A measurement of signal-to-noise ratio versus trackwidth from 128 µm to 4 µm

Abstract

Bulk-erased tape noise will vary as the square root of the reproduce head trackwidth if the noise signal is uncorrelated across the track. Recent models of erased noise involve clusters of interacting particles that could be as large as a few microns. As the trackwidth (TW) approaches the cluster size, the noise should become correlated across the track and the tape-noise-limited signal-to-noise ratio (SNR) should become constant as TW is further reduced. A check of this idea using inductive heads is impractical. Magnetoresistive (MR) heads have very high signal and low noise and so are well suited for this task, but precautions must be taken to minimize thermal and Barkhausen noise. A multichannel MR head having TW from 128 μM to 4 μM was built to explore the areal reproduce density limits of MR heads and measure the bulk-erased SNR versus TW for 3M 5198 tape. Tape-noise-limited performance was achieved with the narrowest 4μM TW channel at a wavelength as short as 1μM. A wideband-equalized SNR of 20 dB was demonstrated with this channel at a wavelength of 1.27 μm or 40 kilo flux changes per inch (40 kFCI), at a tape speed of 38.2 cm/sec. With a 100% guard band, or an 8-μm track-to-track pitch, this corresponds to the very high areal storage density of 127 MFC/in <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> . The SNR was found to vary as <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\sqrt{TW}</tex> down to TW = 4 μm, which indicates that the expected particle cluster size must be smaller than 4 μm in the crosstrack direction.