BRUSHLESS DC MOTORS FOR WINCHESTER DISC DRIVES

George P. Gogue

Abstract:

Spindle motors and in particular Winchester disc drives are becoming very dominant among the many applications of brushless dc motors. The flexibility and simplicity of design makes these motors ideal in an application where management of space is of great importance. The switching circuits necessary for these motors add controllability to the performance and make a high degree of speed-control a possibility. Economically, the brushless motor is acceptable even with the additional cost of the switching circuit. This has been made possible by the lower cost of electronic integrated circuits in the past few years.

Introduction:

The brushless motor is selected for Winchester disc drives for several unique reasons, chiefly, the availability of a low voltage dc supply for these drives. Space limitation requires the motor to have a high diameter to axial length ratio, and the brushless motor inherently allows that. It also is adaptable to accurate (+/- 0.1%) speed control by virtue of its electronic switching circuit. This circuit is a small part of the total electronic circuitry required in these computer peripherals and, therefore, constitutes no substantial burden on the designer especially since the technology involved in the motor switching circuit is compatible with the rest of the electronic circuitry.

Fig. 1 Typical Disc Drive Motor

Fig. 1 shows a typical brushless motor assembly for a 2-disc 5 1/4" Winchester disc-drive of the under-slung type. The discs are mounted on the hub (upper) side of the assembly and the brushless motor is on the other side (lower) of the mounting bracket. The two sides of the assembly are isolated from each other by a combination of a mechanical and magnetic seal. The motor is built with the rotor, which carries the permanent magnets, on the outside and with the winding assembly stationary on the inside. Rotation is, therefore, transferred from the rotating magnets to the hub by the shaft and bearing assembly. The only electronic components required inside the motor are 1, 2 or 3 sensors (depending on the type of switching) to produce position dependent pulses essential for commutating the winding. These sensors are usually Hall sensors producing either sinusoidal or digital pulses.

The winding consists of several coils of copper wire, wound to produce 2, 3 or 4 phases (or windings). The coils are usually wound on a slotted lamination stack of magnetic steel. Alternatively these coils can be wound on a continuous laminated ring in the form of a toroid (1).

The permanent magnets mounted on the inside of the rotor-cup are either ceramic or plastic bonded magnets. These materials are economically acceptable and at the same time produce as much magnetic flux as can be supported by the amount of steel available. Thus making the choice of Rare-earth materials for this application both useless and economically unacceptable.

The complete system of brushless motor and electronic driver is shown in the block diagram of Fig. 2.

Fig. 2 Brushless Motor & Drive

The electronic driver consists of both the logic circuit and power switching devices. The speed-control circuit is designed to operate either on pulse-width-modulation (PWM) or linear control. All the above circuits, however, require feedback signals from the motor which are rotor-position dependent. Hall sensors in the vicinity of the rotating magnets provide these signals (2). The load on the motor is either the coated media or optical discs or a tape cartridge. Operating speeds are 3600 rpm, 1800 rpm and 600 rpm, respectively.

Types of Switching Circuit:

A common situation facing the designer of the Winchester, tape or optical drive is which type of brushless motor switching is most suitable for the application. The main types are:

a) 1-phase switching

b) 3-phase switching

c) 4-phase switching

Each of the above types can be either unipolar or bipolar. This means that for the 1-phase switching, either one winding or both windings are used at any one time. For the 3-phase switching it means either one winding or two windings are used at any one time.

The 4-phase switching can be done either by switching one winding or two windings on at any one time. The selection of winding configuration is based on some or all of the following parameters:

a) The space, i.e. diameter and axial length, available for a motor capable of satisfying the output requirements.

b) The required starting torque at some or all of the possible starting positions.

c) The starting current available from the power supply and whether this current is controlled electronically or by the motor winding resistance.

d) The maximum number of power switching devices allowed as well as their type which determines the voltage drop across them.

e) The maximum tolerable torque ripple under running condition. However, the speed-control circuit is also important in minimizing this ripple.

f) The electrical and mechanical time constants required are based on switching and acceleration requirements. A high inductance can cause switching problems and hence is specified by a low electrical time constant. A high motor moment of inertia can result in a long acceleration time if the mechanical time constant is not specified.

g) Non-repetitive runout of the hub can be affected by the type of switching circuit. The speed-control circuit is also a factor in limiting the runout to a low level.

h) The temperature rise of the motor and the switching circuit reflects directly on the temperature rise of the adjacent parts in the drive unit. Thus the limit on heat dissipation can affect the choice of winding resistance and continuous current and hence the type of winding.

i) The disc mounting configuration on the motor hub determines which type of motor winding is most suitable. The under-slung and in-the-hub type motors make use of the available space in different ways by having different diameter to length ratios.

A different comparison can be made between the three major types of switching circuit based on the ratios of minimum and average torque to the maximum torque. These ratios reflect on the starting torque obtainable from the motor as well as on the percentage torque ripple. Table 1 shows this comparison and also gives the number of switching devices required in each circuit (3).

Table 1

The equations used in calculating the torque ratios in Table 1 are as follows:

In both equations (1) and (2), m is the number of commutations per revolution (for 4-pole design). Changing from unipolar to bipolar switching for any of the three major types may, however, require additional devices. For example in the 1-phase motor, a fifth device of power rating equal to the other four is required. Equations (1) and (2) assume that the e.m.f. wave forms of the motors are sinusoidal. This, however, is not always the case. The ratios given in the table are in fact lower than would be obtained from motors with trapezoidal e.m.f. wave forms and that is why the latter is desired in brushless motor designs.

No account is taken in Table 1 of the effect of the reluctance torque on the total torque, and the ratios given are solely due to the electromagnetic torque. The reluctance torque is the bi-directional torque produced by the interaction of the magnets with the lamination stack, and the electromagnetic torque results from the interaction of the current-carrying winding with the magnetic field.

The requirement for reluctance torque varies with the type of winding configuration. A 1-phase motor requires a high reluctance torque to facilitate starting at the commutation instants when the electromagnetic torque is zero. Whereas in a 3-phase or 4-phase motor it is desirable to minimize the reluctance torque and thus result in an increase in the minimum torque. In fact, a certain family of motors for 3-phase switching has no reluctance torque at all by virtue of its unique magnetic circuit (1).

Other types of switching circuit, not listed in Table 1, can be used with- 3-phase and 6-phase delta-connected windings. Delta connected windings generally have a low terminal resistance for equivalent values of back e.m.f. constant (V/k rpm or V/rad/sec).

The majority of computer peripheral applications use 3-phase star-connected motors with four poles. The low torque ripple can be achieved with a reasonable cost of electronics compared to, for example, a 6-phase motor. The switching circuit used with any motor can be a combination of bipolar start and unipolar run. This results in a high starting torque from the motor or a smaller motor is required to deliver the specified starting torque, whichever is more beneficial. Since the running condition of the motor is at very light load, the unipolar operation is adequate while reducing the number of switching devices being used and hence the total power dissipation.

Brushless Motor Performance:

The performance of a brushless dc motor can best be described by its speed/torque characteristics. These are results of tests performed on the motor between no-load and stall conditions. Fig. 3 shows typical speed/torque and current/torque characteristics of a brushless motor (4).

Fig. 3 Speed/Torque/Current Curves

The linear variation of speed with torque is typical of brushless dc motors as seen from the following equation:

Where:

n : speed in rpm

V : supply voltage in Volt

: e.m.f. constant in V/k rpm

: generated torque in oz.in

: motor terminal resistance in Ohm

: torque constant in oz.in/A

m : no. of power switching devices ON

: saturation voltage across each switching device in Volt

Equation (3) does not include the effect of high current on the speed/torque curve known as armature reaction. The magnetic flux produced by the winding weakens the main magnetic field causing the speed to drop faster at high values of current and torque. Equation (4) shows the linear relationship between torque and current, again ignoring the effect of armature reaction at high values of current.

A useful equation for the estimated required can be developed using equations (3) and (4) and Fig. 3. The unknown values in this equation (5) can usually be assumed based on prior experience or published data.

Where:

: no-load current in Amp

: generated torque at stall in oz.in

: no-load speed in rpm

An estimate of the required starting torque can be made using equation (6) at the desired acceleration time t.

Where:

: motor moment of inertia in oz.in.

: load moment of inertia in oz.in .

: angular acceleration in rad/

: friction torque of motor in oz.in

: load torque and stiction in oz.in

The angular acceleration can be calculated from the following equation:

Where t is acceleration time in seconds to reach speed n in rpm linearly.

Motor Performance in Complete Drive:

The primary requirement of a brushless dc motor in this application is to satisfy the parameters discussed above. These are: a minimum amount of starting torque at a specified starting current and a certain value of motor resistance. This should be done within the restrictions of space, switching method, configuration etc. as discussed earlier. However, to make a good drive, several or all of the following items also must be satisfied to varying degrees.

1. Acoustic performance:

Acoustic tests on the complete drive assembly are usually performed in a fully anechoic chamber and noise pressure in DB is measured at frequency bands with center frequencies typically at 500 Hz, 1 kHz and 2 kHz. The weighted average noise level in DB(A) is also measured to include the effect of all frequencies of the spectrum. Applications such as a half-height Winchester disc drive require typically under 36 DB in the frequency bands and under 40 DB(A) in the weighted average, both at a distance of 1 meter from the microphone.

Experience has shown that to achieve low levels of noise, the motor and the drive have to be analyzed simultaneously. The drive can resonate differently with different motors and excite a certain frequency which results in a high overall noise level. These frequencies can be electrical (due to linear or PWM switching and the number of commutations per revolution), magnetic, or simply mechanical in origin. The key to a quiet drive is to avoid exciting fundamental frequencies or their harmonics between, say, 500 and 5000 Hz.

2. Runouts:

The definition of runout in computer peripherals is the amount of variation within which a certain point on the recording surface comes with respect to a theoretical position in space as the motor rotates. This amount is usually measured in micro-inches and can be either repetitive, i.e. every revolution, or non-repetitive, i.e. happening at random. The maximum variations are specified in two axes, axial with the motor shaft and radial with the movement of the recording head. The runout that must be controlled to a low value (typically under 50 micro inches) is the radial non-repetitive which is affected by a number of design and operating parameters. The most obvious is the quality of bearings and the accuracy with which parts are machined and assembled. However, there are other factors of equal importance such as the design of the magnetic circuit, the choice of materials, the isolation between the electrical and mechanical parts, and the choice of switching circuit. The latter refers to the choice of speed-control method (linear VS. PWM) as well as whether unipolar or bipolar, 1-phase or 3 phase, etc.

3. Leakage Flux:

The amount of magnetic leakage flux outside the motor and in the recording-head zone is usually limited to a few Gauss. This level is affected by the design of the magnetic circuit but controlled by shielding and material choice. The magnetic shields used are placed between the motor and the recording areas and are made from high permeability materials. The shaft material used is preferably nonmagnetic stainless steel, however with certain types of seals, a magnetic shaft might be required. Comparatively speaking, the toroidal motors have a high level of leakage flux and in-the-hub motors have a low level, with conventional slotted under-slung motors in the middle.

4. Pulse Symmetry:

The pulses produced by the motor are usually from Hall devices and are used for both commutation and speed-control. The symmetry of these pulses is determined by the magnetic circuit of the motor and whether the main flux or the leakage flux is used to trigger these devices. Other factors are the dimensional tolerance of the magnets and their positional accuracy. The quality of the sensors and their hysteresis levels naturally play a part as well. Asymmetrical pulses result in erroneous commutation of the winding which could cause the starting torque to fall.

Conclusion:

The most commonly used spindle motors are brushless dc motors which are usually supplied in the form of complete assemblies ready for accepting magnetic and optical discs or tape cartridges. These motors have the advantage of long life, low noise, small runouts and simple construction. The choice of switching circuits, power devices and speed-control methods is wide enough to provide an optimum design for each application. In general the brushless motor is well suited to miniaturization and a variety of configurations.

References

[1] J. T. Jackson, Jr., "Wide air gap permanent magnet

motors", US Pat. no. 4,445,061 1984

[2] W. F. Lundin, "Performance of micro switch's new bipolar Hall effect sensor in brushless motors", Proc. of 7th International Motorcon Conf., Hanover, West Germany, April 22-24, 1985

[3] G. P. Gogue, Ph.D. Thesis, "Commutation of a brushless motor using power transistors", University of Aston, UK 1980

[4] Electro-Craft Corp. Handbook, "DC motors, speed controls, servo systems", 5th Ed. 1980