1 Introduction
While section 1 introduced the selection and specification of EV motors and control circuits, this section shows how system and detail design can in themselves produce very worthwhile improvements in efficiency which can define the viability of an EV project. The section opens with discussion of the recently introduced brushed DC motor, by Nelco Ltd, for electric industrial trucks, then considers three sizes of brushless DC machine for electric and hybrid drive cars, before examining the latest developments in motor controllers.
2 Electric truck motor considerations
EV motor makers Nelco say the requirements for traction motors can be summarized as light weight, wide speed range, high efficiency, maximum torque and long life. The company recently developed their diagonal frame Nexus II motor, for general electric truck operation.
In this motor, Fig. 3.1, active iron and copper represent 50 and 30% respectively of the motor weight. Holes in the armature lamination, (a), have resulted in some weight reduction and the use of a faceplate commutator, (b), has also helped keep weight down - with only 30% of the copper required for a barrel-type commutator - because the riser forms part of the brush contact face. With use of aluminum alloy for the non-active parts, such as brush holders (c) of the motor, weight of the 132 L motor is held to 80 kg, a power to weight ratio of 450 watts/kg. Tolerance of high accelerations comes from perfection of the faceplate commutator to retain brush track surface stability. Usually the constraint on high power at high speeds, particularly when field strengths are reduced, is commutation ability, Nelco maintains.
The patented segmented frame of the Nexus, (d), makes the provision of interpoles quite an easy option - to optimize commutation at all current loadings, so reducing brush heating losses and compensating for interpole coil resistance losses. As output torque is a function of armature current, flux and the number of conductors, all these must be maximized. Short time high current densities, over the constant torque portion of the performance envelope, are possible given adequate cooling. Cost is held down by such measures as use of a segmented yoke/pole assembly, (e); extruded brush holders are also used, (f). Figure. 3.2 shows rating and efficiency curves for the N180L machine.
3 Brushless DC motor design for a small car
In this case study of the design of a 45 kW motor 1 commissioned for a small family hatchback - the Rover Metro Hermes - the unit was to give rated power from 3600-12 000 rpm at a terminal voltage of 150 V AC. The unit has been tested on a dynamometer over the full envelope of performance and methods for improving the accuracy of measurement are discussed below. The results presented show a machine with high load efficiency up to expectations and the factors considered are important in minimizing losses.
3.1 BRUSHLESS MOTOR FUNDAMENTALS
A key aspect of motor design for improved performance is vector control, which is the resolution of the stator current of the machine into two components of current at right angles. Id is the reactive component which controls the field and Iq is the real component which controls the power. Id and Iq are normally alternating currents. In this example, Fig. 3.3, the machines being considered are of the rare-earth surface-mounted magnet type with a conventional 3 phase stator and a rotor consisting of a magnetic flux return with a number of motor pole magnets mounted on it. The open loop characteristics of the machine are considered as follows: if the shaft of the motor is driven externally to 12 000 rpm a voltage of 260 V will be recorded, (a). In this condition with full field at maximum speed, iron losses will be high and the stator will heat up very quickly. At this operating point the motor could supply about 135 kW of power. However, this is not the purpose of the design, (b).
Fig. 3.3 Example brushless motor characteristics: (a) no-load terminal voltage when machine is operated as a generator;
(b) variation of machine terminal voltage with torque and speed (left) with variation of power factor with torque and speed (right); (c) vector diagram (right) of PMB DC motor (left), in field weakening condition 12 000 rpm no-load.
Id
The torque-speed requirement for a typical small vehicle is shown to be constant torque to base speed (around 3600 rpm) then constant power to 12 000 rpm. This assumes a fixed ratio design speed reducer. During the first region the voltage rises with speed. In the second region the voltage is held constant at 150 V by deliberately introducing a circulating current - Id which produces 152 V at 12 000 rpm to offset the 260 V produced by the machine, to leave 150 V at the machine terminals. The circulating current produces this voltage across the inductance of the machine winding. It also produces armature reaction which weakens the machine field; total field = armature reaction + permanent magnet field gives a lower air gap flux and lower iron losses. This mode of operation is known as vector control. What happens if we reverse the direction of Id? Theoretically we strengthen the field. However, with a surface mounted magnet motor the machine slows down due to the effect of the circulating current on the machine inductance. However, the torque per amp of Iq current remains constant.
If we supply the motor from a square wave inverter we observe some interesting phenomena when we vary the position of the rotor timing signals. In the correct position the stator current is very small. When the current lags the voltage the motor slows and produces current with sharp spikes and considerable torque ripple. When the current leads the voltage the motor runs faster and produces a near sine wave with smooth torque output. It is the field weakening mode we wish to use in our control strategy, (c).
3.2 MOTOR DESIGN: METHOD OF MEASUREMENT
In the following account details are given of the motor design, Fig. 3.4, and of the predicted and measured efficiency maps. The measured efficiency maps were carried out using a variable DC link voltage source inverter. Polaron conducted the trials with two waveforms: a square wave with conduction angle 180° and a square wave with harmonic reduction, conduction angle 150°, the purpose being to assess the effects of the harmonics on motor performance, (a).
--
a = 150° audible a = 180° audible SPEED V I P noise V I P noise 1000 29V 7.3A 75W 52dB 28V 12.4A 72W 54dB 2000 55V 8.1A 216W 54dB 55V 12.8A 216W 54dB 3000 82.6V 8.4A 396W 56dB 84V 13.2A 405W 55dB 4000 113V 9.12A 540W 56dB 110V 13.6A 630W 56dB 5000 138V 9.12A 765W 58dB 137V 13.8A 900W 57dB 6000 150V 25A 990W 59dB 150V 24A 1080W 58dB 8000 150V 87A 1440W 60dB 150V 84A 1800W 63dB 10000 150V 122A 2250W 67dB 150V 123A 2700W 69dB Stack OD 220 mm Stator mass 14.1 kg Stack ID 142.5 mm Rotor mass 4.12 kg Length 80.5 mm Total mass 34 kg Overall length 140.5 mm Rotor inertia 0.016 kg m 2 V Pole number 16 Thermal resistance 0.038°C/Watt Peak torque 200 Nm Thermal capacity 6000 joules/°C Motor constant km RMS 3.03 Nm/sqr (W) Rotor critical speed 21€000 rpm Motor constant km DC2 89 Nm/sqr (W) Nominal speed 12€000 rpm Electrical time constant 10.4 millisecs Back EMF at 12€000 rpm = 260 V Mechanical time constant 1.9 millisecs Winding resistance 0.096 ohm Friction 0.171 Nm Winding inductance 100 micro-henries Motor torque constant 0.3 Nm/A Vector control voltage 150 V Winding star connected RMS line to line
(a) (b)
--
Fig. 3.4 Motor design data: (a) XP1070 machine data; (b) no-load losses (machine only).
The measurement of electrical input power is accurately achieved using the 'three wattmeter' method. Measurement of mechanical power is more difficult. Polaron found it necessary to mount the motor into a swing frame with a separate load cell to obtain accurate results at low torque.
Even so, other problems such as mechanical resonances and beating effects at 50 Hz harmonics require care in assessing results. The operating points were on the basis of maximum efficiency below 150 V AC terminal voltage.
Results are in the form of three efficiency maps which give predicted and measured performance on both waveforms. The losses in this type of motor are dominated by resistance at low speed and iron losses at high speed. What the results show is that low speed performance was accurately predicted but high speed performance was less efficient especially at light load. The reason for this is that the iron loss at 10 000 rpm, no-load, should be about 1000 W, sine wave, (b). With 150 V terminal voltage the measured figure was 2200 W. The following paragraphs discuss the factors affecting this result but it is believed that the main contributors are larger than expected hysteresis losses due to core steel not being annealed, and larger than expected eddy current losses because of lower than specified insulation between laminations.
Annealing causes oxidation of the surface of the steel, leading to improved interlayer insulation.
Polaron subsequently coat the laminations with epoxy resin then clamp them in a fixture to form a solid core for winding.
3.3 MOTOR DESIGN FACTORS AFFECTING MACHINE EFFICIENCY
For the stator the important factors are: (i) shape of lamination - optimized lamination has a much larger window than 50 Hz induction motor lamination and a bigger rotor diameter relative to the stator diameter; (ii) use of high nickel steels is counteracted by poor thermal conductivity. Thin silicon steel with well-insulated laminations gives best results. Laminations should be annealed and not subjected to large mechanical stresses. The core can be a slide fit in casing at room temperature as expansion due to core heating soon closes the gap. Stator OD should be a ground surface; (iii) winding must be litz wire and vacuum impregnated to ensure good thermal conductivity.
Varnish conducts 10 times the heat of air gap.
For the rotor the main ones are: (i) if magnets are thick (10 mm in this case) mild steel flux return is satisfactory; (ii) magnets are unevenly spaced to remove cogging torque; (iii) individual poles must not contain gaps between magnet blocks making up the pole. Such gaps lead to massive high frequency iron losses. This can be checked by rotating the machine at lower speed and observing the back-EMF pattern. If there are sharp spikes in the wave form the user will have problems with losses.
3.4 MOTOR CONTROL
Battery operated drives must make optimum use of the energy stored in the battery. To do this, the efficiency of both motor and driveline are critically important. This is especially true in vehicle cruise mode typically two-thirds speed one-third maximum torque, therefore Polaron proposed to build a drive with two control systems: (i) current source control in constant torque region and (ii) voltage source operation in constant power region. At 45 kW 6000 rpm we would expect I L 175 A,
V AC 150 V; inverter switching loss 10 kHz, 1.8 kW; converter saturated loss 0.9 kW, using PWM on the windings and IBGT devices.
If, however, we use a square wave at the machine frequency, Fig. 3.5, and the machine operates with a leading power factor, the switching losses are greatly reduced for additional iron loss, of 225 W, at top speed. The inverter efficiency increases from 94% to 97%. In the low speed constant torque region there is no alternative to using PWM in some form.
Fig. 3.5 Motor line current waveforms.
4 Brushless motor design for a medium car
4.1 INTRODUCTION
Here the task is to optimize the 45/70 kW driveline for the family car of the future 2 . This involves improvements in fundamental principles but much more in materials and manufacturing technology.
The introduction of hybrid vehicles places ever greater demands on motor performance.
It is the long-term aim of the US PNGV program to reduce the cost of 'core' electric motor and drive elements to 4 dollars per kW from around 10 dollars charged in 1996 for introductory products supplied in volume. The price may be reduced to 6.5 dollars using new manufacturing methods to be reviewed below. Further savings may come from very high volume production.
This will require significant investment which will not occur until there is confidence in the market place and technical maturity in a solution. In terms of design, we may increase speed from 12 000 to 20 000 rpm. For reasons to be explored, a further increase becomes counterproductive unless there is a breakthrough in materials. In the inverter area Polaron believe the best cost strategy is to use a double converter with 300 V battery, 600 V DC link and 260 V motor. This assumes power levels of 70 kW.
The motor can be induction type or brushless DC. Induction is satisfactory in flat landscape/ long highway conditions. For steeper terrain, and shorter highways as exists in Europe brushless DC is more suitable - especially for high performance vehicles and drivelines for acceleration/ braking assistance in hybrid vehicles. Excellent progress has been made in the silicon field. The introduction of high reliability wire bonded packaging in association with thin NPT chip technology for IGBTs is reducing prices and improving performance. Currently a 100 A 3 phase bridge costs around $100 in volume. The arrival of complete 3 phase bridge drivers in a single chip at low cost is a further improvement in this area. Individual driver chips provide better device protection and drive capability at this time.
Great progress has been made in batteries in recent years. However, the time has come for a change in emphasis. Previously the pure battery electric was seen as the desired solution. Even if the remaining technical issues can be addressed, we are still impeded by weight and cost of such a solution. Consequently Polaron believe they should focus on hybrid solutions and this needs batteries optimized for peak power not energy capacity. It requires batteries with geometries optimized for peak power - ultra-low internal resistance and perhaps high capacitance at the same time. It will certainly require new packaging. A capacity of 2 kWh at 2 minute rate would be adequate for the average family car. It will also require a low cost short-circuit device to bypass high resistance cells in long series strings.
There is now little doubt that brushless DC machines offer the best overall performance when used in vector control mode, with high voltage windings, Fig. 3.6. The reason is that the brushless DC motor offers the lowest winding current for the overall envelope of operation. An electric vehicle has to provide a non-linear torque/speed curve with constant power operation from base speed to maximum speed. In a brushless DC motor, the motor voltage may be held constant over this range using vector control. In an induction motor, the motor voltage must rise over the constant power speed range. If V and I are the voltage and current at maximum speed and power the values at base speed are V (Base Speed/Max Speed)1/2 , I (Max Speed/Base Speed)1/2 . If maximum speed / base speed = 3.5 times, the current at base speed is 1.87I. Consequently the induction motor inverter requires 1.87 times the current capacity of the brushless DC motor inverter.
Fig. 3.6 Current designs of vector controlled brushless DC machines. The most significant improvement recently for brushless DC machines has been the development of the Daido magnet tube in Magnaquench material. This product offers the benefits of high energy magnet and containment tube. This leads to a third benefit which is not immediately obvious but very significant. Surface magnet motors usually employ a containment sleeve which adds several millimeters of air gap to the magnetic circuit. Since magnet tube does not require the same air gap flux density. The benefit is reduced magnet weight for a given motor design.
For example, 140 mm diameter Daido grade 3F material with a 5 mm wall will operate unsupported to 13 500 rpm.
The rotor of the machine, Fig. 3.7, is assembled with the magnet tube glued to the flux return tube, with the magnets de-energized. The pole pattern is applied with a capacitor discharge magnetizer from inside the flux return tube. The end plates and motor shafts are then fitted using a central bore for precise axial alignment. Use of a solid rotor is not practical unless a rotor material which does not saturate until 3 tesla is used. Since such material costs
$50 per kg the hollow tube is the best alternative. The use of magnet tube makes complete automation of rotor construction possible achieving significant savings in labor costs, Fig. 3.7a.
Fig. 3.7 Rotor design and machine performance: (a) a 150 kW, 20 000 rpm brushless DC stator-rotor; (b) power/speed for brushless DC motor with 3.5:1 constant power speed range. Many designers are attracted by the possibility of running motors faster than the current 12 000 rpm. The objective is to reduce the peak torque requirement in an effort to reduce weight and cost of active materials. One obvious method is to compromise the constant power over the 3.5:1 speed:range requirement. Polaron's own investigations into faster speed suggest any increase above 20 000 rpm will be counterproductive. There are many reasons for this:
(a) The maximum frequency of operation is limited to 1500 Hz using Transil 315 in 0.08 mm thickness (3.15 W/kg at 50 Hz). Most designers are concerned with no load line losses and are endeavoring to optimize this.
(b) Consequent on (a), as the speed rises above 20 000 rpm the pole count has to be reduced from 8 to 6 to 4 poles. This results in thicker magnets and longer flux return paths.
(c) Optimum machine geometry is rotor OD = stator length. The Polaron 70 kW machine has rotor OD = 140 mm and rotor length of 95 mm which is close to optimal. The machine has 8 poles and gives 70 kW from 4000 to 13 500 rpm.
(d) Machines that are below 100 mm rotor diameter are not easy to make as the windings cannot be inserted by automatic machinery. This is especially true of heavy current windings.
(e) Machines with low pole count have poor rotor diameter to stator diameter ratio, which increases the mass of stator iron and results in large winding overhangs increasing copper losses.
(f) Laminations for these machines should have a large number of teeth to reduce the thermal resistance from copper to water or oil jacket. The limitation is when the tooth achieves mechanical resonance in the operating frequency range of the machine. Typically it is the 6f component that causes excitation (6f = 6 times motor frequency). Silicon steel (Transil) has good thermal conductivity. High nickel steels such as radiometal exhibit poor thermal conductivity but lower a sleeve if used within its speed capability, a thinner magnet tube is possible whilst maintaining iron losses. Machines with a high peak torque requirement are better in Transil where the copper losses of peak torque can be safely dissipated.
(h) If a better core material at a sensible price were available it would be a real boon. This is one area where there is much room for improvement. Polaron are aware of powder core technology using sintered materials but the tooth tip flux density is only 0.8 tesla. Ferrites are worse at 0.5 tesla.
(i) If makers are prepared to use containment sleeves, a power-speed graph for high speed radial brushless DC machines would look like that in Fig. 3.7a (based on 3.5:1 constant power torque/speed curve). This is the maximum power achievable in consideration of dynamic stability requirements. This graph assumes two point suspension and that the first critical speed must be 20% higher than the top speed of operation (25 kW rotor from 25 000 to 80 000 rpm would be 57 mm OD 100 mm long).
(j) One problem with high speed machines is the increased kinetic energy stored in the rotor.
This can place a severe strain on subsequent speed reducers unless torque limiting devices are provided.
(k) Acoustic noise is often severe at high speed. For a reduction try: (i) impregnation of stator;
(ii) removing sharp edges on outside of rotor; (iii) operating rotor at reduced pressure using magnetic seals or (iv) using machine with liquid cooling jacket.
(l) Speed reduction is another difficult area at high speed. Since torques are low, friction speed reducers are quieter than gears by a factor of ten.
(m) Bearings and mechanical stability are challenging problems at turbo-machinery speeds.
Polaron believe the best cost/performance ratio can be achieved for 70 kW system by: (1) using a Transil 315 stack 0.08 mm thick made as a continuous helix using the punch and bend technique;
(2) using a rotor made from 5 mm magnet tube of surface mount structure mounted on 12.54 mm of 14/4 stainless steel; (3) magnetizing the rotor after assembly to flux density of 3 tesla for 2 milli-secs for maximum flux density; (4) choosing a stator frequency of less than 1500 Hz, mean air gap flux density 0.6 tesla; (5) using a liquid cooled stator; (6) insulating the stator from earth with low capacitance coupling; (7) choosing stator of 215 mm OD with 48 teeth stack of 95 mm giving 70 kW from 4000 to 13 500 rpm. Alternatively a stator of 185 mm with 24 teeth and rotor of 110 mm OD x 140 mm long will give 70 kW from 6000 to 20 000 rpm; (8) winding the machine for 460 V in constant power region (460 V at 4000 rpm) with machine driven as a generator open circuit. This gives good efficiency and substantial winding inductance to minimize carrier ripple, Fig. 3.7b.
5 Brushless PM motor: design and FE analysis of a 150 kW machine
5.1 INTRODUCTION
High speed permanent magnet (PM) machines with rotor speed in the range from 5000 to 80 000 rpm have been developed 3 , applications of which include a gas turbine generator with possible application in hybrid electric vehicles. The motor considered below runs at infinitely variable speeds up to 2 kHz at full power and has been designed for different requirements at an output power of 150 kW. Machine parameters have been calculated from software package 141 developed at Nelco Systems Ltd.
The drive system of this design consisted of a brushless DC machine and an electronic inverter (a chopper and DC link) to provide the power. The performance parameters set out are aimed to producing a design specification of the machine shown in Fig. 3.8, Fig. 3.9 showing the machine controller.
In the initial stage, a detailed specification was set out for the peak torque performance of 150 Nm from 10 000 to 20 000 rpm, the no-load back-EMF at 20 000 rpm of 600 V(RMS); the total number of poles are 8 (1.33 kHz at 20 000 rpm), and the maximum total weight is 45 kg.
5.2 ROTOR AND STATOR CONFIGURATION
Main constraints were found to be weight and inductance; in the high speed application it is important to keep the weight to a minimum, therefore a ring design is the most suitable which means a sufficient number of poles is required on the rotor, 8 poles in this case. The main advantage with this configuration is that the return path for the magnetic circuit in the core and yoke is much smaller in cross-section area (the thickness of the ring has been considered within the customer's shaft requirements). While 8 pole design was found to give the best solution, a 16 pole design was also considered which resulted in lower weight, but was rejected because the return path had to be increased, in that area, to give sufficient mechanical strength to the unit. In general a machine of high number of poles, at high frequency, produces high specific core loss and the reduction in the stator mass meant that the total core loss was a few watts more. To achieve the required winding inductance, careful attention had to be made to the shape of the stator lamination so as to reduce the slot leakage. The reduction of current density in the copper conductors has also been considered, but the slot shape and area have had an effect on the winding inductance. The final lamination design has been optimized for minimum slot leakage, to achieve the required performance.
Fig. 3.8 150 kW PM brushless machine.
5.3 MAGNETIC MATERIAL SELECTION
High energy-density rare earth magnets, of samarium cobalt, have been chosen in this design because of the material's higher resistance to corrosion, and stability over a wide temperature range. Also it has a high resistance to demagnetization, allowing the magnetic length of the block to be relatively small. This shape of block lends itself to being fixed onto the outside diameter of the rotor hub, to produce the field in the d-axis, which gives advantages of a greater utilization of the magnet material with lower flux leakage, the low slot leakage resulting in low winding inductance. The magnets in this application have been fitted with a sleeve on the rotor outside diameter, for mechanical protection and to physically hold the magnets in place. A carbon-fiber sleeve was chosen for this application; it offers at least twice the strength of the steel sleeve in tension, so a much greater safety factor can be achieved. The sleeve on the rotor increases the effective air gap but an unloaded air gap flux density of 0.6 tesla was achieved from this high energy density rare earth magnet. The core loss in the stator, due to high frequency, is considered and must be kept to an acceptable level. The grade of material considerable is radiometal 4550.
This alloy has a nominal 45% nickel content and combines excellent permeability with high saturation flux density.
5.4 MAGNETIC CIRCUITS
The magnetic circuit for this design was calculated using the Nelco software. The most important parameters in the design of the magnetic circuit were weight and to keep the core losses down to a minimum whilst reducing the slot leakage to minimize the winding inductance. This is achieved when a compromise has been reached in which the flux density in the teeth is 1.15 tesla, the density in the core is 0.8363 tesla, and the yoke flux density is 0.78 tesla.
5.5 BRUSHLESS MACHINE DRIVE
The machine drive consists of a polyphase, rotating field stator, a permanent magnet rotor, a rotor position sensor, and the electronic drive. During operation the electronic drive, according to the signals received from the rotor position sensor, routes the current in the stator windings to keep the stator field perpendicular to the rotor permanent field, and consequently generates a steady torque. Conceptually, the drive operates as the commutator of a DC machine where the brushes are eliminated. The main advantage here is that no current flow is needed in the rotor.
As a result, rotor losses and overheating are minimal, the input power factor approaches unity and maximum efficiency is obtained. This is especially relevant in continuous duty applications, where the limiting factor of traditional induction drives is invariably the difficulty of removing rotor losses.
Fig. 3.9 Machine controller.
5.6 MOTOR DESIGN: FINITE-ELEMENT MODELLING
3D finite-element modeling (FEM) was not required, as the topology of the machine in x-y plane is the same along the axial length, except at each end where the end turns winding exists. However, a 2D finite-element model has been employed for the machine to calculate and analyze the flux distribution in it, Fig. 3.10. This is done to facilitate the rotor movement relative to the stator, so that the characteristics of interest such as the flux modulation due to slot ripple effect on the magnet and the rotor hub can be examined. To carry out this kind of analysis, several meshes have to be created, one for each rotor position, and then each solved in turn. The software program has a facility for coupling meshes, using Lagrange multipliers. This technique has been used to join the independent rotor and stator meshes at a suitable interface plane, a sliding Lagrange interface being placed in the middle of the air gap. The view at (a) shows a close-up view of the joined meshes for the machine, and in (b) is the rotor of the machine at 45° from base (half of the rotor mesh is missing for clarity).
Fig. 3.10 Finite-element modeling: (a) the coupling meshes; (b) rotor at 45 o from base; (c) air gap flux waveforms; (d) contour and vector flux.
The stator winding flux linkage waveforms of the machine have been calculated from the time transient solution, as the rotor speed is dynamically linked to the program, at 20 000 rpm. The experimental phase flux linkage has been deduced by integration of the phase EMF generated from the machine at no-load. These EMFs are shown to be within 8% difference, the value calculated by FEM being the higher. The flux in the air gap was measured using a search coil that is inserted on the stator side. From this search coil, a flux waveform was recorded and it is shown together within the flux calculated from FEM in (c). The flux plot, as contours and vectors at 0° rotor position for the machine, is shown at (d).
6 High frequency motor characteristics
In the 1970s motor designers were introduced to Bipolar Darlington transistors which permitted switching up to 2 kHz at mains voltage. In the 1980s insulated packaging was mastered and motor costs have been reduced. In the 1990s we have the IBGT which permits operation to 16 kHz for the first time at high power. This gives the designer a new freedom 4 . Hitherto the market sector has been dominated by 50 Hz machine designs. Now we can choose our operating point so the question must be asked: what is the optimum point and which is the best type of motor? There is no simple answer to this question. We have several types of machine each with characteristics which are good in particular tasks. What is certain is that whatever type of machine is used, it can be made smaller than its 50 Hz counterpart by using a high frequency design.
During the next ten years lies the challenge of the hydrogen economy with an increased demand for electric drives. IBGTs make new inverter topologies possible. The inverter on a chip in the back of the motor is now a reality.
6.1 HF MACHINE PROPERTIES
Motors designed for high frequency operation are of many types; however, they all share common design attributes. The 50 Hz motor designer will be used to the idea that at the full-load operating point copper loss = iron loss. This is not true for HF machines - iron loss dominates, accounting for up to 80% of the losses. Another factor is the power density which is in general 5-20 times greater.
The use of HF windings means that the number of turns on a winding is reduced. So a high frequency motor can be expected to have much lower winding resistance and inductance than a 50 Hz machine.
For good loss management it is necessary to minimize the weight of core material. Generally, the flux density at the tooth is greater than in the main body of the core. It is common for all 50 Hz machines to use 2 or 4 pole windings; on HF machines, 8-32 poles are much more common.
Machines with a high pole number have a much smaller diameter build-up on the rotor; for a given stator OD the designer achieves a bigger rotor diameter which gives more torque and reduces stator mass. Machines with large numbers of poles are much easier to wind with only short winding overhangs. This is important because the overhangs contain the winding hot spots. See the example below.
--
Dl60 frame IM 380 V 50 Hz motor 1500 RPM 12 000 RPM
Power 11 kW 60 kW (air cooled) (water cooled) Frequency 50 Hz 400 Hz Resistance 0.5 WL/L 0.2 WL/L Inductance 2.5 mH 400 mH Stator flux density 1.5 tesla 0.75 tesla
- Currently 500-1000 Hz represents the optimum operating point for stator iron. HF machines are very suitable for use in non-linear torque speed regimes because it is possible to operate at much higher flux densities at low speed. We therefore need to investigate vector control characteristics.
6.2 VECTOR CONTROL
Understanding of this subject has been delayed by years with torturous mathematical explanations of how it is achieved. In practice, vector control is a powerful technique because: (1) the full power of the stator controller can be brought to bear on the field system; (2) only a single winding set is involved. The stator current has two components: (a) field component Id and (b) real power component Iq. As these axes are at right angles, they may be independently controlled so long as the field is capable of supporting the demanded torque.
Vector control is nothing more than power factor control. The reactive element controls the field and the real power element controls the generated torque. In induction motors there is an added complication; there has to be slip between the rotor and stator to create rotor current for producing the field. This involves an axis transformation which makes for all the difficult mathematics. Synchronous motors are much easier; vector control only involves manipulation of phase shift.
Permanent magnet machines offer great flexibility because it is possible to manipulate the field with vector control currents. This has no damaging effect on the magnets so long as the material has a recoil permeability of unity (or a linear 2nd quadrant demagnetization curve) such as ferrite and samarium cobalt.
However, the level of ampere turns needed to control the field varies dramatically between different types of machines in accordance with the magnetic reluctance in the d-axis. It may be seen that this becomes more critical in HF machines which have smaller numbers of turns on the stator, for example a machine with four sets of windings per phase. If windings are arranged in star, Fig. 3.11a, generated back-EMF is 380 V at 2800 rpm, or by letting the circulating current at 100% field be 1 and rearranging the windings in parallel delta, as at (b), an alternative situation arises.
Now 380 V is produced at 30 000 rpm and the current for 100% field increases to 4(3)1/2 I or 6.82I.
To give some idea, I is approximately 30 A for a 500 nm surface mounted PM magnet machine.
It may seem attractive to do away with the permanent magnets altogether. In practice this is not a good idea because the machine has a poor power factor and requires an oversize inverter. However, there is a variation on the concept which is possible, called the switched reluctance motor. This machine ignores Fleming's LH rule and instead relies on the attraction forces between an electromagnet and soft iron. The problem is that the production of torque is not smooth; however, they are suitable for use in difficult environments.
Fig. 3.11 Star, left, and parallel-delta, right, winding.
6.3 OPEN LOOPS OR CLOSED LOOPS FOR INDUCTION MOTORS
In the early days of inverter drives, open loop operation of induction motors was the main objective. Generally this is satisfactory over a 10:1 speed range but is problematical at slow speed due to: harmonic torques; stability problems - especially with low load inertias; and lack of rotor cooling.
Vector control may be used to improve stability and can be applied on an open loop basis. To do this, estimates are used for the load inertia/rotor current and lead to errors where fast dynamics are involved. However, Jardin and Hajdu wrote one of the leading papers on this subject whilst developing the Budapest Tramcar drive system 6 .
As the motor frequency is increased, the low winding resistance makes eddy current losses, induced by DC circulating currents in the windings, a problem. Voltage source inverters without active balancing are unlikely to be satisfactory.
One machine which gives excellent performance on an open loop control is the buried PM motor developed by Brown Boveri/CEM/Isosyn and now also produced by GEC/Alsthom Parvex.
Ken Binns at Liverpool University is a well-known authority in this area.
For fast dynamics and tight control there is no substitute for proper closed loop operation of a permanent magnet machine using vector control. Such an arrangement can give a constant power torque/speed range of 4:1 and this can be increased by using winding switching up to 70:1. Such systems are ideal for traction drives in vehicles with torque bandwidths of up to 1 kHz.
6.4 INDUCTION MOTORS
Of the various types of motor, Fig. 3.12, induction motors (IMs) are practical up to about 30 kW and 15 000 rpm. Beyond these limits exhausting the rotor losses is generally a problem (at 1500 rpm megawatt level machines are commonly constructed). At 15 kW IMs run satisfactorily up to 100 000 rpm but special motor construction techniques are needed to give strength to the cage.
Water cooling is used at high power. A typical specification, for example, might be 36 kW, 12 000 rpm, PF 0.9 at 400 Hz, 36 kW - 4 pole: efficiency 0.9 at 400 Hz, 36 kW, 380 V, 68 A line current;
slip (pure aluminum cage) 50 rpm cold, 70 rpm hot - torque 29 Nm (0.7 tesla), hot rotor diameter 6 in, active length 4 in, stator 9 in OD, peak torque at low speed 100 Nm (1.5 tesla), rotor cooling 8 CFM compressed air at 17 Psi, stator cooling 4 liters water/minute.
6.5 SURFACE MOUNTED MAGNET SYNCHRONOUS MACHINES
This design is first choice for high power drives. The rotor consists of a steel sleeve to which the magnets are glued and a containment band fitted on the outside. This fits inside a standard stator with water jacket. This design is practical up to 0.5 megawatts at up to 100 000 rpm and is used for traction drives.
Another benefit is that the output frequency is no longer related to shaft rpm and multi-pole designs/speeds over 3000 rpm may be considered. Using vector control, voltage and frequency may be separately controlled and much faster speed of response can be achieved. Many people are wary of PM designs because of concern about high temperature performance. The latest Nitromag alloys operate up to 250 o C. These use nitrogen as the alloying element and are being investigated as part of the Joint European Action on Magnets Program.
Most commercial motors use samarium cobalt of the 1/5 variety which has superior mechanical properties to the 2/17. Generally speaking, alloys of 20 MGO are in common use and the trick is to design rotors around standard size blocks, 1 x 1/2 x 6 inches thick. Modern high coercivity magnets need very large currents to demagnetize the magnets and typically 3 tesla are needed to achieve full initial magnetization for about 1 millisecond.
Fig. 3.12 Motor types. (a) INDUCTION MOTOR (c) CLAW TYPE PM MAGNET MOTOR; (b) SURFACE MOUNTED PM MAGNET MOTOR (d) UNIO TYPE MOTOR (Contentional construction); (e) BURIED MAGNET MOTOR (4 POLE) (g) HOMOPOLAR MACHINE; (f) SWITCHED RELUCTANCE MOTOR (h) RELUCTANCE MOTOR
Typical machine specification for 60 kW, 10 000 rpm (surface mounted) would be: stator OD 10 in, rotor diameter 7 in, active length 3 in; operating point 0.7 tesla at 666 Hz, 8 poles, 380 V, 103 A, efficiency 0.97, power factor 1; winding resistance 0.015 L/L, winding inductance 300 mH L/L; iron loss 1.5 kW at 666 Hz, core Transil 270 0.35 mm non-orientated; load torque 57 Nm, peak torque 150 Nm, vector control current 100 amps for 0.7 tesla.
6.6 IRONLESS PM SYNCHRONOUS MOTOR
This machine has been developed by UNIQ (USA) for hub mounted motors for use in electric vehicles. It consists of a machine with both an internal and external rotor which are mechanically linked and a thin stator winding which is usually fabricated using printed circuit techniques. The result is a lightweight machine with a very high power density and low winding inductance since there is no stator iron. Performance is largely determined by the quality of permanent magnet used. The d-axis reluctance is high due to the double air gap so that the currents needed for vector control can be large compared with a conventional PM machine. Such machines have been built up to 40 kW rating at 7500 rpm with epicyclic speed reducers that are wheel-mounted.
At present such machines are costly to manufacture because of the large amount of PM material involved, which has to be of the cobalt/neodynium variety to achieve good performance. Losses are all due to stator copper which is generally operated at extremely high current density to give a very thin stator.
6.7 WOUND ROTOR SYNCHRONOUS MACHINE WITH BRUSHLESS EXCITATION
This machine is sometimes used for inverter drives in addition to the well-known use as an electricity generator. The presence of the exciter/rectifier means that this solution is applied at higher powers.
The rotor can be salient pole or of surface slot construction at high speed. Whichever solution is chosen, the full field thermal loss in the motor is significant and a particular problem if the machine is to be run slowly at high load torques. This type of machine is used in traction drives using thyristor-based converters.
Fig. 3.13 Wound rotor synchronous machine with brushless exciter. SHAFT WOUND ROTOR; (SALIENT POLE TYPE)
7 Innovative drive scheme for DC series motors
Many DC brushed motor drive schemes for EVs use a DC shunt motor and it has been suggested that such a solution is the most appropriate 5 . This section investigates an alternative solution.
There are many railway locomotives which successfully use series wound motors and we hope to establish that indeed this is the best solution for electric vehicles.
7.1 MOTOR DRIVES: WHY CHANGE THE SYSTEM?
Because the system is already subject to change brought about by new requirements and developments. First, we have the introduction of sealed battery systems. These will permit much higher peak powers than hitherto possible and consequently will run at high voltages. 216 V DC is a common standard working with 600 V power semiconductors. Second, we have the introduction of hybrid vehicles. This will result in the need for drives and motors to operate for long sustained periods - previously batteries did not store enough energy. Third, the DC series motor has the right shape of torque-speed curve for traction, constant power over a wide speed range. Fourth, DC series field windings make much better use of the field window than high voltage shunt windings where much of the window is occupied by insulation. The series field winding is a splendid inductor for use in battery charging mode. Losses in series mode are significantly reduced.
Fig. 3.14 Motor characteristics. Torque Speed Curve
An example specification is typified by the Nelco N200, Fig. 3.15(a), which compares with a 240 mm stack, Fig. 3.15(b):
Shunt field Series field
N = 227 N = 12 Hot resistance 7 Hot resistance 0.014
Watts 700 at 10 A Watts 500 at 189 A So why hasn't somebody attempted to use series motors in EVs before? They have for single quadrant low voltage systems but not on multi-quadrant, high voltage schemes. This account proposes a new control concept akin to vector control for AC machines. We will show how it is possible to achieve independent control of field current I f and armature I a , with very fast response, using a transistor bridge.
7.2 VEHICLE DYNAMICS AND MOTOR DESIGN
A vehicle represents a large inertia load with certain elements of resistance some of which increase with speed; see section 8. For a small family car, mass = 1250 kg at 60 mph (26.8 m/sec) typical cruising speed. Windage accounts for 6 kW, rolling resistance 2 kW and brake drag 2 kW, a total of 10 kW in steady state conditions. Windage varies as the 3rd power of vehicle relative velocity with respect to the wind.
Kinetic Energy = 1/2 MV 2 , where M = mass = 1250 kg and V = velocity in meters/sec. So we have:
SPEED (MPH) 10 20 30 40 50 60 70 80 (m/sec) 4.5 8.9 13.4 17.8 22.3 26.7 31.2 35.6 KE (kilojoules) 12.5 49.5 111 198 309 446 607 792. What this illustrates is that recovered energy below 20 mph is small, consequently regeneration only matters at high speed. It also illustrates that the inertia load, not the static resistance, is the main absorber of power during acceleration.
7.3 MOTOR CHARACTERISTICS
These are shown in the following table:
Voltage 216 V Rated power 45 kW, 1250-5000 rpm Frame D 200 M- 4 pole with interpoles Weight 170 kg
Fig. 3.15 Field windings: (a) shunt field machine; (b) 3 state strategy for series field machine.
Cooling air forced, separate fan Winding, series field 245 A/216 V full load Efficiency at full load 85% Field Resistance 10 milliohm, inductance 1.2 mH Armature Resistance 30 milliohm, inductance 260 mH inc. brush-gear interpoles Dimensions A = 490 mm, B = A + shaft, C = 335 mm, D = 350 mm; see Fig. 7.14 This illustrates that when the field current is strengthened in the constant power region, the armature voltage can be made to exceed the battery voltage and regenerative braking will take place. Below 1250 rpm plug braking must be used; however, the energy stored at this speed is small.
7.4 SWITCHING STRATEGY (SINGLE QUADRANT), FIG. 3.15
Figure 3.15(a) shows the arrangement for a 216 V, 45 kW shunt field machine with separate choppers for field and armature. There are some disadvantages with this scheme: (a) field is energized when not needed; (b) forcing factor of field is small - for a 45 kW shunt field, R = 7 ohm, I = 10 A nominal, L = 1.2 henries, t = 0.17 seconds; (c) when extended to multi-quadrant design two bridge chopper systems are needed if contactor switching is to be avoided; (d) extensive modifications are needed to provide for high power sine wave battery charging; (e) field power losses are significant (3 kW at max field).
Figure 3.15(b) illustrates the proposed new circuit which has a single 3 state switch: state (1) open-circuit; state (2) armature + series field; state (3) armature. So as an example, consider the following situation:
Full load torque at standstill Field voltage for 245 A = 2 V Armature voltage for 245 A = 16 V
Fig. 3.16 Three state circuit expanded to 4 quadrant operation.
D so with 216 V battery:
D = 2/216 in state 2
D = 16/216 in state 3 The balance of the time will be off (D = duty cycle ratio for chopper).
It can be seen that by manipulating the relative times spent in each of the states, separate control of field and armature currents may be exercised.
When the speed of the motor exceeds the base speed (1250 rpm) the back-EMF is equal to the battery voltage and the switch henceforth operates only in states (2) and (3).
Let D = duty cycle for single quadrant chopper, then V out /V in = D, hence
D 2 (V B -5 -K A wI f I a R a L a dI a /dt) = I f R f + L f dI f /dt and
V B 5 =(K A wI f + I a R a + L a dI a /dt) (D 2 + D 3 ) where
= motor speed, rads/sec
V B = battery voltage
K A = armature back-EMF constant V/amp/rad/sec (D 2 + D 3)
D 2 = duty cycle state 2
D 3 = duty cycle state 3 Other symbols are self-explanatory.
7.5 MULTI-QUADRANT STRATEGY
Figure 3.16 illustrates the 3 state circuit when expanded to 4 quadrant operation: state 1 is all switches off; state 2 either S l /S 4 or S 2 /S 3 on and state 3 is either S l /S 2 or S 3 /S 4 on. As is clear, the third state is produced by having a controlled shoot-through of the transistor bridge. It may be considered that with two transistors and two diodes in series, voltage drops in the power switching path make the circuit inefficient. In fact with the latest devices: V_ce sat for switches = 1.5 V at 300 A; V f for diodes = 0.85 V at 300 A, giving a total drop = 4.7 V. So (4.7/216) 100 = 2.3% power loss.
Fig 3.17 4 quadrant circuit. When the motor loses 15% this is a small deficiency. It represents 1.2 kW at full power. As the table illustrates in Fig. 3.16, all states of motoring and braking can be accommodated. The outstanding feature of this scheme is that the full power of the armature controller can be used to force the field, giving very fast response. From Fig. 3.16, it will be seen that the 4 quadrant circuit consists of a diode bridge D l -D 4 and a transistor bridge S l -S 4 (D 5 -D 8 ). D 9 acts as a freewheel diode when the transistor bridge is operated in shoot-through mode. Bridge D l /D 4 is required because the direction of armature current changes between motoring and braking. Control in braking mode is a two-stage process. At high speed the armature voltage exceeds the battery voltage and the battery absorbs the kinetic energy of the vehicle. At low speed the field current is reversed and plug braking of the armature to standstill is achieved via D 9 .
7.6 DEVICE PROTECTION IN A MOTOR CONTROLLER
Switches S 1 -S 4 form a bridge converter and the devices require protection against overvoltage spikes from circuit inductances. The main factors are: (1) minimize circuit inductances by careful layout. The key element is the position of D 9 and associated decoupling capacitor relative to D l -D 4 ; (2) fit 1 mF of ceramic capacitors across the DC bridge S 1 /S 4 plus varistor overvoltage protection.
D l -D 4 can be normal rectification grade components but D 9 must be a fast diode with soft recovery. D 5 -D 8 are built into the transistor blocks.
7.7 SINE WAVE BATTERY CHARGER OPERATION
With little modification the new circuit, Fig. 3.17, can be used as a high power (fast charge) battery charger with sine wave supply currents. The circuit exploits the series field as an energy storage inductor. S l and D 6 are used as a series chopper with a modulation index fixed to give 90% of battery volts. This creates a circulating current in the storage inductor. Switch S 4 and diode D 7 function as a boost chopper operating in constant current mode and transfer the energy of the storage inductor into the battery. Charging in this manner is theoretically possible up to 250 amps but will be limited by: (a) main supply available and (b) thermal management of the battery.
Fig. 3.18 Full circuit diagram of combined chopper/battery charger. Experience shows that charging at 30 amps is possible on a 220 V, 30 A, USA-style house air conditioning supply. Charging at greater currents will require special arrangements for power supply and cooling. One advantage of the scheme presented is that it may be used on any supply from 90 V to 270 V.
It is also possible to adopt the circuit for 3 phase supplies in one of two ways: (1) add an additional diode arm - this would produce a square wave current shape on the supply; (2) fit a 3 phase transistor bridge on the supply - this would permit a sine wave current in each line at a much increased cost.
7.8 POWER DIAGRAM FOR MOTORING AND CHARGING
Figure 3.18 presents the combined circuit diagram for motoring and battery charging. Reservoir capacitors and mode contactors have been added. The capacitors function as snubbers when running in motoring mode. As drawn, to adapt to battery charging, the battery plug is moved to outlet D and the mains inserted into plug B, alternatively contactors could be used to do the job. Battery safety precautions comprise: (1) the battery is connected via a circuit breaker capable of interrupting the full short-circuit current of a charged battery; (2) this circuit breaker is to contain a trip to disconnect battery by mechanical means only; (3) battery/motor/controller are each to contain 'firewire' to disconnect the circuit breaker; (4) circuit breaker is to be tripped by 'G' switch when 6G is exceeded in any axis.
Fig. 3.19 Control system in motoring mode. BASE DRIVE; BRAKING ON ACCELERATOR TO SIMULATE NATURAL ENGINE BRAKING
7.9 CONTROL CIRCUIT IN MOTORING MODE
Figure 3.19 shows the block diagram of the controller for motoring mode. The heart of the system is a memory map which stores the field and armature currents for the machine under all conditions of operation. These demands for I f and I a are then compensated for in accordance with the battery voltage before conversion into analogue form, to be passed to operational amplifier loops which drive the modulators. Current feedback is provided by Hall effect CTs. The torque loop has input from two pedals and a feedback from a torque arm attached to the motor. Above the base speed there is no open circuit condition and the armature loop error is used to control the field.
7.10 CONTROL CIRCUIT IN BATTERY CHARGING MODE
The control circuit for battery charging is shown in Fig. 3.20. When the battery is below 2.1 V per cell and 40°C it is charged at the maximum current obtainable from the supply. Above 2.1 V/cell the battery is operated at reduced charging up to 2.35 V per cell, compensated at -4 mV/°C for battery temperature. This data assumes lead-acid cells.
As can be seen from the block diagram there are two separate loops for the buck and shunt choppers. The fast current loops stabilize the transfer function for changes in battery impedance.
The current limit function must be user-set in accordance will supply capabilities.
Fig. 3.20 Block diagram of battery charging controller.
References
1. Hodkinson, R., Operating characteristics of a 45 kW brushless DC machine, EVS 12, Aneheim, 1995
2. Hodkinson, R., Towards 4 dollars per kilowatt, EVS 13, Osaka , 1996
3. Al'Akayshee et al., Design and finite element analysis of a 150 kW brushless PM machine,
Electric Power Transactions, IEEE, 1998
4. Hodkinson, R., The characteristics of high frequency machines, Drives and Controls Conference, 1993
5. Hodkinson, R., A new drive scheme for DC series machines, ISATA 24, Aachen , 1994
6. Jardin and Hajdu, Voltage Source Inverter with Direct Torque Control, IEE PEPSA, 1987
Further reading
Alternative transportation problems, SAE, 1996 The future of the electric vehicle, Financial Times Management Report, 1995 Battery electric and hybrid vehicles, IMechE, 1992 Electric vehicle technology seminar report, MIRA, 1992 Electric vehicles for Europe conference report, EVA, 1991