Introduction - 應用順滑模態控制於微型音圈馬達摩擦力補償之研究

1.1. Research background and recent development

Compact camera module (CCM) has embedded in Mobile phone to provide camera function for years. The pixel size of CCM’s image sensor has increased from 0.3M in the early stages to 5M presently. User’s demand for better image quality of those cameras with image pixel size larger than 3M triggers CCM makers to build auto-focus (AF) function into high-end CCM.

AF function is established by an actuator, which carries lens, plus an image sharpness identifier (ISI). The actuator carries lens to different positions in the optical stroke and stop at the position with best image sharpness according to ISI’s instruction. Camera’s focus range can be from 10cm to infinity with the AF function. The opposite of AF is fixed-focus (FF).

The focus range of FF CCM is specified for macro (10cm to 40cm), or infinity (90cm to ∞). It means the image quality is poor at macro if the FF CCM is specified for infinity. Under this condition, the high-pixel image sensor will not achieve the expected performance. That is why AF is strongly recommended for high-end CCM today.

Due to pursuing thin profile for mobile phone, CCM is required to be small. This limits the actuator solutions for the AF CCM. MEMS, piezo and mini voice-coil motor (VCM) are the available solutions in the market. The strengths of MEMS solutions are more precise positioning, better repeatability and lower power consumption. The weaknesses of MEMS solutions are poor reliability and higher cost. Piezo solutions also have the strength of low power consumption, but they have concerns about reliability and cost. VCM is the most popular solution for AF CCM due to its compact size, cost competitiveness and satisfactory performance. The size of VCM is down to 8.58.50.46mm³ with 0.35mm full stroke for now and will become smaller in the future.

VCM solutions can be separated to two groups. One called tradition VCM, which builds in springs and adopts open-loop control. Its positioning is settled when the Lorenz force

yielded by the current of coils and the spring-force yielded by deformation are balanced. The other is closed-loop control VCM, which builds in position sensor and requires closed-loop controller. One weakness of the traditional VCM is that its spring would malfunction in continuous video mode. In this mode, the VCM may continuously work with max spring deformation. In the meantime, coils are heated up by the continuous driving current. And the heat is transferred to the spring because spring and coils are connected. The constant deformation plus the heat makes the spring not be able to restore to original position when the driving current is removed. On the contrary, closed-loop VCM will not have this problem because it does not embed with spring. Closed-loop VCM also provides better performance in transient response and positioning.

One kind of the closed-loop VCMs, as shown in Figure 1, supports good tilt angle performance by implementing guide pins ((b) of Figure 1), which directly contacts with actuator and results in nonlinear friction force ((g) of Figure 1). A well-known solution for decreasing the effect of nonlinear friction force is applying lubrication oil on the guide pins.

But the optical system would fail if it is polluted by oil. So that taking care of the lubrication oil is an extra burden for manufacturing CCM. Furthermore, the effect of the nonlinear friction force will become more significant when the size of VCM is shrunk smaller (which is an inevitable trend). [13] even achieves power saving purpose by designing the static friction force to hold actuator at arbitrary position with zero supplied current. This characteristic is welcome by portable devices, such as CCM, because of power saving consideration. But it does add challenges on controller design because the effect of nonlinear friction is increased.

So that developing a cost-effective friction compensation scheme with satisfactory performance is crucial for the future success of the mini VCMs embedded with guide pins.

Friction compensation has been studied for years in other fields of application, but it is new to study friction compensation scheme on the mini VCM applied to CCM. A comprehensive survey of friction compensation schemes is presented in [1]. In which, type A) and B) solutions are suitable for cost-sensitive applications because of its limited calculation burden. [11] proposed a nonlinear proportional controller with bang-bang force in specified region to compensate the sticking force, where [12] demonstrated a look-up table position controller, that has higher gain when the position error is smaller and lower gain when the

following control (AMFC) to overcome the loading variation, and [10] considered an anti-windup PI controller, incorporated with the disturbance observer, to control the VCM.

Figure 1 : Guide pins directly contacted with actuator result in significant nonlinear friction force.

1.2. Research motivation, objective and approach

The worldwide sales volume of mobile phones is over one billion in 2008, in which the high-end models all build in AF CCM. If closed-loop mini VCM solution can be more competitive in performance and cost, then its market share can be increased in AF CCM. Or even be able to convince middle-end (main stream) mobile phones to adopt AF solutions.

Furthermore increase the market share of the mobile phone embedded with AF CCM. Then the whole vertical solution providers, such as VCM manufacturers, controller chip suppliers and AF CCM makers, will benefit from it.

We intend to improve the competitiveness of closed-loop mini VCM by developing a simple and effective control scheme to compensate the nonlinear friction between actuator and guide pins. Our goal is to remove the lubrication oil from the VCM. This can save the

(a)

(b)

(e)

(f)

Friction

Velocity

(g) (a)

(b)

(e)

(f)

Friction

Velocity Friction

Velocity

(g)

cost of materials and manufacturing process. This also can improve the reliability of the VCM because lubrication oil is sensitive to temperature.

In this thesis, a sliding mode controller is developed to compensate the mismatched nonlinear friction force of the VCM, which does not apply lubrication oil on the guide pins.

The controller only uses the information of static friction force of the VCM. With pole placement of the sliding mode state equation, steady state position error is predictable and is able to be controlled to be arbitrarily small. To the best of our knowledge, this kind of approach has not yet been studied and applied to the mini VCM system. The simulation results prove that the developed algorithm has better performance than classic controller, and the experiment results demonstrate the stick-slip oscillation is avoided. We also present an experimental environment (Figure 2), including PC motherboard, in which control algorithm is executed, and FPGA board, in which digital driver (full bridge) with synchronous current sampling [4] is implemented. The FPGA board connects to the motherboard through parallel-ATA (Pparallel-ATA) cable. Position feedback, current feedback and controller output traffic on the cable with standard PATA PIO protocol. This experimental platform provides the best cost/performance ratio comparing to others, such as ARM-base development board. Its high calculation power is especially suitable for advanced algorithm study. With this experimental environment, a more effective design flow for developing controller ASIC is also presented.

1.3. Thesis Organization

This thesis is organized as follows. In section II, physical VCM system modeling is described. In section III, a sliding mode control law for friction compensation is proposed, while simulation results are demonstrated in section IV. In section V, experimental results are provided with the brief implementation of PC-based experimental environment and design flow of controller ASIC. Finally, conclusions are presented in section VI.

Motherboard (Controller)

FPGA Board

ATA bus X86

CPU ATA

Full VCM Bridge

IIR Filter PWM

IIR Filter

i d FPGA

A_f