Erich P. Valstyn

Performance of single-turn film heads

This paper begins with a review of previously published work on film recording heads and then reports on write-read experiments with single-turn film heads at a medium velocity of 40 m/s (1575 in/s). A typical film head is compared to a conventional ferrite head with regard to readback pulse shape, transition-density response, and efficiency. The results are in agreement with theoretical predictions.

The write field of a magnetic-film recording head

Magnitude and distribution of the write field generated in the gap region of a miniaturized single-turn recording head are computed on the basis of a micromagnetic analysis. The head consists of a copper strip conductor coated with a high-permeability film containing a gap. Computations predict that magnetic-film heads can produce fields adequate for writing on conventional recording media with currents of a few hundred milliamperes. Also, the results indicate that the write resolution of such heads may be expected to be better than that of heads now in use.

INTEGRATED HEAD DEVELOPMENTS

Miniaturized recording heads fabricated by techniques similar to those used in integrated-circuit technology, namely evaporation, sputtering, plating and photoetching, were first proposed by Gregg,’ Barton and Stockel * and Barcaro and associate^.^ Gregg’s patent was filed in 1961, and the paper by Barton and Stockel appeared in 1964; hence, the idea is comparatively old. However, theoretical and experimental work dealing with feasibility and performance of such devices was not published until less than two years ago.

Coupling between flat films

The magnetization distribution in a rectangular film, magnetized parallel to one side, is approximated by assuming that the magnetization \overrightarrow{M} has the same direction everywhere in the film and that its magnitude rises from zero at the edge to the saturation magnetization M s at a distance c from the edge, maintaining this magnitude throughout the remainder of the film area. The function representing the magnitude of \overrightarrow{M} in the edge region is chosen in such a way that the resulting field has no singularities, c is determined by the coercive force, which is set equal to the sum of the maximum demagnetizing field and an applied reversing field. This model lends itself better to the discussion of demagnetizing effects in single or coupled films than either the ellipsoid or the line-charge model. It takes account of the fact that the width of the edge region, where curling and domain walls occur, varies with the applied field, and, in this respect, it agrees fairly well with experiment. It, furthermore, permits the definition of a disturb sensitivity for memory elements consisting of single or coupled films.

Williams-Comstock model with finite-length transition functions

The theory of the third-order-polynomial (TOP) and fifth-order-polynomial (FOP) magnetization transitions is presented. These transitions have a finite length, rather than an asymptotic approach to /spl plusmn/M/sub r/., which is the case with some widely-used transition functions. The Williams-Comstock model is used to obtain the transition parameters, which are equal to half the transition lengths, resulting in quadratic equations and simple expressions. In this analysis, the write-field gradient is maximized with respect to the deep-gap field, as well as with respect to the distance of the transition from gap center, which results in a higher gradient than is obtained with the original Williams-Comstock approach, at the expense of a higher write current. Analytic expressions are obtained for the read pulses of inductive and shielded magnetoresistive heads, and equations for nonlinear transition shift are derived for the arctangent and the TOP transitions. The results are compared with those obtained using arctangent and tanh transitions and with experiment. In addition, certain aspects of the write-process Q function and the optimum deep-gap field are discussed.

Write noise in magnetic-film memories due to current diffusion in the ground plate

In magnetic-film memories which have the bit current returning through a ground plate, the write noise, i.e., the sense-line voltage caused by the bit-current pulse, contains a slowly decaying component. Theoretical and experimental results show that the time derivative of this component, which should be as small as possible for best rejection by the sense system, can be reduced by 1) ap- propriately stratifying the ground plate, 2) keeping the ground plate at a low temperature, or 3) terminating the bit-sense line with a suitable RC network. The experiments were carried out using a scale model of the bit-sense line, with the physical dimensions, the time scale, and the resistivity of the conductors all being increased by the same factor. Electrical properties of such a scale model are identical to those of the simulated system.

Bit-line interaction in magnetic-film memories

The noise generated in the bit-sense lines of a magnetic-film memory during the write operation consists of two components: the drive noise due to energizing the line under consideration, and the interaction noise caused by coupling between the bit-sense lines. The latter component is the subject of this paper. By an analysis of pulse propagation in a transmission system consisting of two identical parallel strip lines above a stratified ground plate, it is shown that the relative significance of the interaction noise increases with line-to-ground spacing and that its time derivative at read time can be made to vanish by an appropriate stratification of the ground plate. The stratification which achieves this is close to the one for which the time derivative of the drive noise has a minimum. Hence the time derivative of the write noise, which should be as small as possible for best rejection by the sense system, can be reduced by ground-plate stratification. It can also be reduced by keeping the ground plate at a low temperature.