Browse Prior Art Database

Head-Positioning System Employing Ternary-Valued Position Error Signal

IP.com Disclosure Number: IPCOM000060908D
Original Publication Date: 1986-Jun-01
Included in the Prior Art Database: 2005-Mar-09
Document File: 3 page(s) / 51K

Publishing Venue

IBM

Related People

Wallis, CN: AUTHOR

Abstract

A head-positioning system for a small hard disk file uses a single servo sector with a ternary-valued position error signal and a stepping mechanism able to step to fractions of a track. An initial approximation of the offset of the real track position from the nominal is made on the basis of a thermal expansion model by assigning values to constants in a model equation. If an access to the track concerned does not produce the null position error signal (PES) value, the head is stepped in the appropriate direction until null is reached. The model constants are updated in accordance with the actual null position so that subsequent accesses should reach the track first time. The steps for computing the necessary constants are as follows.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 3

Head-Positioning System Employing Ternary-Valued Position Error Signal

A head-positioning system for a small hard disk file uses a single servo sector with a ternary-valued position error signal and a stepping mechanism able to step to fractions of a track. An initial approximation of the offset of the real track position from the nominal is made on the basis of a thermal expansion model by assigning values to constants in a model equation. If an access to the track concerned does not produce the null position error signal (PES) value, the head is stepped in the appropriate direction until null is reached. The model constants are updated in accordance with the actual null position so that subsequent accesses should reach the track first time. The steps for computing the necessary constants are as follows. Calculation of Thermal Shift We assume that the amount by which the head needs to be moved from the nominal position (as defined by the stepping-motor) is a linear function of track number. We will refer to this expression as the 'Offset Characteristic.' A higher-order approximation could be dealt with, but is probably unnecessary. That is to say, S = A.n + B where S is the magnitude of the thermal shift n is the track number A,B are parameters which have to be generated by an algorithm from measurements made on the disk. We start by assuming that A and B are zero. Each time we seek to a track, we apply the shift calculated from the offset characteristic above. As a result, we find the head in a position which gives one of the three possible outputs from the servo demodulator: in, out, or null (I, O, or N). ***** SRR ORIGINAL DOCUMENT ***** The positive signs are chosen if the demodulator output is I, and the negative if it is O. E is chosen to be small compared with the permissible TMR (track misregistration). In the simplest case, E would be constant, but if the speed of convergence is important, it could easily be varied to give optimal binary search behavior. This algorithm is a simplified version of a recursive least- squares fitting algorithm, which needs only one multiplication (the value of n/N) to be done in the microprocessor. This is an appropriate compromise between complication and performance. If it is desired to avoid multiplications altogether, fixed increments for A and B could be chosen which would converge eventually, but would be slower still. PES Generation There is one servo sample per revolution recorded on each track, conveniently immediately after Index. The servo sample consists of three bursts of sine-wave, each at the same frequency, but having different phases. The way in which the phase variations are disposed is shown in Fig. 1. The frequency is chosen to be as high as possible consistent with the constraint that the track-to-track phase error introduced by clocking error when the servo sample is written must be a small fraction of a cycle. If the clocking accuracy is poor, and the frequency is cons...