Paper submitted wit thin the scope of the Master’s Thesis Maste of Industrial Sciences er GROUP T – Leuve Engineering Co G en ollege – 2009-2010 0
Design of a high power co n h ontroller f BLDCfor -motors on FPGA a dsPIC n and
Supinya Piampongsan *, Athanasio Sarafianos*, Dr. Ir. Lam nt os mbrechts Marc†, Prof. Ir. Bi c ienstman Luc‡
*
Master st tudent of Industr Engineering: Electronics, GR rial : ROUP T – Leuven Engineering College, Vesaliu usstraat 13, 3000 0 Leuve Belgium en, † Actu uator Division, M Melexis N.V., Trasportstraat 1, B B3980 Tessender Belgium rlo, ‡ Unit Electronics, GROEP T – L Leuven Engineer ring College, Ve esaliusstraat 13, 3 3000 Leuven, Belgium <Luc.Bienstman@groept.be e>
Abstract DC motors have ga ained popularity y Brushless D (BLDC) m in automot tive applicatio in recent years; system ons ms that were formerly im mplemented m mechanically o or lly are being converted into electrical g i l, hydraulical BLDC mo otor-driven sys stems. The re easons for thi is shift inclu ude the motor higher effi r's iciency, highe er torque per v volume and ex xtreme durabili ity.
The cost of such chan nges is the ne for a mor eed re complex control algorith than that o conventiona hm of al brushed DC motors an an increas in required D nd se hardware. To address t this issue, Melexis N.V., a ased manufact turer of integra ated circuits fo or Belgium-ba automotive applications, developed a p e pre-driver ASIC C that works in conjunctio with a moto controller to on or phase BLDC m motors. run three-p This pap outlines th implementa per he ation of such a motor cont troller. Various algorithms u used to run thi is type of motor includ ding six-step commutation n, ed re sinusoidal control, and field oriente control ar ed red GA implemente and compar on an FPG as well as a dsPIC. The pre-driver A e ASIC provided by Melexis i d is g integrated on a custom-m o made PCB capa of handling able high power and a graphic user interface is developed r cal in LabVIE EW to control and observe the program e m running on the logical unit. n Keywords: BLDC motor dsPIC, FOC, FPGA, r, , sinusoidal c control Introducti ion Brushless direct current (BLDC) moto is a type o or of electromoto that has become prevalent in th or he automotive industry over the past decad It has taken e r des. n over tasks previously performed by its brushed s counterpart and is now competing ag t gainst inductiv ve motors for dominance in car technology y. BLDC motors are used in electric powe s er Today, B assisted ste eering systems gearbox hy s, ydraulic pumps s, air compr ressors, as well as nu umerous othe er componen nts. They are preferred ov brushed D ver DC motors d due to their relative rob bustness again nst mechanical failings an are favored over ind
ucti nd d ive because of t their relativel high pow ly wer motors b efficiency and cooler op y perating temper ratures. The tec chnical disadvantage of BLD lies in the DCs eir dependence on comp putationally h heavy electron nic Unlike brushed DC motors wh hose turnings a are control. U facilitated by mechan d nical brushes BLDCs a s, are electronic cally commuta ated; they do not turn witho n out the assista ance of electro onic hardware called a mot e, tor controller which sends electrical sign to the mot r, nals tor in a part ticular sequen nce. In doing so, the mot g tor controller requires know r wledge of the m motor’s positio on, whether from position sensors or a c f complex position estimation algorithm. Various ty n ypes of mot tor controller and control algorithms exi each with its rs ist, own stre engths and w weaknesses, which will be w elaborated up in this doc d cument. The assig gnment To ease the motor co ontrol unit des sign process f for ve , sed automotiv applications, Melexis NV, a Belgium-bas microelectronic compan has designed an applicatio ny, onintegrated circ cuit (ASIC) that acts as an t specific i interface b between a low-voltage digital motor controll l ler and the high-voltage analog motor peripherals. It r l uld performs certain signal processing tasks that wou ecuted by discre components eet s. otherwise have to be exe约束椅
Figure 1 A sample motor con unit with Mel ntrol lexis’s 15120 ASIC C
2
Figure 1 illustrates how Melexis’s15120 ASIC chip connects to a motor controller to a high power inverter module, which is then connected to a three-phase BLDC. Among other tasks, the ASIC converts high voltage position sensor signals from the motor to logic-level 5V for motor controllers. It also provides current measurement via a shunt resistor, over-current safety mechanism, and protection against shootthrough transient voltages. More details on the ASIC and its functional blocks can be found in the appendix. The aim of this thesis is to create a demonstration board for the ASIC, using it to control a BLDC motor in a variable speed drive. In doing so, the students are presented with a number of possible motor controllers and control algorithms. Theoretical background: DC motors and technology A brushed DC motor (hereby referred to simply as a DC motor) is composed of two general parts: the stationary part, or stator, and the rotating part, or rotor. The behavior and interactions between these parts are detected and controlled by a commutator and a mechanical brush. An example of such a motor is given in Figure 2.
a part of this force acts as useful torque on the rotor. The optimum angle between the stator and rotor fields is 90 degrees, which results in all of the force generated being used as torque on the rotor, giving maximal torque and efficient use of current. The motor is started by energizing the stator and letting current flow through the rotor windings, thus activating the stator and rotor fields respectiv
ely. As seen in Figure 3, Position1, the two fields are initially orthogonal, creating maximal torque which acts on the rotor, causing it to turn. As the rotor turns, the rotor field also changes direction, aligning itself with the stator field generated by the permanent magnet. When the rotor reaches Position 2, the angle between the two fields has converged to zero degree; no torque acts on the rotor and the motor is motionless. This is where the brush and commutator enter the scene.
Figure 2 An example of a DC motor with a stator magnet and rotor armature windings; a commutator and brush can be seen physically contacting to the rotor
The static part of a DC motor consists of components capable of generating an electromagnetic field. This may be a permanent magnet or, as is the case in large systems, electromagnets or coils. The stator can generate a magnetic field of a certain magnitude at a fixed direction. The dynamic part of the motor, or rotor, consists of windings, spun around a shaft. Typically it is surrounded by the stator and its magnetic field. When electrical current flows through these rotor windings, an electromagnetic field is induced. The magnitude and direction of such a field is dependent upon the magnitude and direction of the rotor winding’s current. When positioned within each other’s vicinity, the rotor and stator fields interact with each other in such a way that results in a force as long as there is an angular difference between the two. The direction of this force depends on the angle betw
een the fields, and
Figure 3 Depiction of the commutation of a one-coil DC motor, using brushes and a commutator; the brush is connected to the supply voltage and ground.
When the rotor approaches Position 2, the brush, which makes mechanical contact with the rotor’s coils, detects this approach and causes the commutator to change the polarity of the rotor windings’ current. This changes the direction of the rotor field, causing the angle between stator and rotor fields to be, once again, ideally 90 degrees. Thus, a sufficiently large force is generated for the rotor to continue to turn. The change in current polarity according to rotor’s position is called a commutation; when done with a commutator and brush, it is called mechanical commutation.
3
BLDC motor A brushless DC motor, on the other hand, can be viewed as a DC motor turned inside-out. The stator is made up of the coils that generate an electromagnetic field of controllable magnitude and direction, while the rotor is a permanent magnet that generates a magnetic field of constant magnitude (see Figure 4).
can be used in volatile environments and EMI-sensitive environments. More comparisons between brushed and brushless DC motors are presented in Table 1.
Table 1 Comparison of BDC and BLDC motors
Characteristic Mechanical structure Brushless motors Field magnets on the rotor Similar to AC synchronous motor Distinctive Quick response and Long‐lasting features excellent controllability Easy maintenance (usually no maintenance required) Winding Ring connection The highest grade: Δ or Y‐ connections connected three‐phase The simplest: Δ‐ connection connection Normal: Y‐connected three‐phase winding with grounded neutral point, or four‐phase connection The simplest: Two‐phase connection Commutation Mechanical contact Electronic switching using method between brushes and transistors commutator Automatically detected Hall element, optical Detecting by brushes encoder, etc. method of rotor's position Reversing By a reverse of terminal Rearranging logic method voltage sequencer Conventional motors Field magnets on the stator
Figure 4 A brushless DC motor
未载入sso模块This approach has certain implications: 1. The coils no longer rotate. Thus, mechanical brushes and commutator become obsolete. The commutation of the field must be done electronically, not mechanically. 2. There is no brush to automatically cause commutation as the rotor reaches a certain position. Thus, the electronic commutator must obtain information on the rotor’s position, and consequently the direction of the rotor field, by some other means. 3. For optimal torque efficiency, the stator field should be kept ideally at 90 degrees in front of the rotor field, thus rotor position must be known with some precision. Position sensors that detect rotor position at higher the resolutions are better suited for such purpose. They enable higher resolution control of field angle and allow the motor to rotate more smoothly. Hall Effect sensor, which can sense magnetic fields, is the most commonly used sensor for rough position sensing. When a higher resolution is desired, encoders or resolvers are preferred. Each of these sensors requires a number of wires. When this must be reduced to an absolute minimum, there exist sensorless approaches, which will be discussed more in detail in a later section. 4. Since the commutator switches polarity via a mechanical contact, sparks are generated. Additionally, brushes and a commutator are mechanical parts that wear over time. These components are absent in BLDC motors, so they require relatively little maintenance, generate little to no electromagnetic interference (EMI), operate very quietly, and
BLDC structure and topology The most common BLDC motors are constructed as a wye-topology, th
ree-coiled system: the stator part consists of three coils connected in a star-like fashion (see Figure 5). Delta-connected motors also exist but do not differ from wye-connected ones in how they are controlled.
Figure 5 Equivalent circuit of a three-phase stator connected to half bridges; terminals A, B and C are the physical connections between the bridge and motor windings
Each stator coil is commonly referred to as a phase. Motors with more or fewer phases than three exist, but are out of the scope of this thesis. From Figure 5, one can identify the three phases: A, B and C. An inverter module, or half bridge, is used to supply current to each stator phase. In Figure 5, each phase is shown connected to such a bridge: two transistors, one connected to the supply voltage while the other connects to ground. They are commonly referred to as ‘high side driver’ and ‘low side driver’ respectively. By turning on the high side driver of one phase and the low side driver of another, one causes current to flow in a certain direction (Figure 6). Thus, control of current entering and exiting each phase can be achieved by turning high and low side drivers on or off in a particular sequence. In practice, this is usually done by applying pulse width modulated (PWM) signals.
called six-step commutation because there are six discrete ‘states’ to drive the inverter bridge. In each state, one stator phase is connected to ground, one is connected to supply voltage, and the last is left floating, with both of its half-bridge transistors turned off. Figure 7 shows all the possible states.
Figure 7 The six states of a six-step commutation. Current always flows into one winding and out of another. The third winding is left unconnected. Figure 6 An example of how the current flows into one winding and out of the other
半带滤波器The number of poles in the rotor magnet also contributes to the motor’s operation. A single pair of north and south poles is referred to as a ‘pole pair’. The significance of pole pair number is elaborated upon in a later section. Electronic commutation Various control algorithms exist to drive BLDC motors via electronic commutation. Depending on the application and desired results, these are the most common options: Six-step commutation , also referred to as block commutation Sinusoidal commutation Field oriented control or vector control These approaches are described in more detail below. Six-step commutation or block commutation Six-step or block commutation, is the most widespread control algorithm because of its ease of implementation and acceptable results. This method is
The sequence of states executed is determined by Hall Effect sensor signals. Typically, three Hall Effect sensors are mounted onto the stator at 120 electrical degrees apart around the shaft. Each sensor sends an ‘on’ signal when electromagnetic field is near and an ‘off’ signal otherwise. With three sensors, eight combinations of sensor states are possible. Of these, states '000' and '111' should never occur and are regarded as error states.
Table 2 Hall sensor conditions represented digitally with corresponding output states; in one electrical commutation, each sensor is on half of the time.
1111sHall 1 2 3 0 0 0 1 1 1 0 1 1 1 0 0 0 1 0 1 0 1 1 1 0 0 0 1 Phase A High Low side side OFF ON OFF ON OFF OFF ON OFF ON OFF OFF OFF Phase B High Low side side OFF OFF ON OFF ON OFF OFF OFF OFF ON OFF ON Never occurs Never occurs Phase C High Low side side ON OFF OFF OFF OFF ON OFF ON OFF OFF ON OFF
To turn the motor, a state of the six-step commutation is activated according to the current Hall sensor state. Thus, each of the six valid states of the
5
Hall sensor signals corresponds to one particular stat r te of the six-s commutati (see Table 2). step ion With six discrete comm x mutation steps, the rotor turn , ns for one ele ectrical revolution. The numb of electrica ber al revolutions required for one mechan s r nical revolution n depends on the number o rotor pole pairs: each pai n of p ir requires one electrical revolution p o per mechanica al revolution. Thus, a three e-pole pair rot magnet wil tor ll ee evolutions (3*6 steps) per on 6 ne require thre electrical re mechanical revolution. A complete di l iagram with al ll input, outpu voltages can be viewed in Appendix A. ut n Recall th for maxima torque effici hat al iency, the stato or field shoul lead the rotor field by 90 degrees. With ld 0 h six states per electrical revolution, the three Hal l ll rovide knowle edge of rotor position at a sensors pr resolution of 60 electric degrees. Thus, for a six cal xn H can ct step driven motor, the Hall sensors c only detec with certain that the sta field leads the rotor by an nty ator n angle betw ween 60 and 120 degrees A possible d s. e, undetectable error of m maximum 30 degrees from m gle o e is desired ang will lead to small torque decay. This i expressed b the followin equation wh T is torqu by ng here ue as a functio of stator fiel leading an on ld’s ngle:
Aside fro using the Hall sensors for positionin om ng informatio one can als discard with sensors entire on, so h ely and opera on sensorle control. Fo this, the ba ate ess or ack electroma agnetic force (B BEMF) genera by the mot ated tor is measu ured on the floating p e phase of ea ach comm
utat tion state. Here commutation occurs when a e, n zero-cross sing signal is detected (see Figure 9). Su uch signal occ every 30 e curs electrical degre ees. Sinusoida commutatio al on The six-st commutati is thus bes suited for high tep ion st speed app plications, wh here torque rip pple is of litt tle importanc and hardly n ce noticeable due to the inertia of e the system at high speed At low spe torque ripp m ds. eed, ple becomes prominent, m making the sy ystem harder to control. For these typ of applica F pes ation, sinusoid dal commutat tion can be use ed. Six-step commutatio can be con p on nsidered a low wresolution control algorithm as com n mmutation occu urs only at every 60 el lectrical degrees. It is th his roughness or low resolu s, ution, that gen nerates the torq que ripple. In contrast, sinu usoidal commu utation is a hig ghn this algorithm each phase is m, resolution control. In t energized by a sine-shap analog vol d ped ltage, shifted 12 20 degrees ap (see Figure 10). part
is here is a notic ceable ripple in n With thi approach, th the torque output wit magnitude up to 13% e th % maximum torque. Figure 8 il llustrates thi is on. served midway y phenomeno Maximum torque is obs between ea commutatio when the le angle is 90 ach on, ead 0º
Figure 8 Torque ripples observed when applying six-step a commutation a dashed line indicates a comm n; mutation step.
rrgggTorque ripple hinder control at lower speeds rs s, making thi control strat is tegy unsuitable for low-speed e or precise-position co ontrol applic cations. Othe er re lified for such h algorithms exist that ar better qual purposes. Sensorless control s
Figure 10 Black lines represent sine waves voltag 0 e ges generated by sinusoida commutation Dashed lin al n. nes mmutations. Gre lines represe ey ent represent 60-degree com enerated by six-s control. step voltages ge
The sin nusoidal voltag shape is gen ge nerated by PW WM signals. Thus, voltage a each phase v T at varies graduall ly, in contras with the disc st crete, 60-degre steps of block ee commutat tion, which ten to cause ha nds armonics. Torque generated b sinusoidal control can be e by expressed in the followi equation: d ing Ttotall=K*I*(sin(θ)+ sin(θ+120)+ sin(θ+240)) + s Where K is the moto constant and I is the avera or d age hat ases. The above current th flows into the three pha equation can be simplifi to: c ied
Figure 9 Vol ltage on one phase solid line repres e; sents actual voltag ge measured; do line is an appr otted roximation of the b EMF back
6
Ttotal=1.5*K*I This simplified equation shows that the resulting torque is no longer dependant on the position of the rotor. This implies that with a correctly executed sinusoidal commutation, the resulting torque vector is constant and the motor rotates smoothly. The position of the rotor is still needed, however, to determine when to commutate and to maintain a leading angle of 90 degrees between the stator and rotor fields. Three Hall sensors do not provide a high enough resolution. An optical encoder or a resolver mounted on the shaft can be used. Higher position sensor resolution allows for higher resolution sinusoidal signals, which result in smoother operation (see Figure 11).
possible to use optimize motor efficiency. Information on the current in each phase of the motor is obtained either by measuring the current through each shunt resistor in a three-shunt bridge or by reconstructing the current through each coil using information from a single shunt resistor (see Figure 12). This is done by measuring current through the single shunt resistor at strategic times.
Figure 12 (a) Three-shunt bridge (b) Single-shunt bridge
Figure 11 (a) 48-step sine wave; values are observably discrete (b) 192-step sine wave; the wave is visibly smoother
It is possible to execute sinusoidal commutation using three Hall sensors and some peripherals of th
e dsPIC architecture or the same sensors and an FPGA. These approaches are elaborated upon in the implementation sections. Field oriented control Field oriented control (FOC) operates based on the principal that only the portion of the stator current vector that is perpendicular to the rotor helps generate torque to propel speed. Thus, to increase torque and power efficiency, it is practical to control the current vector in such a way that the stator’s current vector is perpendicular to the rotor’s position at all time. The part of the current that is parallel to the rotor, called direct current (Id), should be kept ideally at zero, while the part of the current that is perpendicular to the rotor, called quadrature current (Iq), is dependent on desired motor speed. Thus, by measuring and adjusting stator current relative to the rotor, it is
In this thesis, a single-shunt bridge is used and the current through each stator coil is reconstructed. Implementation details can be found in the FPGA implementation section. To know and control the current relative to the rotor, a number of axis transforms is employed. First, the Clarke transform is used to convert the measured stator current values from the three-phase stationary axis corresponding to each coil of the stator to a twophase stationary axis (see Figure 13).
Figure 13 The Clarke transform converts current vector from three-phase stator reference axis (a, b, and c) to twophase stator reference axis (beta and alpha); is represents stator current vector
Park transform is then used to transfer the current vector from the two-axis stator reference to a two-axis rotor reference (see Figure 14). As the motor is on,
豫公网安备41160202000603
站长QQ:729038198
关于我们
投诉建议