DEVELOPMENT OF A SILENT BRUSHLESS DC MOTOR DRIVE S. Camilleri, D. Patterson & H. Pullen NT Centre for Energy Research, Australian CRC for Renewable Energy Northern Territory University Darwin, N.T. 0909 Email : Steven.Camilleri@darwin.ntu.edu.au Abstract Brushless DC motors are known for their versatility and efficiency. Unfortunately the cheap and common method of using trapezoidal current control is not suitable for all applications because of the distracting acoustic noise made during commutation, known as 'ticking'. This paper outlines some methods of controlling the brushless motor to produce much less acoustic noise, including using a DSP for producing commutation period current wave-shaping and a sinusoidal Voltage/Frequency amplitude control. 1. INTRODUCTION & BACKGROUND The target application for this system is a high efficiency ceiling fan. This fan uses an unconventional blade design with extremely high pitch that produces standard levels of airflow at only half the rotational speed, compared to a conventional flat blade design. The result of this is an extremely low 'wind' noise, due to the improved aerodynamics. A picture of the fan blades is shown in Figure 1. the low noise blade design. In addition, the motor controller design would have to be as cheap as possible to make the final product cost comparable with existing technology. In addition to this application, another motor was developed to drive a standard steel bladed ceiling fan. The control methodology would be the same for this application, but the speed would be twice as much as the low speed research fan. Again, low acoustic noise operation is required. When the fan was first tested with a standard trapezoidal current brushless motor controller, it produced noise in a wide spectrum. The loudest part of the noise was naturally at the frequency of commutation at around 100 Hz in a 16 pole machine, running at 125 RPM. Figure 1 : Reseach Fan blades To further improve the fan system, it was decided to use a high efficiency brushless DC motor with a power electronic drive instead of the standard split phase induction motor. This would improve the system efficiency significantly, provided the brushless motor could be driven to produce a low noise output to match To reduce the noise, a simple circuit was devised that would anticipate an impending commutation period (via mechanically shifted hall effect sensors) and control the motor current to zero, reverting it to the original value once that period was over. In effect, this circuit would act as a 'soft commutation' or 'deticking' circuit. Some waveshaping was required to reduce and increase the current in a manner that would not produce extra acoustic noise, for which the accurate control of a TMS320C30 DSP was required. A block diagram of the system is shown in Figure 2.
Figure 2 : Block Diagram of Deticking Circuit In practice, the soft commutation circuit did reduce the noise but not to the level required. Spectral analysis was performed using a quality microphone and Hewlett-Packard 3561A signal analyser, in a standard audible range (20 Hz to 20 khz). As can be seen from Figure 3, the deticking circuit was only effective on two areas, 250 Hz and the range 900 Hz to 1300 Hz. The largest part of the noise, below 150Hz, was nearly unaffected. Overall, the deticking circuit did not reduce the noise produced by the fan to an acceptable level in fact at the high frequency range above 1300 Hz, the noise was increased. 0-10 -20-30 -40-50 -60-70 -80-90 -100 0 500 1000 1500 2000 2500 Frequency(Hz) Figure 3 : Fan Motor Noise Spectrum Deticking Circuit Ticking Deticked Various wave shapes were used for increasing and decreasing the current at the commutation period, all accurately controlled using the DSP. Figure 3 shows the noise reduction results for the best shape (among many tested) which was in essence an extract of a sinusoidal curve between 90 o and 270 o. Of all the wave shapes tried, this shape produced the best results. 2. THEORY OF VOLTAGE/FREQUENCY CONTROL After abandoning the previous control method, a totally new approach was desired. The trapezoidal method of current control requires rapid changes in current in the machine, which in turn produces the acoustic noise. Control of the current in a sinusoidal manner appears to be a straightforward way to reduce the rapid di/dt's in the motor and correspondingly the acoustic noise produced. Sinusoidal control is difficult and complex to do using a closed loop current control system (usually involving the use of expensive position encoders) and since the system was designed to be cheap, an open loop strategy was decided upon, namely Voltage/Frequency (V/F) control. This method is most common in induction machines due to the simplicity of the control system and since it allows variable speed control of the machine. In basic terms, V/F control uses a microcontroller or similar digital control device to generate a sinusoidal Pulse Width Modulated (PWM) drive for which the fundamental amplitude varies proportionally to the fundamental frequency. For an induction machine, this means that the microcontroller starts with marginally above zero frequency and low sinewave amplitude, increasing both at a rate that the mechanical system can handle to accelerate them. The amplitude of the sinewave is directly proportional to the current through the windings of the machine at any one frequency, otherwise the current is inversely proportional to the frequency for a fixed amplitude. The main difference between this drive developed for a brushless machine and the one used for an induction machine is the use of 'slip'. An induction machine provides torque in proportion to the slip factor, which is related to the ratio between the rotating magnetic field on the stator and the mechanical speed of the rotor. As the rotor slows, slip increases and more torque is provided. With a brushless machine, the rotor has permanent magnets attached and hence if a pole slip occurs, negative torque will be generated. This causes the machine to stall, and after a stall has occured the machine will not restart using open loop control but simply stop and 'vibrate' as the rotating magnetic field repeatedly passes over the permanent magnets, generating no useful torque. It is therefore essential to ensure that the permanent magnet rotor stays completely synchronous with the rotating magnetic field on the stator. Fortunately for a ceiling fan system the load & speed profile is known and can be stored in a digital control device, so stalls under normal operation can be easily prevented.
It is still neccesary to include a very basic form of closed loop control for situations that are unexpected (i.e. something comes into physical contact with the rotating fan, causing it to stall) due to the fact that in a stall condition, if a high ampltiude sinewave is applied to the fan motor then the currents through the motor may be extremely high and may damage the motor or controller. an isolated power stage for driving the motor from a 50V DC bus. The power section constructed for development of the prototype uses a standard six switch bridge comprised of SGS-Thomson STN3E06L MOSFETs, which are small (supplied in an SOT-223 surface mountable packages), efficient and cheap. A simple diagram of the bridge is shown in Figure 4. To this end, a single hall effect sensor can be used for a speed reference signal. Once the speed has been measured by the digital controlling device, it can be used to look up the appropriate section in the V/F table, guaranteeing that the currents will never be high for very long. In addition, using this methodology allows stall recovery. 3. VOLTAGE/FREQUENCY CONTROLLER DESIGN Listed below are the specifications of the motors used in the development of this control system. RESEARCH FAN Figure 4 : Basic diagram of Power Stage Design of the power stage requires a working knowledge of the power range of the system, in this case the prototype system characteristics are : RESEARCH FAN Motor (1) : Three Phase Permanent Magnet Brushless 16 Pole - Prototype Application speed 140 RPM ( 18 Hz electrical frequency) 16.5 Ω phase resistance average 53 mh phase inductance average Full speed : Current at full speed : Voltage at full speed : Power at full speed : 135 RPM 520 ma from DC supply 490 ma RMS /φ 52 V from DC supply (equates to 18.4 V RMS / φ) 27 W STANDARD BLADE FAN STANDARD BLADE FAN Motor (2) : Three Phase Permanent Magnet Brushless 16 Pole - Prototype Application speed 280 RPM ( 36 Hz electrical frequency) 3.7 Ω phase resistance average 12.8 mh phase inductance average Full speed : Current at full speed : Voltage at full speed : Power at full speed : 260 RPM 550 ma from DC supply 520 ma RMS /φ 42 V from DC supply (equates to 14.85V RMS / φ) 23 W Note that both of these motors are still in the development stage, hence their parameters are still subject to future improvement. Both motors will use the same control system, running with a different V/F characteristic and maximum speed. Electronic requirements for a V/F converter of this nature can be divided into two modules - A digital control device to generate the PWM waveforms and The MOSFET was selected on the basis of cost and size, rather than efficiency. Since either motor will not draw much more than 0.5 A, the resistive loss in the MOSFET will be negligible. Using an equation for MOSFET conduction loss P C : P C = I DS 2 R DS ON (1)
Where I DS is the Drain-Source current, and R DS is the Drain-Source resistance. Therefore, for this calculation we need to determine MOSFET current I DS(RMS), which will work out to half the RMS motor phase current (by basic KCL). This works out to 0.26 A RMS through each MOSFET on average. For the 100 mω STN3E06L, using equation (1), this comes out at 6.8 mw of conduction loss in each MOSFET, a total of 40.8 mw through each leg in the bridge. Switching and gate losses are found to be about equal for this device, switching at a rate of 23.4 khz. The total loss in the power section is therefore negligible and can be assumed to be a maximum of about 100 mw, giving a system efficiency above 99.6% at full output. During testing, the external case temperature of the MOSFETs could not be measured above ambient, thus they did not require heatsinking. The digital control section is more complex. There are many devices on the market that are ideally suited to the task of motor control, but in this case there are the heavy restrictions of cost and physical size that reduce the options significantly. The choice is between a cheap DSP and a cheap microcontroller - and the microcontroller wins due to it's simplicity and reduced size. In any case, there are currently no DSP systems available on the market that are as cheap and easy to use as a well chosen microcontroller. The prototype system was initially based on a microcontroller that was readily available, the Siemens 80C537. This microcontroller itself is not ideally suited to the task, but shares a common language with many other more suitable Siemens microcontrollers such as the SAB-C504 Motor Controller. Programs have been developed with the 80C537 that perform all of the requirements mentioned previously and there have been no major problems with the system. A block diagram of the prototype drive is shown in Figure 5. Figure 5 : Block Diagram of Prototype V/F Controller 4. CONCLUSION While some of the initial methods tried such as deticking circuits with DSP control did not work to the standard required, the prototype V/F controller developed for the Brushless DC machine used in both the research and conventional ceiling fans has been tested to the levels required by the respective systems and found to work very well. Construction of a standard six switch block was simplified by the known parameters of the system and the low motor currents, also allowing a very efficient power stage. Selection of a digital control device for attaching to this power stage is not a simple procedure as the market for small and cheap motor control chips is extremely limited, however there are several that suit the task adequately. The Siemens range of microcontrollers has been tested using this control scheme and performs well. 5. REFERENCES [1] John G. Kassakian, Martin F. Schlecht & George C. Verghese, Principles of Power Electronics, Addison-Wesley Reading, Massachusetts 1992. [2]Werner Leonhard, Control of Electrical Drives, 2 nd Ed. Springer-Verlag Berlin Heidelberg 1996. [3]T. Kenjo & S. Nagamori, Permanent Magnet and Brushless DC Motors, Clarendon Press, Oxford 1985. [4]M.F. Rahman, L. Zhong and K.W. Lim, A Comparison of Two High Performance, Wide Speed Range Drive Techniques for Interior Magnet Motors, 1998 International Conference on Power Electronics Drives and Energy Systems for Industrial Growth, Volume I, pp. 276-281, December 1998.
[5]K. Schmidt & D. J. Patterson, Performance Results for a High Efficiency Ceiling Fan and Comparisons with Conventional internet at Fans, 1999 World Renewable Energy Congress. Available at : http://ee.ntu.edu.au/ntcer/pdf_docs/