pulse width modulation): This is the classical and most widely used method of pulse width modulation <cite>Newton</cite>. It consists in comparing a reference signal (modulator) to a triangular carrier with F_{swi} as a switching frequency. The principle of the CB-PWM is to deliver at each switching frequency an average voltage on the machine's terminals that is equal to the reference value. During a switching period (T_{swi}), an active inverter leg applies two consecutive switching states of opposite voltage polarity (result of the comparison). When using the CB-PWM characterized by subcycles of constant duration, two salient sidebands, centered about the carrier frequency (and additional frequency bands around integral multiples of the carrier).are reflected in the harmonic spectrum <cite>J. Holtz pwm</cite>. Figure 4.5 shows the PWM principle.

� Precalculated PMW: This method calculates the commutation angles in order to eliminate specific harmonics of the spectrum.

� SVM (Space Vector modulation): The space vector modulation technique averages three consecutive switching state vectors (the vector null and the two active vectors limiting the sector that contains the voltage reference vector) over the time interval that is the sampling period. The resultant average voltage vector must equal the sampled reference vector <cite>Zhou</cite>.

An equivalent strategy to the SVM consists in adding the zero-sequence signal to the modulator and to compare it to a triangular carrier as the CB-PWM. The obtained performances with this method are exactly the same of the SVM <cite>wissem</cite>.


[Figure 4.5: (a) CB-PWM functional schematic of an inverter leg A, (b) Principle of CB-PWM]

For the simulation tests performed in section <ref>simulation results</ref>, the PWM modulator is a carrier based PWM of 20 kHz with zero-sequence signal addition, that is equivalent to the SVM.

As far as the calculation of the PI current regulators is concerned, it was described in Appendix A.

Simulation results <label>simulation results</label>

The complete model of the BSSG (detailed in Chapter 1, section 1.3) together with the vector control detailed in this chapter were developed with Matlab/Simulink environment. A start-up of the machine up to 8000 rpm was simulated. Simulation results are presented in Figure 4.6.


[Figure 4.6: Simulation results: (a) I_{d}^{*} and I_{d}, (b) I_{q}^{*} and I_{q}, (c) G_{e}^{*} and G_{e},(d) Mechanical speed, (e)- Stator currents, (f) Stator currents zoom, (g) V_{ab}, (h) Electrical position]

Figures 4.6- (a) and (b) represent the reference and actual d-q currents respectively. The d-current is kept zero and the q-current is constant until the flux weakening zone is reached where both currents start decreasing. The reference and actual torque are shown in Figure (c). The mechanical speed which is the result of the torque control is presented in Figure (d). It can be seen that the machine's speed runs from zero up to 8000 rpm. The three phase currents are shown in (e) with a zoom in (f). The phase to phase voltage V_{ab} is illustrated in Figure (g), while the electrical position is given by Figure (h).

Implementation on FPGA board

As mentioned in Chapter 3, the vector control of the brushless synchronous starter generator was implemented on an Actel FPGA board.

Vector control implemented architecture

Figure 4.7 presents the architecture of the implemented vector control, developed by following the methodology explained in Chapter 3 section 3.2.3.


[Figure 4.7: Current vector control algorithm scheme]

The control unit ensures the control and the sequence order of the different integrated modules. When activated and after conversion time t_{AD}, the control unit in turn, activates the Park Transformation module (see Figure 4.7, block "123-dq"). This transformation calculates the d-q components of the stator current during t_{P} time. When this module indicates the end of the calculation, the comparison module is activated in parallel on both axis, thus having the same calculation time of t_{C.}

At this stage, it is important to recall that Actel FPGA doesn't include integrated multipliers. For resources concern, most of the multipliers were factorized in order to reduce their number, despite of parallelism as stated in Chapter 3.

The PI controller of d-axis is activated when the comparison is finished. When this step is achieved (after t_{PId} time), the PI controller of q-axis is activated. The end of the calculation of the latter (after t_{PIq} time), allows the activation of the Park inverse transformation that generates the three phase reference voltage. The calculation time of the Park inverse transformation is t_{IP}. Once these voltages are calculated, they are compared to the PWM carrier in order to generate the adequate command signal HP_{x} and BP_{x} (x refers to a phase number).

Then these command signals are treated, as explained in section <ref>modified pwm</ref>, by taking into account all the undesired cases that may produce any damage to the inverter's drivers and/or IGBTs.

In this implementation scheme, the back emf compensation was not considered. During the experimental results, the machine mainly operated in the low speed region where the back-emf can be neglected. Therefore, the compensation of the back-emf was not a primary concern.

The timing diagram of the implemented vector control is shown in Figure 4.8.


[Figure 4.8: Timing diagram of the implemented vector control]

Table <ref>timing table</ref> details the calculation time required by each module when using a FPGA of Actel A3P1000 type with a clock of 40MHz.

[B]<LaTeX>\begin{table}[H] \centering</LaTeX>

Module Latency Computation time

AD Interface 16 t_{AD} = 0.4 �s

abc-to-dq (Park) 20 t_{P} = 0.5 �s

Comparison 8 t_{C} = 0.2 �s

PId 12 t_{PId} = 0.3 �s

PIq 12 t_{PIq}= 0.3 �s

dq-123 (Inverse Park) 13 t_{IP} = 0.325 �s

t_{CR} = t_{P} + t_{C} + t_{PId} + t_{PIq}+ t_{IP} t_{CR} = 1.625 �s

Execution time T_{ex} = t_{AD} + t_{CR} T_{ex} = 2.025 �s

Number of core cells 24332 out of 24576 (99%)

Consumed resources Block rams 6 out of 32 (18%)

Multipliers (16x16) 3

[caption]<LaTeX>\caption{FPGA time/area performances of the vector current controller}</LaTeX><label>timing table</label>[E]<LaTeX>\end{table}</LaTeX>

The calculation time of the vector control is then 1.625�s. By adding the conversion time t_{AD}, the total execution time of the algorithm is of 2.025�s.

Another performance criterion to consider when using FPGAs is the consumed resources. The consumed FPGA resources are obtained for 16-bits fixed point format.

Modified PWM principle <label>modified pwm</label>

For the implementation, the CB-PWM was selected for it simplicity and efficiency. The frequency of the carrier was set to 20 kHz based on the used Semikron IGBT referenced SEMIX553GD128DC. However, and in order to get better performances, this PWM was slightly modified.

Figure 4.9 illustrates the principle of the used PWM.


[Figure 4.9: Secured PWM signals management scheme]

In the positive half-period, the current flow of the phase A is ensured via T_{a+} (see Figure 4.5) and the commutation is therefore inhibited in the lower leg T_{a-}. Therefore, T_{a-} is not conducting during the whole positive half-period. After completion of the positive half-period and before the lower leg starts conducting, a dead time of 5�s is inserted in order to avoid any short circuit incidence. Then, when the current becomes negative, only T_{a-} will commute.

Moreover, each command signal deduced from the comparison between the reference voltage value and the carrier (HP_{x} or BP_{x}) is split into two command signals (one signal in conduction mode (HP_{x}_ON or BP_{x}_ON) and another one in non-conducting mode (HP_{x}_OFF or BP_{x}_ON)). Each IGBT receives the corresponding two signals.

The "inverter's leg module" is then responsible for treating the command signals before application on the inverter's IGBTs. It receives the different HP_{x} and BP_{x} from the PWM block (see Figure 4.9) and treats them in order to create the corresponding signals (HP_{x}_ON and HP_{x}_OFF) and (BP_{x}_ON and BP_{x}_OFF) and at the same time to avoid any disfunctionning of the inverter.

Different fault cases were figured out and treated with the following functions that are also presented in Figure 4.10:

� Glitch numerical filtering block

� Overfrequency fault block

� PWM fault block

� Command driver ON/OFF

� Dead time insertion block

The general_fault signal is active at low level and allows to inhibit all the active command signals of the inverter. Each block of Figure 4.10 is detailed separately in the upcoming subsections.


[Figure 4.10: Inverter's leg block scheme]

Glitches numerical filtering block

The different command signals HP_{x} and BP_{x} might contain undesirable glitches that can produce a dysfunction of the inverter. Therefore, these command signals (HP_{x} and BP_{x}) are treated in the "glitch numerical filtering block" (see Figure 4.10) to eliminate any glitch of a duration less than 200 ns . The resultant command signals are HP_{x}_filtered and BP_{x}_filtered. Figure 4.11 shows an example of HP_{x} signal that is treated by the glitch numerical filtering to eliminate any glitch of a duration less than 200ns.


[Figure 4.11: Example signals treated by the glitch numercial filtering bloc ]

Overfrequency fault block

To reduce the losses of an IGBT device and subsequently of the total system cost and also to reduce volume, an essential parameter has to be considered which is the switching frequency. The maximum admissible switching frequency depends of the type of the used IGBT. In this application, Semikron IGBT referenced SEMIX553GD128DC are used with a recommended switching frequency of 20 kHz .

If for any reason, two consecutive pulses are active in less than 48.8�s , the signal overfrequency fault generated by the "overfrequency fault block" (see Figure 4.10) will be active during 50�s. That will inhibit all the command signals of the 3 inverter's legs. Figure 4.12 shows that overfrequency fault signal is activated when the filtered command signal has an activated pulse in less than 48.8�s .


[Figure 4.12: Example of activation of overfrequency fault]

PWM fault block

When the inverters function, if both upper and lower IGBTs of a same leg are active, then two phases will be short-circuited causing the damage of the IGBTs of the inverter. Therefore, a PWM fault signal is generated by the "PWM fault block" (see Figure 4.10) to avoid this type of events. It forces HP_{x} and BP_{x} to be inactive and this during 50�s as shown on Figure 4.13.


[Figure 4.13: PWM fault block: Exemple of functioning ]

Dead time insertion

Let's study the case when the HP_{x} signal is active. In this case HP_{x}_ON will is equal to 1.


[Figure 4.14: Dead time insertion diagram]

When this IGBT stops conducting, then HP_{x}_ON should normally switch to zero, while an active pulse of HP_{x}_OFF will occur, in order to block the upper leg IGBT. Before the lower IGBT of the same leg starts conducting and in order to eliminate any possibility of short-circuit, a dead time of 5�s is inserted between the falling edge of the pulse HP_{x}_OFF and the rising edge of BP_{x}_ON (see Figure 4.14).

Command driver on/OFF block

To control each IGBT, two command signals are required, the HP_{x}_ON (to activate) and HP_{x}_OFF (to inhibit), both for upper leg and BP_{x}_ON and BP_{x}_OFF for lower leg. This block "command driver on/OFF" (see Figure 4.10) is responsible for generating these signals properly.

Figure 4.15 shows a command signal HP_{xsec ureddt} (obtained after being treated by the different previous blocks) into HP_{x}_OFF and HP_{x}_ON that represent the effective complementary signals that are used to command the same IGBT.


[Figure 4.15: Decomposition of HP_{_{X}}sec ureddt into HP_{_{X}}_OFF and HPx_ON]

HP_{x}_OFF and HPx_ON must respect the basic cycle presented in Figure 4.16.


[Figure 4.16: Basic cycle of HP_{_{X}}_OFF and HP_{x}_ON]

Launching of PWM calculation

When using FPGAs to realize the vector control with CB-PWM, it is essential to respect that:

� The reference voltage compared to the carrier are refreshed at precise instants. Moreover, the sampled reference value is kept constant until the following refreshment time.

� The refreshment time of the reference value, is optimal when it occurs with the carrier vertices (low and high).

The control unit of the algorithm is launched via a start signal generated at each sampling period T_{s} (25�s ). In this case T_{s} corresponds to the half of the switching period T_{swi} (50�s ). It can be noted that, with regard to the carrier vertices, the active level pulses of the start signal are applied a time T_{ex} (see Table <ref>timing table</ref>) before.

This way, the voltages reference values are

refreshed exactly at the carrier (low and high) vertices, which corresponds to a control loop delay that is only equal to T_{ex} as shown in Figure 4.17. These reference values are maintained constant until the following refreshment time <cite>wissem these</cite>. An end signal indicates that the reference voltages are computed and refreshed.


[Figure 4.17: Synchronized PWM sequential timing diagram ]

Experimental results

To evaluate the performance of the vector control, experimental tests were elaborated on the brushless exciter synchronous machine called ALLISON. These results are obtained using PI-based current controllers on both d and q axis, with the PWM detailed of 20 kHz and a sampling frequency of 40 kHz .

Figure 4.18 shows the start and end signals together with the clk_synchro which is the clock that is synchronized with the carrier and has 20 kHz as frequency. The edges of clk_synchro correspond to the carrier vertices. These signals were visualized via the digital to analog converters. This figure proves that the computation was activated in advance of T_{ex} from the clk_synchro edges. It also shows the end signal indicating that the voltage references values are refreshed at exactly the peaks of the carrier (or the rising and falling edges of the clk_synchro).

[Figure 1]

[Figure 4.18: Experimental results: PWM clock at 20 kHz with the start and end signals]

The d-q currents reference values are transmitted to the FPGAs by the control panels via potentiometers. The position of the machine is given with a resolver that was mounted on the shaft of the machine. A resolver to digital converter RDC is used to provide the digital position. This latter is transmitted to the FPGA board using a differential bus and the electrical position is used to achieve the vector control. The currents are measured with LEM transformers having a bandwidth of 100 kHz .

Experimental results at 500 rpm

Figure 4.19 shows the three phase currents obtained with a reference value I_{q}^{*}=36A with I_{d}^{*}=0 at 500 rpm.


[Figure 4.19: Experimental results: (a) d-q currents (reference and experimental), (b) Three phase stator currents at 500 rpm.]

Figure 4.20 shows the A phase current and the corresponding command signals BP_{a} and HP_{a} of the T_{a+} and T_{a-}. It is clear that HP_{a} (BP_{a}) is null when the current phase is negative (positive). When functionning in the half positive (negative) period of the current, HP_{a} (BP_{a}) is nothing but the result of the comparison between the reference value and the carrier (after taking all the protections into account as explained in section <ref>modified pwm</ref>) .


[Figure 4.20: Experimental results: A-phase current I_{a} with the corresponding command signals HP_{a} and BP_{a}]

The measured phase voltage V_{ab} with a DC link of 270 Vdc is shown in Figure 4.21.


[Figure 4.21: Experimental results: Phase voltage V_{ab}]

The machine is still rotating at a fixed speed of 500 rpm (driven by the bench see Chapter 3 section . When I_{q}^{*} is increased, the measured I_{ab} are increased proportionally. When I_{d}^{*} decreases (since I_{d}^{*} is negative) while I_{q}^{*} is maintained constant, the resultant amplitude of the I_{ab} currents will slightly increase. This can be seen in Figure 4.22.


[Figure 4.22 -a: Experimental results at 500 rpm I_{d}^{*}, I_{q}^{*} and the measured I_{ab}]


[Figure 4.22-b: Experimental results at 500 rpm I_{d}^{*}, I_{q}^{*} and the measured I_{ab}]


[Figure 4.22-c: Experimental results at 500 rpm I_{d}^{*}, I_{q}^{*} and the measured I_{ab}]


[Figure 4.22-d: Experimental results at 500 rpm I_{d}^{*}, I_{q}^{*} and the measured I_{ab}]

Experimental results of start-up and deceleration of the machine

In Figure 4.23, the machine was at standstill initially. The d-q currents references are then applied, together with the excitation current in order to develop an electromagnetic torque. When this latter is higher than the load torque of the bench, the machine starts spinning (see Figure 4.23 -(b)) and reaches 600 rpm (see Figure 4.23 - (c)). Then the I_{q}^{*}is decreased to decelerate the machine as seen in Figure 4.23 -(d).


[Figure 4.23-(a): Experimental results: Electrical position and the actual currents I_{ab} and I_{q}^{*} from start-up till the deceleration of the machine]


[Figure 4.23-(b): Experimental results: Electrical position ? and the actual currents I_{ab} and I_{q}^{*} at start-up of the machine]


[Figure 4.23-(c): Experimental results: Electrical position ? and the actual currents I_{ab} and I_{q}^{*} at constant speed of 600 rpm]


[Figure 4.23-(d): Experimental results: Electrical position ? and the actual currents I_{ab} and I_{q}^{*} at deceleration of the machine]

Currents trajectory in (a,�) frame

Figure 4.24 shows the trajectory of the (a,�) currents at 600 rpm. Figure (a) illustrates the trajectory at 25 A , (c) shows the trajectory at 50 A , and (b) represents the transient form 25 A up to 50 A . As for Figure (d), it illustrates the start-up (from 0 to 25 A) and (e) shows the transient from 25 to 0 A.

[Figure 4.24 : Experimental results: (a)- currents trajectory at 25 A, (b) transient state from 25 A to 50 A, (c) currents trajectory at 50 A, (d) start-up from 0A to 25A, (e) deceleration from 25A to 0A]


In this chapter, the vector control of the brushless synchronous starter generator is explained. The control is based on PI current controllers with CB-PWM at a frequency of 20 kHz . The reference currents are calculated in order to develop the reference magnetic torque as specified in MEGEVE project. The simulation results of the vector control describing the start-up of the MEGEVE machine up to 8000 rpm are also presented.

On the other hand, the experimental tests were elaborated on ALLISON machine that is also a brushless synchronous starter generator. The currents references were transmitted via potentiometers. The experimental results concerning the vector control are also presented, including a start-up of the machine up to 600 rpm.