IEEE TRANSACTIONS ON MAGNETICS, VOL. 24, NO. 6, NOVEMBER 1988 2739 zyx MEASUREMENT OF HEADDISK SPACING WITH A LASER INTERFEROMETER L-Y. Zhu, K.F. Hallamasek, and D.B. Bogy Department of Mechanical Engineering University of California, Berkeley, CA 94720 Abstract A multichannel heterodyne laser interferometer was built to measure the head-disk spacing in magnetic disk drives. The main advantage of the present method compared to white light inter- ferometry and capacitive measurements is that neither the disk nor the slider need be transparent or conductive. By the use of two acousto-optic modulators, a HeNe laser beam is separated into a reference and a signal beam whose fre- quencies differ by 2 zyxwvutsrqponm MHz. The signal beam illuminates an area containing portions of the slider and disk. The reflected light is combined with the reference beam and a lens system focuses both beams onto two image planes, creating 2 MHz interference pat- terns. Five fiber optic probes are positioned at selected points on the image plane. Each probe senses the 2 MHz interference that is phase modulated by the displacement of the corresponding point on the disk or slider. By comparing the phase difference from a pair of fibers, the relative displacement between any two points is obtained. Experimental results are presented and we conclude that the present system is capable of the desired head-disk spacing meas- urement with minimum modification of the slider. It has a resolu- tion of 2.5 nm and a bandwidth of DC to about 40 kHz. Introduction Direct, simultaneous measurement of head-disk spacing and slider motions such as pitch, roll and translation normal to the disk surface is highly desirable in the development of magnetic disk drives and has been a great challange in metrology. Many methods have been reported with various degrees of success. For example, the capacitance probe, Lin [l], requires a special slider with conductive plates embedded at four comers; the white light interferometry, Lin and Sullivan [2], requires either a transparent disk or a transparent slider. Both of these methods measure the slider motion relative to the disk surface, not to an inertial frame. The laser Doppler vibrometer (LDV), Miu et al. [3], can measure the absolute motion of the slider, but not the relative displacement between the slider and disk. The multichannel laser interferometer (MCLI) can measure the head-disk spacing and slider motions with reference to both the disk surface and an inertial frame. The only special requirement of the slider/disk assembly is that the back of the slider should be specularly reflective. Our system is similar to, but simpler than, that described in Davidson zyxwvutsrq [4]. We eliminated the laser beam expander and the reference mirror; also we utilized the beam splitting capability of an acousto-optic modu- lator (AOM1) to simplify the beam splitting and recombination scheme. In this paper, we describe our MCLI, including the optics and electronics. Experimental results on head-disk spacing during the take-off, flying and landing processes are presented and discussed. Results showing the multichannel capability of the apparatus are also presented. ExDerimentd ADDaratUS Figure 1 shows the schematic of the optics system. Ignoring temporarily the second acousto-optic mudulator AOM2, the lens L, and the optical fibers FIBERl and FIBER2, we can think of the system as a variation of the LDV (see [3]). A HeNe laser emits a beam that is polarized in the direction parallel to the optics table (P-wave). The beam passes AOMl which is operated at 40 MHz, and splits into a signal and a reference beam. The signal beam is directed by mirrors M1 and M2 through a polarizing beam splitter PBSl to the target. The quarter-wave retarder plate rotates the polarization of the signal beam zyxwvutsrqpon so that after coming back to PBSl, the signal beam becomes an S-wave and is deflected by PBSl to a half-wave retarder plate. The reference beam, which has the 40 MHz modulation, is directed by mirrors M1, M3 and M4 onto the half-wave plate without reaching the target. The signal and refer- ence beams are made colinear at the half-wave plate. They do not interfere because they are orthogonally polarized. To make them interfere, the half-wave plate rotates the polarizations of both beams by 45' so that each of the two beams has now half P- and half S-components. The second polarizing beam splitter PBS2 directs S-components from both beams to an image plane IPl, and P-components to another image plane IP2. If we place one photo- diode at IP1 and another at IP2, both will sense the 40 MHz interference between the signal and reference beams. When the target moves in the direction of the signal beam, the optical path length of this beam will vary, which causes phase and frequency shift of the 40 MHz interference pattern. A LDV detects the absolute velocity by frequency demodulation. To detect the target displacement, phase demodulation is needed. Since phase demodulation of a high frequency carrier is difficult, we create a 2 MHz instead of 40 MHz carrier. This is accomplished by upshifting the signal beam 38 MHz with AOM2. To detect separately the motion of different points within the tar- get, we use a lens system L to form real images of the target at IP1 and IP2. These images are identical but 180 degrees out of phase. Either or both of the images can be used. By inserting fiber optic probes at various points of the image(s), we can meas- ure the motion of corresponding points on the target. We have five (5) photodiodes and three (3) phase demodula- tors in our MCLI. But many more optical fibers can be positioned on the real image(s). Only the ones "looking at" the points of interest need be connected to the photodiodes. Each phase demo- dulator measures the relative motion between two target points and requires outputs from two photodiodes. Through the use of a multiplexer, the output of each photodiode can be connected to more than one phase demodulator. For brevity only two optical fibers and one phase demodulator is shown in Fig. 1. FIBERl and FIBER2 are connected to separate silicon avalanche photodiodes APDl and APD2. Due to symmetry only the path from FIBERl will be described. The output of APDl is a 2 MHz sine wave, which is sent to AGC, an amplifier with automatic gain control. The signal is then sent to a limiting amplifier and becomes a square wave. Similarly light from FIBER2 is converted into a square wave. These two square waves are sent to a phase comparator, whose output voltage is directly proportional to the phase difference between the two input square waves. For small amplitude motions an exclusive-or (XOR) phase comparator is adequate. But for large amplitude motion (greater than 150 nm) a more sophisticated phase comparator is needed. Otherwise the input-output relation is periodic and not monotonic. The output of the phase demodulator is digitized and displayed by a DATA6000 digital oscilloscope and subsequently transferred to an IBM PC for storage and processing. Measurements were made using a 5 1/4 inch untextured thin film disk with a mini-Winchester slider. Since the back of the slider is usually not mirror-like, it must either be polished or coated. One simple way to create a mirror-like surface is to attach a small fleck of a thin silicon chip on the back of the slider. FIRES; FIBER2 i I Figure 1 Schematic of optics and data acquisition systems. zyxw 0018-9464/88/1100-2739$01.00@1988 IEEE