For a 120MB Large Capacity Floppy Disk Drive with fully backward compatibility in write and read, a magnetic head is developed. The magnetic heads consist of 3μm-Sendust Double MIG core for 120MB, polycrystalline Ferrite core for backward compatibility and "U-shape" ceramic slider. To measure ghost pulse causing from the poor interface between metal and ferrite in double MIG, the correlation factor using the mean square root method is applied to the ripple curve. The ripple curve is derived during write and read over changing write frequency. It is found out that the ratios of ghost pulse to main pulse and the factor are good correlation by comparing between modeling and experiment.

View full abstract

In a recent digital magnetic recording system, it has been pointed out that effects of the write current switching time is one of measure of recording performance. The effects of the current switching time on recording characteristics are discussed in this paper with experiments by using an arbitrary waveform current driver. Playback waveforms' changes with various write current switching time are compared for a longitudinal and a perpendicular medium. Pulse width change due to wide variation of write current switching time is discussed for either perpendicular or longitudinal recording. It has been also found that the medium transition noise increases in longitudinal recording, though no transition noise is observed in perpendicular recording, when the switching time becomes longer.

View full abstract

Extremely high areal density recording associated with multi-track submicron trackwidth recording was proposed. In this method, parallel written data among several tracks can be read collectively with a single track head of wider trackwidth. In this report, two new encoding/decoding schemes introducing multi-level partial response are investigated for the head. It is demonstrated by both a simulation and an experiment that recorded information on two tracks can be descriminated even with a single wide track read head by means of their schemes.

View full abstract

Conventionally, the main function of recording codes has been thought to be matching the property of a signal sequence to that of the recording channel. In this paper, we propose to successfully use another function of the recording codes so as to combine them with the Partial Response (PR) method, and hence describe their coding rule by the trellis diagram. We select four FM family codes whose coding rules can be expressed by applying an automaton model. As the PR methods, PR (1,1) and PR (1,0,1) have been selected based on previous reports. Simulation results show that an SN ratio improvement of about 6 dB at a bit error rate of 10^<-4> can be obtained over the conventional PRML methods.

View full abstract

We developed a new robust ternary recording code called a 13B9T code, for increasing the recording density in magneto-optical disks. Like the 4B3T code, it satifies the c constraint property. It also sucessfully attains a k constraint property which is not possessed in both the 4B3T and 3B2T code. In terms of bit error rate (BER) characteristics, The coding efficiency of this code is 1.44. Since the 3B2T code combined with PR (1,1) method is regarded as the most attractive combination for MO recording systems bit error rate characteristics, we investigated the BER of our new code (13B9T code) for comparison. Computer simulation results show that the developed 13B9T code combined with the PR (1,2,1) method gives a 1.5dB better CN ratio than the 3B2T code with the PR (1,1) method at BER=10^<-4>.

View full abstract

We investigated RLL (Run Length Limited) codes suitable for high linear density recording in a light intensity modulated magneto-optical (MO) disk. First, we propose new RLL codes, (1,6) code, VTM (Variable Two Modulation), and VFM (Variable Five Modulation). Next, when using RLL codes, d=2 and d=4,to achieve a linear density 1.34 times larger than conventional CD's, and above, we detected a lower bit-error rate in the case of d=4 when compared to d=2. As a result, with a viterbi detector and a linear recording density of 2.63 bit/μm (λ : 785nm, NA : 0.50,CD×1.55), we achieved a bit-error rate of less than 10^<-4> in (4,19) RLL code.

View full abstract

We studied a PRML (Partial Response Most Likelihood) system for a phase change optical disk. We proposed the PRML system which detected nonlinear distortion by using survival path and controlled expected signal levels adaptively. We confirmed that bit error rate of this system was reduced to 1/5 to 1/100 as compared to bit by bit decoding. And we proposed a new algorithm for adjustment of the sample-timing phase which controlled oscillator with evaluating phase error from survival path.

View full abstract

For 1,7 code and 2,7 code, which can be applyed to the optical disk recorder, we studyed at the computer simulation to record at mark edge method, to play back at PRML (combined partial response adaptive equalizer and viterbi decoder), to compare bit error rate. Next we made recording and playing back experiment at the track pitch of 1.2μm land/groove and minimum bit wavelength of 0.33μm/0.30μm, and analyzed those data. At the result, we found that 1,7 code is more advantageous about 2dB in SN ratio and it is possible to realize minimum bit wavelength of 0.30μm.

View full abstract

We've already developed a DC suppression method which decides the changing way of modulations based on combination of data and channel code to minimize DC. Table information, which shows the changing way of modulations, are multiplexed at the head of each data block. But a mistake of table selection causes error propagation during a data block decoding if table information cannot be read. In this paper we developed a DC suppression method with faculty of error correction by taking advantage of what data, which are Reed-Solomon code, are constituted on Galois field. This method multiplies each byte datain a data block by some constant elements of Galois field, modulates some multiplied data and selects a block of channel codes which has the least DC component. We can do DC suppression effectively without error propagation with this method.

View full abstract