Search

Hard disk drive (HDD) is a mechanical data storage device with head slider flying over disk medium surface. The head medium spacing keeps reducing for the sake of linear density increment. The head flying height (FH) in HDD product is reduced to less than 1 nm of back-off spacing. The intermittent head disk contact happens in HDD and causes additional head motions in 3-dimensions: FH, off-track, and jitter. Increasing FH deteriorates the BER performance. The off-track motion causes track squeeze of adjacent track during writing process and has the potential to loss data in squeezed track. Head jitter motion makes the synchronous writing more challenged for bit patterned medium. It is desired to measure head 3D motion inside HDD.

The lower head–media spacing (HMS) is always expected for achieving better recoding performance at high linear density. Besides the efforts to reduce the overcoat thickness of head and media, the low flying height (FH) and smaller FH modulation (FHM) are necessary to ensure the low HMS and good interface reliability. In the current HDD, the FH change is measured from the readback signal by the Wallace spacing loss equation. The real-time dynamic FH signals consist of repeatable and non-repeatable FH changes, or FHM. The repeatable FHM (RFHM) correlates with the disk morphology variations. It usually has the variation of transition parameter $a$ inside. Because the change of $a$ is the noise of the RFHM, it has to be very careful for the compensation of the RFHM by thermal FH control. Otherwise, it may cause unexpected head–disk contact. The non-RFHM (NRFHM) measurement is able to remove the media-related noise, which has a good testing accuracy of around 1 A. This paper studies both RFHM and NRFHM characteristics of four types of sliders with different air-bearing surface designs.

Head media spacing of hard disk drive (HDD) is expected to continue its reduction in order to support areal density growth. At sub-nm clearance between the head and the disk, the intermittent head disk contact and interactions may happen. This does not only cause the fly height modulation but also induces the off-track motions as well. It is desirable to understand the characteristics of the 2D motions in the near contact regime, so as to enable further reduction of the clearance and to improve the reliability of HDD. This paper presents a method to measure instantaneous fly height (FH) motion and cross track motion concurrently by using read back signal from the dual data tracks written at different frequencies. The method is able to separate the FH motion and cross track motion of the head.

Perpendicular magnetic recording (PMR) in the hard disk drive is approaching its physical limits. In an earlier work, the dedicated servo (DS) recording system has been proposed to provide continuous position error signal for servo, enable higher servo sampling rate, and improve the overall servo performance. A further benefit is that the DS layer results in surface area savings at the data layer. However, it was also reported that the embedded servo layer introduces baseline variation and non-linear transition shift (NLTS) to the readback signal of the data layer. In this paper, we propose novel signal processing techniques to improve the bit error rate (BER) in DS recording. The synchronous averaging technique is proposed to improve the BER in the presence of baseline variation distortions. Further, the servo and data-dependent noise prediction method is proposed to mitigate the effect of the NLTS. Through the use of these techniques the linear density loss from the conventional PMR media is reduced.

The perpendicular magnetic recording (PMR) in hard disk drives is approaching its physical limitation. The emerging technologies, such as heat assisted magnetic recording and microwave assisted magnetic recording have been proposed to record on magnetic media with thermally stable smaller size grains at higher areal density (AD). However, in the media fabrication, achieving well-isolated small size of grains is more challenging than obtaining high ${K_{u}}$ material as recording media. Reducing the number of grains per bit is a major path for keeping AD growth of PMR in recent years. To minimize the SNR penalty at a smaller grain number per bit, pushing more on track density is the right approach. With the 2-D magnetic recording (TDMR) readers for inter-track interference cancellation, the off-track read capability is improved significantly for allowing a narrower track read. In the drive working environment, when the external vibration or other mechanical disturbance happens during the writing process, it creates more track squeeze at adjacent tracks and leaves a very narrow track at some locations of the track. When the track width is narrower than the squeeze to death width in the 747 curve, it causes hard failure in the channel. To solve the track squeeze problem, this paper proposes to add an additional magnetic recording layer in between the data recording layer and the soft underlayer of conventional PMR media. This additional recording layer is used to record servo information only. The continuous positioning error signal is able to improve the servo performance and to provide the real-time monitoring of the positioning error. When it is under bad servo conditions, the writing process can be stopped to avoid nontolerable track squeeze. The continuous servo signals are designed to be of moderate intensity at very low frequency, and its impact on data signal has been minimized. The linear density gap between the dedicated servo media and the conventional PMR media is able to be controlled within 3%. As the dedicated servo system keeps only around 100 wedges of track ID and sector ID at the data layer, the surface area saving at the data layer can break even in capacity. The dedicated servo technology together with TDMR readers is the key technology to achieve ultrahigh track density during both writing and reading processes.

Current typical perpendicular commercial media has grain size of around 8 nm, which the size has remained more or less the same since its introduction. Typical bit lengths for the media can go as low as 15-20 nm. Transition jitter, which is one type of media noise, tends to occur at regions where magnetic transitions could be found. Such noise could affect the storage and readability of the data, especially as more accuracy is demanded when hard disk industries push for higher and higher densities. This work describes the methodology of using averaging methods to measure the writer footprints jitter profiles which are recorded at different writing currents on a Spinstand. Two different commercial heads have been used for the evaluation. Multiple down-track footprint averaging and least square fitting of the individual transition footprint data against the noise-free footprint are used to retrieve the jitter profiles of the writers at different writing conditions. The statistical jitter profiles are then used to plot various cross-tracks and regionalized jitter distribution and probability plots. From the plots, it is quite conclusive to determine that writer A at 50 mA especially at the track center (TC) has the minimal and thus best jitter profile amongst the other writing conditions and heads used.

Writing processes can cause extra medium noise, which is the dominant noise in the perpendicular recording system. This work describes the methodology of using averaging methods to measure the writer footprints and its noise profiles, which are recorded at different writing currents on a Spin-stand. Two different commercial heads have been used for the evaluation. The test is performed on a 2.5 inch commercial disc at the middle diameter (MD) location, rotating at 5400 RPM spindle speed, and at 0 skew angle. The center track is DC erased over an AC erased background and then recorded with a repetitive footprint data pattern and an alignment data pattern. Multiple revolution averaging and down track footprint averaging are used to retrieve the noise free footprint profile of the writer. Using the noise free footprint profile, the statistical characteristics of noise can be derived from the individual noisy footprint which shows the media noise characteristics induced by writing process. Using footprint and noise profile data, signal to noise ratio (SNR) for regions near the trailing edge of the writer profile are calculated and suggests that the ideal writing condition for writer A and B to be 55 and 25 mA respectively.

As the head-disk flying clearance reduces to about 1 ~ 2 nm during read-write operation, the write current induced writer pole-tip protrusion (W-PTP) may affect the reliability of the hard disk drive (HDD) performance. The dynamic temporal heating and cooling profile of thermal actuator and the W-PTP are investigated using the Wallace spacing loss method. The work suggests that the dynamic W-PTP can be closely compensated by the dynamic cooling of thermal actuator.

The head media integration played a very important role to push up the areal density of perpendicular recording successfully. Besides the success of good sensitivity and high resolution reader at desired feature size of recording bit, the advanced writing process is the key enabler to sustain the areal density growth continuously. The sharper field gradient of writer and the narrower switching field distribution of recording media improve the transition jitter and promote the SNR of recording media. Without reducing grain size, the areal density of perpendicular recording has been increased by a few times. In this work, we investigate the performance of writing process via the measurement of head field footprint in recording media on the Spinstand. The feature size variations of head footprint at different writing current and different flying height (FH) illustrate the effective writing process visually. The transition curvature with transition width value at the trailing edge of writer pole is the key indicator to show the linear density capability related to writing process.

Recently, the announcement of the first 8 TB shingled drive to be available commercially for archival purposes by Seagate [1]. In shingled magnetic recording (SMR), a wider writer is able to write narrower track for pushing higher track density. At narrower track width, the jitter noise increases and it results in lower on-track SNR. Usually, the linear density is reduced in SMR and the areal density gain comes from more track density increment. Comparing to PMR, the SMR is able to increase the areal density by 15% to 30% depending on the servo performance of the drive. At reduced linear density, there is a big jump in track density.

The jitter noise is the dominant media noise in perpendicular recording. It is proportional to cross track correlation and the square of transition parameter “a” [1]. Besides the efforts of minimizing the value of “a,” it is also very important to control the variation of “a” in each individual writing process. In this work, from the time domain head media spacing measurement, there is a significant signal amplitude variation arising from individual writing process at the same writing conditions. After removing the change related to disk morphology, the amplitude difference of individual writing reflects the variation of “a.” Reducing such “a” variation is helpful to control the transition noise.

The gradient of magnetic head field is very critical in the writing process and it is one of the key factors to determine the achievable linear density of magnetic recording system. This work proposed a new method to measure the writer field gradient on Spinstand. The even harmonic ripple effect [4] is used to produce the modulated magnetostatic field at the writing transitions generated from the old data pattern at leading edge side of the writer. The newly written single tone signals have the modulated transition shift. Through the duty cycle measurement of the single tone signals, the delta transition shift can be measured. With the calculated magnetostatic field based on the media magnetic properties and the measured size of writer footprint, the writer field gradient equals the ratio of delta magnetostatic field over delta transition shift.

In the hard disk drive (HDD), the Wallace based in-situ flying height (FH) test is widely used for FH measurement since the application of thermal FH control (TFC). The Wallace equation shows the spacing loss of signal amplitude at a particular frequency or wavelength. It has a very simple expression for longitudinal recording and becomes more complex in perpendicular recording with soft magnetic underlayer. In the FH measurement, the signal amplitude at testing frequency has to be in good quality in order to achieve high precision FH measurement. In this work, an integration of power spectral density-based version of the Wallace spacing loss equation is developed. The use of the latter equation enables in-situ FH measurement via the monitoring of relaxed quality (multitone) readback signals over a band of frequency. The new method is able to test the FH change at high precision on AC erased track.

Currently, the areal density grow in HDD still depends on the reduction of grain number per bit. In order to have less SNR penalty, it has to scale more on the track density rather than the linear density. This work proposed to insert an additional servo recording layer between data recording layer and SUL of conventional PMR media. With high quality continuous servo signals, the dedicated servo recording system has better servo performance than the conventional embedded servo for the same servo mechanics. This guarantees the high track density requirement of magnetic recording system.

In perpendicular magnetic recording system, the jitter noise is the dominant noise from the medium. The jitter noise is proportional to the cross-track correlation and the square of transition parameter [1] . For a given head and medium, pushing linear density has more SNR penalty than increasing track density. That is why all the emerging technologies, such as shingled magnetic recording, microwave assisted magnetic recording, heat assisted magnetic recording, and bit pattern media, emphasize more on the track density increment for the growth of areal density. One challenge from narrower track width is the inter-track interference (ITI). In the two-dimensional magnetic recording (TDMR), multiple readers were proposed to reproduce the signals from multiple tracks [2] . The recent study shows that TDMR readers are effective to cancel the ITI from the adjacent tracks and are able to achieve higher track density [3] . The pitch between the readers along cross-track direction needs to be set properly for the optimal ITI cancellation. But due to the skew angle, one of TDMR readers may have to read near the middle of two tracks. When the recording bits on two adjacent tracks are in the opposite polarity, it may create local flux paths between reader hard bias and medium bits. This could generate multi-domains in the read sensor and produces additional reader noises. This paper studies this noise dependence on both the off-track position and the bit length.

A model is introduced in this paper to describe the effects of thermal magnetic noises in the read sensor under off-track reading conditions taking into account the influence of the magnetic field generated by the recorded magnetization patterns on the medium. Our numerical studies show that there exist additional noises in the hard biased reader, depending on recorded patterns on the magnetic media. The influence of the recorded magnetic patterns is not only on the variation in reader resistance but also on the high frequency noises caused primarily by the thermal fluctuations. The theoretical analysis is based on the micromagnetic modeling of the state of magnetization in read sensor taking into account of its external magnetic fields due to both the hard bias and the media magnetization patterns. The effect of the thermal agitation of the gyromagnetical precession of magnetizations is evaluated by considering the angular magnetic noise.

In this study, we investigated the performance of perpendicular magnetic recording with squeezed tracks. Based on spinstand signals, we developed a signal generation module to generate the readback signals for tracks of various squeezed track width and bit length. To mitigate the drop in bit error rate (BER) as squeezed track width decreases, we implemented the bidirectional pattern-dependent noise prediction (BiPDNP) detector with low-density parity check codes. To improve the BiPDNP detector over earlier work, we modified the log-likelihood ratio computation. Our results show that our proposed BiPDNP detector improves the bit error rate, and allows smaller squeezed track width and smaller bit length for the same BER performance.