TY - GEN
T1 - A study on optimal BAR in array head reading
AU - Kondoh, T.
AU - Nakamura, Y.
AU - Nishikawa, M.
AU - Osawa, H.
AU - Okamoto, Y.
AU - Kanai, Y.
AU - Muraoka, H.
N1 - Funding Information:
I. INTRODUCTION Two-dimensional magnetic recording (TDMR) which combines shingled magnetic recording (SMR) [1] with two-dimensional signal processing attracts attention in hard disk drives (HDDs). The SMR is a magnetic recording system which increases the track density by writing tracks without the guard band [1]. Therefore, the influences of writing and reading intertrack interference (ITI) deteriorate the quality of reproducing waveforms. The equalization using two-dimensional finite impulse response (TD-FIR) filter for the waveforms from three adjacent tracks is proposed [2] as a method to reduce the influence of reading ITIs. In this study, we evaluate the bit aspect ratio (BAR) in TDMR R/W channel under a specification of 4 Tbit/inch2 using bit error rate (BER) obtained by computer simulation. II. READ/WRITE SIMULATION We employ the LDPC coding [3] and iterative decoding system including a partial response class-I (PR1) channel [4] with the TD-FIR filter. The input sequence is encoded by a 128/130(0, 16/8) run-length-limited (RLL) encoder [5] and a (30, 3)-regular LDPC encoder. The encoded sequence is recorded on the granular medium model [6] with non-magnetic grain boundaries based on a discrete Voronoi diagram by a write head which has a triangular main-pole, a trailing-shield and a side-shield for SMR [7]. The average grain size, the grain size dispersion normalized by an average grain size and the average width of non-magnetic grain boundaries are 5 nm, 0.2 and 1 nm, respectively. In this study, we set BAR to 1, 2, and 3, respectively under a specification of 4 Tbit/inch2 on the medium. The bit length and the track pitch for BAR = 3 are 7.3 nm and 22.1 nm, respectively. We use an array head with three wraparound type of readers. Each reader scans the center of the intended, previous and following tracks. The widths of shield gap in cross and down track directions are 30 nm and 22 nm, and the width and thickness of the magneto-resistive element are 17 nm and 2 nm. Additive white Gaussian noise (AWGN) is assumed as the system noise. The signal-to-noise ratio (SNR) for the system noise at the reading points is defined as SNRS = 20log10 A3 / σS [dB], where A3 is the saturation level of the reproducing waveform from an isolated magnetic transition in the case of BAR = 3 and σS is the rms value of the system noise in the bandwidth of the channel bit rate fc. A channel bit response including R/W process on the intended track is equalized to the PR1-target by the two-dimensional equalizer composed of the low-pass filters (LPFs) having cut-off frequency xh normalized by fc and TD-FIR filter with Nt taps, where Nt is the number of taps for a reader. The PR1 channel output waveform is iteratively decoded by the turbo equalization [8] performed between an a posteriori probability (APP) decoder and a sum-product (SP) decoder. Here, isp and iglobal stand for the maximum number of iterations in the SP decoder and the turbo equalization, respectively. The channel error rate (CER) at the APP decoder output is calculated by comparing the input sequence of R/W channel with the hard-decided sequence of APP decoder output without the iterative decoding. After the given iteration, the output sequence is obtained through a hard decision unit and the RLL decoder. Then the BER is calculated by comparing the input sequence of RLL encoder with the output sequence of RLL decoder. III. CER AND BER PERFORMANCE Figure 1 shows the CER performance for BAR, where xh = 0.4, Nt = 15 and SNRS = 25 dB. We assume the same system noise sequence in every case in the simulations. As can be seen from the figure, CER is the lowest at BAR = 2. In the case of BAR = 1, the influence of reading ITI increases and can not be fully removed by the TD-FIR filter. In addition, in the case of BAR = 3, the most of reading ITI is removed by the TD-FIR filter, but the influence of intersymbol interference increases with the reduction of the bit length. Therefore, the LDPC coding and iterative decoding system achieves good CER performance at BAR = 2. Figure 2 shows the BER performance for SNRS, where xh = 0.4, Nt = 15, isp = 10, and iglobal = 10. In the figure, the circles and triangles show the BER performances in the case of BAR = 2 and 3, respectively. As can be seen from the figure, the BER performance for BAR = 2 shows about 3 dB better performance compared with that for BAR = 3. ACKNOWLEDGMENTS This work was supported in part by the Advanced Storage Research Consortium (ASRC), Japan, and the Japan Society for the Promotion of Science (JSPS) under Gants-in-Aid for Scientific Research (C) 26420358.
Publisher Copyright:
© 2017 IEEE.
PY - 2017/8/10
Y1 - 2017/8/10
N2 - Two-dimensional magnetic recording (TDMR) which combines shingled magnetic recording (SMR) [1] with two-dimensional signal processing attracts attention in hard disk drives (HDDs).
AB - Two-dimensional magnetic recording (TDMR) which combines shingled magnetic recording (SMR) [1] with two-dimensional signal processing attracts attention in hard disk drives (HDDs).
UR - http://www.scopus.com/inward/record.url?scp=85034645878&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85034645878&partnerID=8YFLogxK
U2 - 10.1109/INTMAG.2017.8007594
DO - 10.1109/INTMAG.2017.8007594
M3 - Conference contribution
AN - SCOPUS:85034645878
T3 - 2017 IEEE International Magnetics Conference, INTERMAG 2017
BT - 2017 IEEE International Magnetics Conference, INTERMAG 2017
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2017 IEEE International Magnetics Conference, INTERMAG 2017
Y2 - 24 April 2017 through 28 April 2017
ER -