WO1996015529A1

WO1996015529A1 - System and method for accurate arcuate scan head positioning

Info

Publication number: WO1996015529A1
Application number: PCT/US1995/014577
Authority: WO
Inventors: Martyn A. Lewis; Paul Stavish
Original assignee: Seagate Technology, Inc.
Priority date: 1994-11-10
Filing date: 1995-11-09
Publication date: 1996-05-23

Abstract

A system and method is disclosed wherein the position of a recording head is kept in proper alignment with the centerline of the track of a recording medium despite any tape wander. Further, the speed in which the tape propagates during both read and write modes is maintained. These functions are performed using the servo information written into the tracks of the recording medium. Tape propagation speed is controlled by adding the servo signal together. Head positioning is controlled by taking the difference between the servo signals. A method using Walsh Transforms is disclosed for normalizing the amplitude of the servo information used to accurately control head positioning and tape propagation speed.

Description

SYSTEM AND METHOD FOR ACCURATE ARCUATE

SCAN HEAD POSITIONING

INVENTORS: Martyn A. Lewis

Paul Stavish

CROSS REFERENCE TO RELATED APPLICATIONS

The present application is related to the following U.S. Patent Applications, all filed concurrently with or prior to the present application:

1. "Arcuate Scan Tape Drive," by John M.

Rothenberg, Joseph Lin, Robert H. Peirce, Richard Milo and Michael Andrews, Serial No. 08/113,996, filed August 30, 1993.

2. "Arcuate Scan Read/Write Assembly," by Gary Nelson and Stephen J. Crompton, Serial No. 08/337,255, filed November 10, 1994.

3. "Economical Wide Range Speed Control System", inventor Martyn A. Lewis, Serial No. 08/337,803, filed November 14, 1994.

4. "Method and Apparatus To Maximize The Top Speed Of Brushless DC Motors", inventor Martyn A. Lewis, Serial No. 08/336,981, filed November 10, 1994.

The above applications are all assigned to the assignee of the present invention and are all expressly incorporated herein by reference. Field of the Invention

The present invention relates to tape drives in general and, in particular, to a system for accurately positioning a transducer head over a recording medium in an arcuate scan tape drive system. BACKGROUND OF THE INVENTION

Presently there exists a need to be able to

accurately position recording heads over a tape

notwithstanding track position and pitch errors and further to be able to control tape propagation speed. When recording information onto a recording medium, i.e. a tape, track position errors may occur. Track position errors are the result of the various sources of errors and disturbances present within an arcuate scan tape drive system, and result in the recording heads being incorrectly aligned with the track centerline of the tape. Some of the sources of track position errors include errors at the encoder frequency and at harmonics of the encoder frequency, errors due to encoder flutter, errors due to capstan motor torque ripple and errors due to cartridge flutter. The total position error of the arcuate scan system is defined as the sum of the

aforementioned errors.

Along with track position errors, track pitch errors can also occur while recording information onto a tape. The track position errors discussed above will give rise to track pitch errors with a magnitude being related to the ratio of the track pitch error frequency to the scanning frequency of the recording head. Track pitch errors caused by errors at the encoder frequency and at harmonics of the encoder frequency are worse than the track position errors caused by the same sources and result in track squeezing which can reduce the available off track margin on the tape. Track pitch errors are also caused by encoder flutter, torque ripple and

cartridge flutter, with total track pitch error being defined as the sum of the aforementioned errors.

Track position errors may also occur when reading information from a tape. There are two categories of tracking errors present while reading information from a tape. The first category is due to the track position errors "frozen in" the tape while recording. The second category is due to encoder errors, torque ripple, and cartridge flutter present while reading information from a tape. These position errors can offset the head from the target track of the tape by several tracks in either direction thereby causing the recording head to read incorrect information.

When reading information from a tape, tracking errors due to errors at the encoder frequency and at harmonics of the tape itself can occur. Tracking errors occurring during reading may also caused by encoder flutter, torque ripple and cartridge flutter. Tracking errors make it extremely difficult for an arcuate scan tape drive system to align the recording heads with the data tracks recorded on the tape. Also, due to the misalignment of the recording heads with respect to the tape, the tape propagation speed may need to be increased or decreased to ensure proper alignment of the recording heads with respect to the tape.

SUMMARY OF THE INVENTION

The present invention eliminates, or greatly reduces the problem of track pitch and position errors by

employing a system which controls the center of rotation of the head assembly comprised of write and read heads so that the recording heads are aligned with the track centerline despite any lateral (i.e., the direction perpendicular to the direction of intended tape travel) tape wander. This is accomplished by having some of the read heads read at the beginning and the end of each scan the servo bursts recorded on the tape having a

microcontroller process the burst amplitudes to obtain upper and lower Position Error Signals (PES_A and PES_B respectively), subtracting PES_A from PES_B and comparing the result PES_tilt to 0, thereby producing a tilt error signal. This error signal is then transmitted to an actuator (i.e. a voice coil motor) which controls the tilt of the head, thereby greatly reducing, or

eliminating head positioning errors relative to the tape. The invention further uses the pre-recorded upper and lower servo burst signals to control the speed at which the tape propagates relative to the head by summing the two position error signals thereby generating a third signal which is transmitted to the capstan motor which controls tape propagation speed.

An advantage of the present invention is the ability to control the center of rotation of the scanner assembly thus keeping the recording heads aligned with the track centerline, despite tape wander.

Another advantage of the present invention is that head tilt motions produce no error signal into the tape speed servo.

Yet another advantage of the present invention is that tape speed variations introduce no error signal into the tilt servo.

Still a further advantage of the present invention is the ability to produce a superior estimate of

normalized amplitudes of the servo burst signals despite any variations between tape samples and head samples.

Yet still a further advantage of the present

invention is that it can be implemented with few low cost parts.

Yet still another advantage of the present invention is that it can be implemented with low cost.

BRIEF DESCRIPTION OF THE DRAWINGS

Additional advantages, features and characteristics of the present invention will become more clearly

understood from the following description of the

embodiments when taken in conjunction with the accompanying drawings, wherein like reference numerals denote like parts and wherein:

Figure 1 is a block diagram of the electro

mechanical system of the present invention.

Figure 2 is a schematic diagram illustrating the servo burst pattern used in conjunction with the present invention.

Figure 3 is a schematic diagram depicting head position and servo information on a magnetic recording medium in accordance with the present invention.

Figure 4 is a block diagram of the tilt positioning servo mechanism of the present invention. Figure 5 is a schematic diagram of the analog tilt servo of the present invention.

Figure 6 illustrates the steps used to calculate the proper amplitude of the position error signal of the present invention.

Figure 7 illustrates the signal patterns resulting from servo bursts being read by transducer heads.

Figure 8 illustrates the two values used during normalized amplitude routine.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

A block diagram of the arcuate scan tape drive system 100 of the present invention is shown in Figure 1. A microcontroller 200, in the most preferred embodiment of the present invention an 80C196 microcontroller manufactured by INTEL, receives commands from a host system via line 301. The host (not shown) sends commands over line 301 to the microcontroller 200 to control the writing of information to a tape 11 and the reading of information from the tape 11 by transducer heads 13. The tape propagation speed is controlled by a capstan motor 210 in combination with an idler wheel 208 and a data cartridge drive puck 206 to propagate the tape 11 in direction P. The capstan motor 210 propagates the tape 11 at substantially 0.5734 ips when the transducer heads 13 are writing information onto the tape 11 and

substantially 0.5734 ips when information is being read from the tape 11. However, speeds up to substantially 72 ips while in a search mode, during which the information on the tape is also read, can be achieved. Servo

information that is written onto the tape 11 is read by two of the transducer heads 13 and transmitted to the microcontroller 200 via preamplifiers 204 of the ASHA assembly 202 to control head positioning and tape

propagation speed, respectively. The ASHA assembly 202 is comprised of the ASHA motor 220, read heads 13, the preamplifiers and write drivers 204, the write heads (not shown) which are disposed to bisect the angles shown between the read heads 13, the bicell transducer 222 and the voice coil motor (VCM) 83, and is coupled to the microcontroller 200 via line 201 and via the DAC's 50. In the preferred embodiment, the scanning frequency of the ASHA assembly 202 is 14334 rpm. The microcontroller 200 contains a mathematical model of the overall electro mechanical system used in the particular application and performs all of the computations described herein below. Also coupled to the microcontroller 200 are DACS 50 for transmitting the analog representation of the digital signals generated by the microcontroller 200 to the various components which comprise the present invention. A Magneto Resistive (MR) encoder 212 that transmits the sine and cosine waveforms representing the capstan motor shaft position is coupled to the microcontroller 200 via lines 213 and 214, respectively. A bicell processor 52 is also coupled to the microcontroller 200 via line 44. The control of the head positioning and tape propagation speed by the microcontroller 200 will be discussed in greater detail below. The servo information that is written onto the tape 11 is shown in greater detail in Figure 2. The data tracks in Figure 2 are shown to be straight lines.

However, in practice, the data tracks are curved in shape. The servo information is written in the form of constant frequency bursts near the top and bottom edges of the tape 11. For example, servo burst 400 has a frequency F1 and a width of approximately two data track pitches, across data tracks 402 and 404, respectively. Servo burst 403 has a frequency F2 that is different from frequency F1 of servo burst 400. In the preferred embodiment, the frequency in burst F1 is one half the frequency in burst F2. The width of the data tracks, numbered 1 through 8, respectively, is normally 600 microinches.

Figure 2 also shows 4 rectangular blocks numbered 1-4, respectively. These blocks illustrate the possible positions of a single positive azimuth transducer head as it scans the tape 11 from the top edge to the bottom edge in direction P. One pair of the transducer heads 13 are positive azimuth heads. The second pair of transducer heads 13 are negative azimuth heads. Only the positive azimuth heads can read the servo bursts. The approximate amplitude of the burst signals, normalized to a full amplitude of 100%, from the particular positive azimuth head when positioned relative to the track centerline 16 of each data track is represented by the table below:

When the transducer head is in position 1 it encompasses all of the data track 404 (track 2) and some of data tracks 402 (track 1) and 406 (track 3), respectively, thereby reading a F1 400 amplitude of 71% and an F2 403 amplitude of 29%. In this position, however, the

positive azimuth read head is unable to read the negative azimuth data on track 2. Therefore, this is not a valid alignment for the head. Similarly, when the head is at position 2 it reads a F1 400 amplitude of 29% and an F2 403 amplitude of 71%. In this case, the positive azimuth read head can read the positive azimuth data, so this is a valid alignment. When aligned in this way, the

negative azimuth heads are disposed to read the negative azimuth data when they scan the tape. The negative azimuth read heads are never involved in reading servo information. At position 3, the head 13 reads an F1 400 amplitude of 29% and an F2 403 amplitude of 71%. Also, when the transducer head 13 is at position 4, it reads an F1 400 amplitude of 71% and an F2 403 amplitude of 29%. All four head positions illustrated in Figure 2 are satisfactory for recovering servo burst information.

However, only positions 2 and 4 are satisfactory to recover data written on the tape 11 because the read heads are positive azimuth heads and can only read positive azimuth data. If the head is located at other than the positions shown, the relative percentages of the amplitudes change in rough proportion to the offset of the head from the track centerline 16.

Figure 3 shows the tape 11, propagating in a

direction P across a transducer head 13 with the servo burst pattern as shown in Figure 2 written on lines 12 and 14, respectively of the tape 11. The track

centerline of the tape 11 is represented by the dashed line 16. Figure 3 also shows two arcs (one solid arc with a center of axis of rotation BB and a dashed arc with a center of axis of rotation AA) crossing the tape 11. The solid arc represents the locus traced out by the transducer head 13 when it is properly aligned with the track centerline 16 of the tape 11. The dashed arc represents the locus traced out by the transducer head 13 when it is misaligned with the track centerline 16 of the tape 11.

A brief description outlining the general operation of the system to control head alignment and propagation speed will be discussed with reference to Figures 1-6. When the tape 11 is moving in a direction P across the transducer heads 13, the positive azimuth heads 13 scans the F1 and F2 bursts near the upper edge 12 and near the lower edge 14 of the tape on the locus (represented by the solid line) from position 17 to position 18. The values read by the positive azimuth head 13 during the scan are transmitted through the pre-amplifier circuitry 204 and the burst demodulator 38 to the microcontroller 200 for determining whether proper head alignment and propagation speed are being maintained. The burst demodulator includes frequency selective filters tuned to frequencies F1 and F2 so that the burst amplitudes in F1 and F2 are independently measured. From the servo bursts F1 and F2 read by the positive azimuth heads 13, position error signals PES_A and PES_B are generated that are used to adjust to the propagation speed of the capstan motor 210 and the tilt of the transducer heads 13 relative to the track centerline 16 of the tape 11. The generation of the PES_A and PES_B values is discussed in greater detail below. The microcontroller 200 takes the difference between the generated position signals PES_A and PES_B to determine if the transducer heads 13 are properly aligned with the track centerline 16 of the tape 11. (This is represented by the solid line with a center of axis of rotation BB). If the difference is zero, no tilt

alignment needs to be done. If the difference is not zero, representing that the transducer heads 13 are misaligned with the track centerline 16, (shown by the dashed line with a center of axis of rotation AA) a control signal derived by the microcontroller from the magnitude and sign the difference is transmitted to the VCM 83 to adjust the head tilt to properly align the transducer head 13 to the track centerline 16 of the tape 11.

The adjustment of the propagation speed of the capstan motor 210 which is coupled to the idler wheel 208 and the cartridge drive puck 206 to propagate the tape 11 is performed in a similar fashion as described above with respect to head alignment except that the two generated position signals PES_A and PES_B are arranged. The

resulting signal is processed by the microcontroller and transmitted to the capstan motor 210 to either increase or decrease its speed. The commutation of the capstan motor 210 is described in co-pending applications that are assigned to the assignee of the present invention entitled "Method and Apparatus To Maximize The Top Speed Of Brushless DC Motors," by Martyn A. Lewis, Serial No.

08/336,981, filed November 10, 1994, and "Economical Wide Range Speed Control System," by Martyn A. Lewis, Serial No. 08/337,803, filed November 14, 1994 and are

incorporated herein by reference.

Figure 4 shows a block diagram of the head tilt positioning system 10 of the present invention. In the preferred embodiment, the head tilt positioning system 10 has two sections, a first section that receives

information from the transducer heads 13 and transmits the information to the microcontroller 200 for

processing, and a second section that transmits the correction signals generated by the microcontroller 200 to the actuator of the ASHA assembly 202 via the

digitally controlled analog closed loop servo of Figure 5 which controls head positioning relative to the tape 11. The servo burst information read from the tape 11 by the positive azimuth head 13 is transmitted to the microcontroller 200 via the burst demodulator 38 on line 35. The burst demodulator 38 produces four signals F1_U, F2_U, F1_L, F2_L denoting the amplitudes of the servo burst signals F1 400 and F2 403 read from the tape 11 as depicted in Figure 2 representing the offset between the transducer head 13 and the track centerline 16. These four signals are processed by the microcontroller 200 to generate the aforesaid upper PES_A and lower PES_B position signals. The steps performed by the microcontroller 200 to generate the PES_A and PES_B position signals are discussed below with reference to Figure 6. When the transducer head 13 is reading information from the tape 11, the capstan driver 209 controls the capstan speed such that forces the average of the PES_A and PES_B signals to 0.71 corresponding to correct alignment of the

transducer head 13 with the track centerline 16. The output of the burst demodulator 38, which measures the amplitudes of the frequency bursts near the upper 12 and lower 14 edges of the tape 11, is passed to the

microcontroller 200 via analog-to-digital converter (ADC) 36 on line 37 whose output is processed by the

microcontroller 200 to produce the PES_A and PES_B values whose difference is then transmitted as one input to a subtractor 34 via line PES_tilt. The second input to the subtractor 34 is the output of the system model 32 on line 33 that is designed to produce a value of 0

representing proper head alignment to the track

centerline of the tape, when PES_A - PES_B is equal to PES_tilt equals 0. The modeling of an electromechanical system is application dependent as generally taught in Franklin, Powell and Workman. The output of the subtractor 34 is passed to a coefficient vector block L_t1 30 via line 39 which dynamically adjusts the error between the system model 32 including the plant (which is defined as the closed loop structure comprising the tilt actuator assembly, the bicell transducer 222, the analog servo loop components 20 (shown in Fig. 5), and the double integration performed by the microcontroller 200) to be more in line with the plant.

In the preferred embodiment, the values of the coefficient vector L_t1 30 can vary depending on the particular application or system and can be determined by the methods described in Franklin, Powell and Workman, "Digi tal Control of Dynamic Systems, " Second Edition, published by Addison Wesley, 1990. The output of the system model 32 is transmitted to another coefficient vector block K_t128 via line 25 whose output is the feedback vector to the system model 32 containing

feedback information on the position, velocity and disturbance estimates of the system model 32.

In the preferred embodiment, the values of the coefficient vector block K_t1 28 can be determined as discussed above with reference to coefficient vector block L_t1 30. The output of the coefficient vector block K_t1 28 on line 27 is compared to a reference signal, REF, at subtractor 26. The value of REF is set to 0. The result of this subtraction, the amount of tilt by which the transducer head 13 is displaced from the track centerline 16, is transmitted as the position error estimate to the system model 32 via line 45. The output of the subtractor 26 is also transmitted via line 45 to a loop compensator 24 which performs a double integration on the tilt error signal transmitted on line 45. The output of the loop compensator 24 is passed to a 12 bit DAC (that comprises one of the DACs represented as block 50 (Figure 1)) on line 23. The input of the DAC 50 on line 23 represents the point where the computations performed within the microcontroller 200 are complete. In the preferred embodiment one Least Significant Bit (LSB or count) of the 12-bit DAC 50 commands the ASHA face to tilt by 9 microinches. The DAC 50 output signal on line 21 is used to adjust the head tilt position using the analog tilt servo 20.

The analog tilt servo 20 causes the ASHA assembly 202 to tilt a prescribed distance for each LSB or count of the 12-bit DAC 50. In the write mode the DAC count is fixed to "cage" the ASHA assembly 202 in a fixed

position. In the read mode, the DAC 50 count is varied so as to tilt the ASHA assembly 202 in accordance with a processed version of the difference between PES_A and PES_B. The analog tilt servo 20 has two other lines coupled thereto, a signal line 42 transmitted from the

microcontroller 200 via DAC 50 and the output of the bicell processor 52 of the analog tilt servo 20 to the microcontroller 200 via line 44. The aforementioned lines, 42 and 44, respectively, are used to provide a means to normalize the closed loop transfer

characteristics of the analog tilt servo 20 so that the aforementioned prescribed relationship is 9 microinches per DAC 50 LSB in the preferred embodiment. This

adjustment, or normalization, is made during

manufacturing.

The analog tilt servo 20 that generates the signals that are transmitted to the voice coil motor VCM 83 to control head position with respect to the tape 11 is shown in Figure 5. One of the DACs which comprise DAC block 50, whose output is used to vary the AGC reference signal from the microcontroller 200 above and below its nominal level is transmitted to the bicell processor 52 of the analog tilt servo 20 via line 51. A second input to the bicell processor 52 is a signal from the bicell (not shown) transmitted via line 53. The third input to the bicell processor 52 is a 2.5 volt power line. The output of the bicell processor 52 is passed to the analog multiplexer 234 of the microcontroller 200 via line 44. The output of the bicell processor 52 on line 44 is also transmitted to a Twin-T Notch Filter 56 that removes certain resonate frequencies that are present in the mechanical structure of the ASHA tilt assembly 202.

The output of the Twin-T Notch Filter 56 is passed as an input to the inverting input of an amplifier 94 via resistor 75 and also a resistive/capacitive loop

consisting of a resistor 77 in series with a capacitor 79 and another capacitor 76. Also coupled to the inverting input of the amplifier 94 is the digitized representation of the signal corresponding to Equation 2 below via DAC 50 and a resistive/capacitive network comprising a resistor 71 in series with another resistor 73. Coupled between resistor 71 and resistor 73 is a capacitor 72. The non-inverting input of the amplifier 94 is coupled to a voltage source (2.5 V).

The output of the inverting amplifier 94 on line 57 is transmitted to both a resistive/capacitive loop consisting of the capacitor 76, the capacitor 79 and the resistor 77 and the inverting input to an amplifier 68 via a resistor 66 coupled to a capacitor 54 in series with a resistor 64. The non-inverting input of the amplifier 68 is connected to a constant (2.5 V) voltage source. Further, the inverting input to the amplifier 61 is also coupled to a network consisting of a capacitor 92 connected in parallel to a resistor 86 in series with another capacitor 88. The noninverting input of the amplifier 68 is coupled to a constant voltage source (2.5 V). The output of amplifier 68 is passed to a bridge driver amplifier 80 via line 63. The amplified signals on lines 43 and 82, respectively are passed to the VCM 83 to control the tilt of the transducer heads 13 relative to the track centerline 16 of the tape 11 and to the inverting input of an sense amplifier 78 whose output is passed to the microcontroller 200 via line 96.

The VCM 83 is coupled to the non-inverting input of the sense amplifier 78 via a resistor 84. The bridge driver amplifier 80, the sense amplifier 78 and the amplifier 68 are all contained within the UC 3173 custom integrated circuit 90 (shown in dashed box) manufactured by UNITRODE. The output of the sense amplifier 78 on line 96 is fed back to the circuit 90 via resistor 76 coupled to network consisting of the capacitor 92

connected in parallel to the resistor 86 in series with the capacitor 88. The values of the resistors and capacitors above for stable analog servo loop operation, are determined by conventional methods known in the art.

The operation of the tilt mechanism 10 of the present invention to accurately position a transducer head 13 relative to the centerline 16 of the tape 11 will now be described with reference to Figures 1-6. The positive azimuth head 13 reads the servo signal bursts F1 and F2 near the upper and lower edges of the tape 11 at positions 17 and 18, respectively, that have been

recorded onto the tape 11 during a previous write

operation. The amplitudes are transmitted to the burst demodulator 38 whose output on line 37 is a high

resolution indication signal of the displacement of the transducer head 13 from the track centerline 16 of the tape 11 based on the servo burst amplitudes F1 and F2 at both edges of the tape 11. The output of the burst demodulator 38 is passed to the ADC 36 of the

microcontroller 200 via line 37 which digitizes the resulting displacement signal transmitted from the burst demodulator 38. The output of the ADC 36, after

processing by the microcontroller, produces the PES_tilt signal, equal to the difference between PES_A and PES_B, where PES_A represents the PES_tilt signal, is represented by the difference between PES_A and PES_B ,where PES_A

represents the PES value generated from the servo bursts present on the upper edge 12 of the tape 11 and the PES_B represents the PES value generated from the servo bursts present on the lower edge 14 of the tape 11 where the PES value is generated by:

where the variable A represents the amplitude of the F1 bursts, and the variable B represents the amplitude of the F2 bursts, of the respective upper and lower servo burst segments of the tape 11. This PES_tilt value is compared with the theoretical or model value of the position of the head obtained from the system model 32 at subtractor 34 which tends towards 0. The signal

representing the difference between the actual position and model position of the transducer head 13 is passed to the coefficient vector block L_t1 30 on line 39 whose output on line 40 is used to dynamically update the system model 32. The output of the system model on line 25 is passed to the coefficient vector block K_t128, whose output is transmitted to the subtractor 26 via line 27 for a comparison to a reference signal, REF, whose value is set to 0. The result of this comparison, defined as the tilt error of the transducer head 13 is transmitted as a second input to the system model 32 and as an input to the loop compensator 24 which performs a double integration on the tilt error signal on line 45 which is represented as follows:

where Y(k) represents the output of the loop compensator 24, X(k) represents the tilt error signal transmitted on line 45, and t_st represents the time between the samples. The output of the loop compensator 24 is transmitted to the 12 bit DAC 50 on line 23 whose output has a tracking resolution of about 0.5%, then is transmitted to the analog tilt servo 20, whose output is transmitted to the voice coil motor VCM 83 via lines 43 and 82 to adjust the tilt of the transducer head 13 relative to the tape 11. By taking the difference between the PES_A and PES_B values and using that difference to control the tilt of the head 13 relative to the tape 11, the tilt servo has been rendered impervious to capstan 210 motion and speed variations. Thus, essentially decoupling the tilt servo of the arcuate scan tape drive 100 from the capstan servo.

The operation of the system 10 to adjust the speed of the capstan motor 210 will now be described with reference to Figures 1-6. As discussed above with reference to controlling the position of the transducer head 13 relative to the track centerline 16 of the tape 11, the recording head 13 reads the servo burst

amplitudes F1 and F2 at both the upper edge and the lower edge of positions 17 and 18, respectively, while scanning the information on the tape 11. The servo burst

amplitudes are transmitted to the burst demodulator 38 whose output is passed to the microcontroller 200 via line 37. The microcontroller 200 generates the PES values from the servo signal values transmitted from the transducer head 13, as described above with reference to controlling head positioning and averages the respective PES_A and PES_B signals together. If the result of the addition equals 0.71, then no adjustments have to be made to the capstan motor 210 speed. If the result of the addition is not 0.71, a control signal is transmitted to the capstan driver 209 whose output is transmitted to the capstan motor 210 to control tape propagation speed.

Thus, the control of tape motion is effectuated by the average of PES_A and PES_B.

By controlling the capstan speed from the average of the PES signals derived from the servo bursts near the upper 17 and lower edges 18 of the tape 11, respectively, the capstan servo has been decoupled from the tilt servo and is rendered impervious to tilt variations. By decoupling the capstan servo from the tilt servo, the settling time for the servos present in the system 100 are thereby reduced.

Burst amplitudes from F1 and F2 may vary due to differences in head samples and tape samples. To ensure that servo loop gains are predictable in the face of such variations, the servo burst amplitudes are normalized automatically each time a data cartridge is inserted in the tape drive 100. As previously described, each of the upper and lower F1 and F2 burst amplitudes vary with misalignment of head 13 to track centerline 16. In fact, when the tape 11 is moved very slowly while the heads 13 are scanning, each of the four burst amplitudes will exhibit a substantially triangular variation, ranging from a minimum to a maximum voltage, with each period of the triangle representing a movement of four tracks.

Therefore, by executing a slow movement of at least four tracks each time a cartridge is inserted, the entire range of servo burst amplitudes is present, and an estimate of the peak-to-peak amplitude and mean value of each servo burst can be acquired. Measuring the maximum and minimum values of the servo burst amplitudes leads to an exposure to errors due to noise. To minimize noise errors, the minimum and maximum servo burst amplitudes values are estimated using the information contained in the servo burst amplitude samples acquired while slowly moving a distance of four tracks.

The method for determining the normalized amplitude values A and B of the PES signals is shown in Figure 6. The tape is moved at any very slow but uniform speed under the control of the encoder servo loop, for a few seconds. The middle portion of the move, after the acceleration phase, will be at a uniform speed, and is timed to correspond to move a distance of 4 tracks. The amplitude samples of the demodulated bursts are deinterleaved to segregate the burst amplitudes of

individual heads 13. A slow speed is chosen to result in a few hundred samples for each head, with a sampling interval corresponding to one revolution of the ASHA assembly 204. A triangular shaped signal, as illustrated in Figure 7, is generated as the tape 11 is moved. It is desired to estimate the peak amplitude of the triangular waveform in the presence of noise.

The samples of the triangular wave (Figure 7) are first summed over the middle portion of the move. If the number of samples in the 4 tracks during which the speed is uniform is N, the mean value of the triangular signal is obtained by dividing the sum by N. The mean value is then subtracted from each of the N sample values. The resulting dc free triangular signal is now modulated with in-phase and quadrature Walsh functions of period equal to the time occupied by the 4 track move. This performs a synchronous rectification and therefore suppresses noise at frequencies other than at the frequency of the triangular waveform. The results of the modulation are two sequences of values, which are independently summed, giving two signals, I_s and I_c, shown in Figure 8,

proportional to the area under the two modulated

waveforms.

The formulae for the two areas are given below as a function of the peak amplitude, A, and the phase

discrepancy, k, between the triangle wave and the Walsh functions, k is expressed as a fraction of a period (assumed to be unity for the purposes of analysis) :

Now the value of A must be determined as a function of I_s and I_c; the value must be independent of k.

I_s as function of A and k

Solve for k:

Substitute k from above into equation for I_c:

leading to:

Simplify the above equation, leading to:

Eliminate the radicals by squaring:

Solve for A as a function of I_s and I_c:

Equation for estimated amplitude Am, as a function of actual peak amplitude, A, and phase, k is represented as:

The above algorithm is successful only in the appropriate quadrant. The steps performed to establish the correct quadrant are:

First, move the capstan motor 210 at a slow enough speed to move the tape 11 four tracks.

The corresponding calibration speed of the tape is :

The time period while at uniform speed over four tracks is:

The tape moves four tracks:

x_cal : =4·W X_cal=2.4·10^-3 inches (15)

Second, take 256 samples of data from the upper edge 12 of the tape 11 on each of the two position azimuth read heads 13. Each head will produce a complete

"triangle" of burst signals. For read head position l, find and save the peak value of the A burst signal, and record the sample number where the peak value was

discovered designated AMAXCNT. It is assumed that the peak value recorded at head position 3 will occur at roughly the same time as the peak burst value for head at position 1.

The sample numbers are kept in a countdown counter that is initialized to 1536. After this first phase is complete, the counter will be at 1280.

Third, change the value of the countdown counter as follows:

counter = 2464 - AMAXCNT

The capstan is then moved at the same speed as before until the counter, still counting down once per ASHA revolution, reaches 1024. For example, suppose the first phase of the software detects the burst signals as the following function of the sample number:

At the end of the first phase the countdown counter will have a value of 1280, and the A burst signal maximum will have been detected when the count was at about 1366. The second phase would like to position the tape in the middle of the illustrated "sweet spot" for maximum error tolerance. The code estimates this center to have occurred when the count was 1366 + (3/8) *256, since the center of the target region would have occurred 3/8 of a cycle before the maximum value for a periodic triangular signal. It now tries to move to the region of the next cycle corresponding to this point discovered in the first phase. To do this, the code adjusts the countdown counter so that it will be 1024 when the burst position signal is at this optimum point. This adjustment

calculation is made as follows:

current count : = 1280

time_to_next_optimum : = 256 - (sweet_spot - current count) new_count : = 1024 + time_to_next_optimum

Thus, the new count is generated as a function of AMAXCNT:

new_count : = 2464 - AMAXCNT

In this example,

new_count = 1098

Fourth, take the modified first-order Walsh

Transform of the recorded data.

The incoming A and B burst signals are sampled and divided by 4 for accumulation. This factor of 4

compensates for the fact that 4 complete burst

"triangles" are taken. Data is taken and accumulated for

1024 points, which corresponds to the counter value at the initialization of this phase. Each triangle has four separate phases, each corresponding to an individual quadrant on the Cartesian plane. To determine which quadrant is which, the counter value is ANDed with O ×

60, resulting in (for burst signal A measurements)

In addition, data is taken for servo burst B. Since the burst B pattern should be exactly opposite that of the servo burst A pattern, the updated IcO and IsO for burst B are found using the opposite sign as those for the burst A. In addition, for each of the burst signals, offset data is accumulated by adding the current burst signal to the accumulated total.

For this phase, data is taken until the counter, which decrements after each sample, is at zero. When this happens, the final normalization calculations are made.

Fifth, calculate the servo burst A and B effects by taking the accumulated effect data and dividing by 1024, keeping in mind that this integration was done on the original burst signals, not those divided by 4 before accumulation into the Walsh transform. Make the

following "simplications" :

I_c0=256·I_c I_s0=256 ·I_s (20)

Sixth, calculate the amplitude of the burst signals a) Calculate

This yields a double-word number that is guaranteed to be positive.

b) Take the high word of the above, and shift right by 3 bits, in effect, letting

c) Use the value determined in step (b) as an index into a table which approximates

The factor of two in the denominator is due to the fact that the table is word-addressed, not byte-accessible, so any offset into the table must be preshifted left by 1 to get the proper alignment. This table consists of 1024 values to handle the above function for inputs linearly distributed from 0 to 1023.

d) Get

This is the estimate of the burst signal amplitude A. A total of four of these amplitudes are taken,

corresponding to the A and B servo burst signals taken near the upper edge of the tape 11 by read heads 1 and 3. Note that the trailing signals are not normalized, as their amplitudes are assumed equal to their respective leading signals. The amplitude of A calculated in equation 3 above

represents the value of A used in equation 1 above to determine the amount of tilt by which the transducer head 13 is displaced with respect to the rack centerline 16 of the tape 11.

The procedure described above for determining the normalized amplitude A of the upper burst signal F1 is also performed to determine the normalized amplitude B of the position signal F2. This value of B is used in place of A in equations 3, 4 and 5 above, for determining the amount of tilt by which the transducer head 13 is

displaced with respect to the track centerline 16 of the tape 11.

The foregoing description of the preferred

embodiment of the present invention has been presented for purposes of illustration and description. It is not intended to be exhaustive nor to limit the invention to the precise form disclosed, and obviously many

modifications and variations are possible in light of the above teaching. The system and method for accurately positioning a head relative to a recording medium and for controlling the tape propagation speed was chosen and described in order to best explain the principles of the invention and its practical application to thereby enable other skilled in the art to best utilize the invention in various embodiments and with various modifications as are suited to the particular use contemplated. It is

intended that the scope of the invention be defined by the claims appended hereto.

What is claimed is:

Claims

1. A head positioning apparatus, comprising:

a motor operative to propagate a storage medium;

a head for reading information from and writing information to said storage medium;

a position transducer, coupled to said head, to position said head relative to said storage medium; and control circuitry operative to control said position transducer, said control circuitry further operative to control the speed in which said storage medium propagates.

2. The apparatus of claim 1, wherein said control circuitry comprises a tilt servo operative to control said position transducer, said tile servo including a first loop and a second loop.

3. The apparatus of claim 2, wherein said tilt servo comprises a first loop operative to read

information from said head, said first loop including means for determining track position.

4. The apparatus of claim 2, wherein said tilt servo comprises a second loop operative to transmit said signal to said position transducer, said second loop including a loop compensator and a voice coil magnet.

5. A method for controlling head position relative to a tape, which comprises the steps of:

(1) reading the beginning of tape servo mark;

(2) reading the end of tape servo mark;

(3) subtracting said beginning of tape value from said end of tape value;

(4) transferring the value determined in step (3) to a position transducer; and

(5) positioning said head relative to said tape.

6. A method for controlling tape propagation speed which comprises the steps of:

(1) reading the beginning of tape servo mark;

(2) reading the end of tape servo mark;

(3) adding said beginning of tape value and said end of tape value; and

(4) transferring the value from (3) to a capstan motor.