1M. Suk Effect of mechanical design of the suspension on dynamic loading processReceived: 2 July 2003 / Accepted: 24 February 2004 / Published online: 3 August 2005_ Springer-Verlag 2005Abstract: In designing a load/unload system utilized in hard disk drives, necessary care needs to be taken to ensure that the slider does not damage the disk surface during loading and unloading processes. However, a small deviation in the design point of the preload between the load-dome and flexure can lead to undesirableloading processes resulting in an adverse number of slider/disk contacts. In this study, we show that if the preload between the load-dome and flexure is too low, the slider can oscillate causing the corners of the slider to contact the disk multiple times even though the slider is a few microns away from the disk. In addition, the slider can be sucked down towards the disk resulting in a complete separation of the load-dome from the flexure assembly leading to uncontrolled loading conditions.This separation occurs while the suspension is still on the ramp, and thus no preload is exerted on the slider immediately following the separation. Consequently, the slider flies at a flying height higher than the design point until the gap between the load-dome and flexure closes. Hence, the suspension must be carefully designed to suppress slider oscillation and to ensure that the loaddome does not separate during the loading process.1 IntroductionOne of the requirements in designing a load/unload system utilized in hard disk drives is ensuring that the slider does not damage the disk surface during loading and unloading processes. Since it is difficult to avoid slider/disk contacts in entirety, however, the system is designed to minimize the number of slider/disk contact events and to lessen the consequences when contacts do occur. The likelihood of slider/disk contacts depends on the loading speed, disk speed, static attitude of the slider, air-bearing roughness, slider geometry, etc. For example, sliders with a large radius of curvature at its corners can eliminate disk damage by reducing the contact stress between the slider and disk surface (Suk and Gillis 1998). Many recent studies have considered the effect of suspension, limiter, and air-bearing designs on the robustness of the loading and unloading processes (Bogy and Zeng 2000; Hua et al. 2001; Liu and Zhu 2001; Zeng and Bogy 2000). Howeve2r, most of these studies have primarily focused on the unloading process since this part of the sequence usually reveals interesting dynamic processes due to the effects of negative pressure airbearing designs. The negative pressure region of the airbearing resists the unloading action resulting in storage of potential energy in the flexure and suspension assembly.When the slider is finally pulled away from the disk and the potential energy is released, the slider can oscillate violently (Fig.1). On the other hand, for areasonably well-designed system, the loading process does not exhibit such a behavior. Hence, most have primarily investigated the unloading process giving onl a cursory attention to the more critical loading process. Most designers of load/unload systems will find that the loading process can be more troublesome compared to the unloading process. Besides potentially causing damage to the disk, other problems can be encountered during the loading process. For example, in some instances, the slider may never load to the designed flying height, but rather, load at flying heights on the order of 1 lm (Suk et al. 2004). In this paper, we show how a small deviation in the mechanical design of the flexure/suspension assembly can increase the probability of slider/disk contacts that can lead to a significant number of disk contacts in one single load cycle. Specifically, we show that a suspension system with low preload between the flexure and loaddome can lead to loading of the slider at an uncontrolled static attitude and velocity. The problems associated with this particular aspect of design can be easily identified by measuring the full-body capacitance during the loading process.2 Description of experimentThe slider loading dynamics was investigated using a laser-Doppler vibrometer (LDV), 62 kHz frame rate high-speed camera, and full-body capacitance. The experimental setup consists of a standard load/unload tester. The capacitance meter measures the full-body capacitance between the slider and disk while the slider is loading onto the disk, similar to the one used in (Suk et al. 2004). The slider was loaded onto and unloaded from the disk using a moving ramp wh3ile keeping the slider/suspension assembly fixed over the OD region of the disk. The vertical motion of the trailing edge of the slider was measured using an LDV. All tests were carried out using an 84 mm glass disk and a negative pressure bobsled type slider with the disk rotating at 10 krpm. The pitch-static attitude (PSA) of the sliders used in the experiment was between 1 and 2_. To show the effect of lack of preload between the flexure and load-dome, we chose two suspension assemblies that are essentially identical with the exception of the preload. Since the difference in the magnitude of the preload is difficult to measure, only the existence of substantial difference is verified. To do this, we mount the head suspension assembly with normal preload (NPHSA) onto the ramp. A small weight, that is sufficient to cause load-dome separation from the flexure, is then attached to the flexure. The amount of separation is measured with a properly positioned CCD camera.Similar measurement is made for a head suspension assembly with low preload (LP-HSA). Figures 2 and 3 show optical images of the load-dome and flexure taken under the same conditions for both NP-HSA and LPHSA, respectively. A greater load-dome separation from the flexure is observed for LP-HSA than the NP-HSA, confirming that LP-HSA has lower preload than NPHSA. 3 Results and discussion negative pressure sliders The slide4r loads onto the disk and then follows the runout of the disk as expected. The bottom plot in Fig.3 is the corresponding full-body capacitance measurement, which shows a single jump in the capacitance at the moment the slider loads onto the disk. A similar measurement for LP-HSA is shown in Fig.5. In this case, the slider oscillates before loading onto the disk unlike the case with a higher preload between the load-dome and flexure. Furthermore, the slider’s vertical loading velocity suddenly increases when the slider is about 50 lm away from the disk. Associated with this sudden increase in the velocity, the capacitance measurement reveals multiple sharp transitions. Following the transitions, the capacitance does not reach the maximum value for another 1 ms or so. These observations indicate a problem, but it is difficult to ascertain the precise dynamics due to the low measurement bandwidth. Higher resolution measurement reveals that the slider contacts the disk multiple times (Fig.6)—note that this exact behavior does not occur for every suspension assembly, but varies from one suspension to another. Figure6 shows simultaneous measurement of full-body capacitance and LDV during loading for LP-HAS immediately before fully loading onto the disk surface. Capacitance measurement shows some oscillation about 2 ms before a step-like jump is observed. Note that the average height of the slider during these oscillations is on the order of a few microns. At this height, the suspension preload (not the preload between the flexure and load-dome) is still supported by the ramp. The LDV measurement shows that the slider actually contacts the disk and bounces on-and-off the disk oscillating at the same frequency as that of the measurement made with the capacitance meter. The slider then settles into what appears to be a loaded position, but the capacitance measurement shows that the slider has not fully reached the nominal flying height position—the capacitance measurement is slightly lower in magnitude than the final value. It takes another 4 ms or so before the slider finally loads fully into the nominal flying height. Surprisingly, LDV is also able to measure this latter process as well. The corresponding arm mounted acoustic emission measurement shows slider/disk contact Fig. 4 Top LDV measurement of the loading motion of the trailing edge of the slider for a system with normal preload between the load-dome and flexure. Bottom Full-body capacitance measurement, which shows a single sharp transition as the slider loads onto the diskverifying the LDV and capacitance measurements of slider–disk contact (Fig.7). The sligh5t delay in the AE signal is due to the fact the sensor is mounted at the suspension mount point, which is far removed from the location of the contact point. Another example of the loading process is shown in Fig.8 showing a similar behavior.The bounce followed by oscillations and slow settling into the nominal flying height has not been reported before. The reason for the observed deviation is due to the lack of preload between the slider and load-dome. During the loading process, the lack of preload results in oscillation of the slider as seen in Fig.5. This oscillation results in the slider corner contacting the disk multiple times when the slider comes close (on the order of a few microns) to the disk. Then, as the slider comes even closer to the disk, the negative suction force pulls the slider towards the disk separating the load-dome from the flexure. Under certain circumstances, the slider actually can also contact the disk during this phase of the process while the load/unload tab is still sliding on the ramp and the slider is a fraction of a micron away from the disk (Fig.9). This phenomenon is easy to see using a high-speed camera. A set of images captured with a high-speed camera for LP-HSA case is shown in Fig.10. It clearly shows load-dome separation from the flexure resulting in a partial loading on the disk while the load/unload tab is still on the ramp. In this particular case, we were unable to capture the slider/disk contacts using the high-speed camera. The initial phase of the measurements shown in Fig.5 is quite repeatable, i.e. the initial oscillation can be observed every time. However, the slider disk contact is not fully repeatable since this depends on many other parameters, such as, the vertical velocity of the disk at the time of loading and random excitation of the system due to airflow and mechanical vibrations. The suction force that causes the slider to jump towards 6the disk is due to a negative pressure force resulting from negative PSA of the slider relative to the disk surface. The relative PSA is usually negative while the suspension is on the ramp although the absolute PSA may be positive. As the suspension moves across the ramp, the relative PSA constantly changes ultimately reaching the absolute PSA value immediately before loading. During the time the relative PSA is negative, the negative pressure force will try to pull the slider towards the disk. If the sum of the flexure stiffness and the preload between the flexure and load-dome is less than this negative force exerted on the slider, the slider will move towards the disk at speeds higher than the desired speed separating the flexure from the load-dome. Furthermore, since the load-dome is separated from the flexure seen in Fig.10, there is no preload on the slider to push the slider towards the disk. As the gap between the load-dome and flexure closes and the preload of the suspension is transferred from the ramp to the slider, the slider is finally pushed into the nominal flying height as indicated by the final small increase in the capacitance and decrease in height as shown in the LDV measurements(Figs.4, 5, 7, 8). 4 Summary and conclusion Recent articles on load/unload have mainly dealt with the unloading process since the unload dynamics of negative pressure slider reveals an interesting behavior unlike the loading process. However, much more attention to detail is required for the loading process than the unloading process, since the affinity to cause disk damage is much greater during the former process than the latter. In this 7paper, we show that a small deviation in the design point of the preload between the load-dome and flexure can lead to adverse loadingprocesses resulting in an undesirable number of slider/ disk contacts.We show that if the preload between the load-dome and flexure is too low, the slider can oscillate and contact the disk multiple times even when the slider is a few microns away from the disk. Furthermore, we show that the slider can also be pulled down towards the disk completely separating the load-dome from the flexure assembly. This results in slider contacting the disk at an uncontrolled speed that can also lead to disk damage.The separation occurs while the suspension is still on the ramp, and thus there is no preload on the slider following the separation. This lack of preload allows the slider to fly at high flying heights until the gap between the flexure and load-dome closes. Hence, a prudent design of the suspension assembly is required to ensure that the combination of the flexure stiffness and the preload between the load-dome and suspension will be significant enough to defeat the negative pressure force keeping the load-dome attached to the suspension at all times and to suppress slider oscillations before loading.8ReferencesBogy DB, Zeng QH (2000) Design and operating conditions for reliable load/unload systems. Tribol Int 33(5–6):357–366Hua W, Liu B, Sheng G, Li J (2001) Further studies of unload process with a 9D model. IEEE Trans Magn 37(4):1855–1858Liu B, Zhu LY (2001) Experimental study on head disk interaction in ramp loading process. IEEE Trans Magn 37(4):1809–1813Suk M, Gillis D (1998) Effect of slider burnish on disk damage during dynamic load/unload. ASME J Tribol 120(2):332–338Suk M, Ruiz O, Gillis D (2004) Load/unload systems with multiple flying heights (presented at the 2002 ASME/STLE international tribology conference, Cancu n, Mexico). ASME J Tribol 126(2):367–371Zeng QH, Bogy DB (2000) Effects of certain design parameters on load/unload performance. IEEE Trans Magn 36(1): 140–1479M. Suk D. Gillis影響機(jī)械設(shè)計暫停動態(tài)加載過程收稿： 2003 年 7月 2 /接受： 2004 年 2月 24日/網(wǎng)上公布： 2005 年 8月 3 _斯普林格 2005年摘要：設(shè)計一個加載/卸載系統(tǒng)中使用的硬盤驅(qū)動器，必要時需要注意，確保在裝貨和卸貨過程不會損害滑塊碟片的表面。因?yàn)?，在設(shè)計點(diǎn)的預(yù)負(fù)荷之間的穹頂和彎曲的一個小偏差可能會導(dǎo)致不良進(jìn)程載入，造成一些滑塊/磁盤不利的接觸。在這項研究中，我們發(fā)現(xiàn)，如果預(yù)之間的負(fù)載圓頂和彎曲太低，滑塊的擺動可能會造成角落的滑塊接觸磁盤過多，使滑桿遠(yuǎn)離磁盤幾微米。此外，滑塊可吸入下跌對磁盤造成了完全分離的負(fù)載圓頂，使柔性裝配導(dǎo)致失控的負(fù)載條件。這種分離的情況仍然暫停在坡道，因此沒有施加預(yù)壓的滑塊立即分離。因此，滑塊蒼蠅在飛行高度高于設(shè)計點(diǎn)，直到負(fù)載圓頂和彎曲之間的差距為零。因此，必須認(rèn)真地暫停旨在制止滑塊振蕩，并確保不單獨(dú)在負(fù)荷盤加載過程。1導(dǎo)言 其中一項要求設(shè)計 一個加載/卸載系統(tǒng)中使用的硬盤 驅(qū)動器是確保在裝貨和卸貨過程滑 塊不會損害碟片表面。因?yàn)檫@是難 以避免滑塊/磁盤接觸的全部內(nèi)容， 因?yàn)?，該系統(tǒng)是為了盡量多的減少 滑桿/磁盤接觸事件和接觸的后果 的發(fā)生?；瑝K/磁盤接觸發(fā)生接觸 的可能性取決于加載速度，硬盤速 度，靜態(tài)的態(tài)度滑塊，空氣軸承粗 糙度，滑塊幾何等。例如滑塊大曲率半徑的彎道可以消除磁盤損害，降低接觸應(yīng)力之間的滑塊和磁盤表面（Suk and Gillis 1998） 。許多最近的研究認(rèn)為，影響暫停與限制器和空氣軸承設(shè)計的魯棒性和裝卸過程有關(guān)（Bogy and Zeng 2000; Hua et al. 2001; Liu and Zhu 2001; Zeng and Bogy 2000） 。不過，這些研究主要集中在卸貨的過程，因?yàn)檫@部分序列通常揭示有趣的動態(tài)過程的影響和負(fù)壓空氣軸設(shè)計。負(fù)壓區(qū)域空氣軸抗拒卸貨行動導(dǎo)致的潛在能量儲存在彎曲和懸掛裝備中.當(dāng)滑塊終于脫離磁盤，勢能釋放，滑塊振蕩劇烈。另一方面，10為合理的設(shè)計系統(tǒng)，加載過程并沒有表現(xiàn)出這樣的行為。因此，大多數(shù)國家都已經(jīng)在主要調(diào)查卸載進(jìn)程給予粗略注意更重要的加載過程。大多設(shè)計師的加載/卸載系統(tǒng)會發(fā)現(xiàn)，加載過程可以比卸載過程更麻煩，。除了可能造成損害的磁盤，其他問題都可以遇到的加載過程。例如，在某些情況下，滑塊可能永遠(yuǎn)無法達(dá)到負(fù)荷的設(shè)計飛行高度，而是在飛行高度負(fù)荷的命令 1流明（Suk et al. 2004） 。在本文中，我們顯示一個小偏差的機(jī)械設(shè)計的彎曲/暫停大會可以增加概率滑塊/磁盤接觸，可能導(dǎo)致大量的磁盤接觸單一負(fù)載周期。具體來說，我們表明，懸掛系統(tǒng)，低預(yù)彎曲之間和負(fù)荷盤可能導(dǎo)致裝載的滑塊不受控制靜態(tài)的態(tài)度和速度。相關(guān)問題這方面的設(shè)計可以很容易地確定測量全身電容在加載過程。2描述的實(shí)驗(yàn)滑塊載入中動態(tài)進(jìn)行了研究用激光多普勒測振儀（激光多普勒） ， 62千赫的幀速率的高速攝像頭，全身電容。實(shí)驗(yàn)裝置包括一個標(biāo)準(zhǔn)的加載/卸載測試。電容米措施全身電容之間的滑塊和磁盤而滑塊裝上磁盤，一個類似于用在（Suk et al. 2004） ?；瑝K裝上和卸下磁盤使用移動坡道，同時保持滑塊/暫停固定的外徑地區(qū)的磁盤。垂直運(yùn)動的后緣的滑桿是用激光多普勒測量。所有的測試使用 84毫米玻璃磁盤和負(fù)壓雪橇型滑塊與磁11盤旋轉(zhuǎn) 10 krpm進(jìn)行。球場靜態(tài)態(tài)度（簡稱 PSA ）的滑塊用于實(shí)驗(yàn)是介于 1和 2_ 。顯示效果缺乏預(yù)彎曲之間和負(fù)載圓頂，我們選擇兩個暫停集會是基本相同，除預(yù)裝。由于不同程度的預(yù)是難以衡量的，只有存在大量不同的是核實(shí)。要做到這一點(diǎn)，我們掛載頭部懸掛大會正常預(yù)（ NPHSA ）進(jìn)入坡道。一個小型的重量，這是足以造成負(fù)載穹頂脫離彎曲，然后附在彎曲。分離的數(shù)量來衡量一個適當(dāng)?shù)奈恢?CCD相機(jī)。類似的測量是用于頭部暫停低大會預(yù)裝（唱片白蛋白）。圖2和圖3顯示的光學(xué)圖像的負(fù)載圓頂和彎曲采取相同的條件下為 NP一人血清白蛋白和 LPHSA分別。更大的負(fù)載穹頂脫離彎曲是觀察唱片白蛋白比 NP一人血清白蛋白，確認(rèn)唱片白蛋白低預(yù)比 NPHSA 。 3結(jié)果與討論負(fù)壓滑塊滑塊負(fù)載到磁盤，然后按照跳動磁盤預(yù)期。底部的陰謀是在圖 3的相應(yīng)全身電容測量，這表明在一個單一的跳轉(zhuǎn)電容此刻滑塊負(fù)載到磁盤。類似的測量唱片- HSA的是顯示在圖 5 。在這種情況下，將滑塊振蕩在裝貨前到磁盤的情況不同，具有較高的預(yù)壓荷載圓頂和彎曲。此外，滑蓋的垂直加載速度突然增加時，滑塊約為 50流明遠(yuǎn)離磁盤。與此相關(guān)的突然增加，速度，電容測量顯示多個急劇轉(zhuǎn)變。繼過渡，電容不能達(dá)到的最高值為另一個 1毫秒左右。這些意見表明一個問題，但很難確定確切的動態(tài)，由于低測量帶寬。更高分辨率的測量表明，滑塊接觸磁盤多次（圖6 ） ，注意，這完全行為不會發(fā)生，每暫停大會，但不同暫停到另一個。圖6顯示同步測量全身電容和激光多普勒在裝貨唱片，已立即在完全加載到磁盤的表面。電容測量表明一些振蕩約2毫秒之前的一個步驟樣跳轉(zhuǎn)得到遵守。請注意，平均身高滑塊在這些振蕩是對秩序的幾個微米。在這一高度，暫停預(yù)（而不是預(yù)之間的柔性和負(fù)載圓頂）仍然是支持的坡道。在激光多普勒測量結(jié)果表明，滑塊實(shí)際接觸磁盤和彈跳上和12從磁盤振蕩在相同的頻率，在測量與電容米?；瑝K然后解決什么似乎是一個加載的位置，但電容測量表明，該滑塊沒有完全達(dá)到了標(biāo)稱飛行高度位置的電容測量略低規(guī)模比最后的值。另需4 毫秒或之前滑塊最后負(fù)荷充分融入名義飛行高度。令人驚訝的是，激光多普勒也是能夠衡量后者的進(jìn)程。相應(yīng)的胳膊安裝聲發(fā)射測量表明滑塊/磁盤聯(lián)絡(luò)圖。 4頂級激光多普勒測量負(fù)荷運(yùn)動后緣的滑塊的系統(tǒng)之間的正常預(yù)負(fù)荷圓頂和彎曲。底部全身電容測量，這顯示出急劇轉(zhuǎn)型滑塊負(fù)載到磁盤核查激光多普勒和電容測量滑塊磁盤接觸（圖7 ） 。稍有延誤，聲發(fā)射信號的原因是傳感器安裝在減震器點(diǎn)，這是遠(yuǎn)離的位置，聯(lián)絡(luò)點(diǎn)。另一個例子是在加載過程中顯示圖8顯示了類似的行為。隨后的反彈振蕩和緩慢的名義解決飛行高度還沒有報告過。的原因，觀察偏差是由于缺乏預(yù)之間的滑塊和負(fù)載圓頂。在裝載過程中，缺乏預(yù)結(jié)果振蕩滑塊看到的圖。這種振蕩的結(jié)果滑塊角落接觸磁盤多次當(dāng)滑塊接近（在命令幾微米）的磁盤。然后，將滑塊來更接近盤，負(fù)吸力量拉動滑塊對分離的磁盤負(fù)載圓頂從彎曲。在某些情況下，將滑塊還可以聯(lián)系實(shí)際的磁盤在此階段的進(jìn)程，而加載/卸載選項卡仍然是滑動的坡道和滑塊的一小部分微米遠(yuǎn)離磁盤（圖 9 ） 。這種現(xiàn)象很容易看到使用高速攝像頭。一組拍攝 與一個高速攝像機(jī)唱片- HSA的案件中顯示圖。這清13楚地表明負(fù)載穹頂脫離彎曲造成了部分負(fù)荷，同時在磁盤上的加載/卸載選項卡仍然在坡道。在這種情況下，我們無法捕捉滑塊/磁盤接觸，利用高速攝像頭。初期階段的測量顯示在圖 5是相當(dāng)重復(fù)的，即 初始振蕩可以看到每一次。然而，滑塊磁盤聯(lián)系不完全重復(fù)的，因?yàn)檫@取決于許多其他參數(shù)，如垂直速度的磁盤在裝載時和隨機(jī)激勵的制度，由于氣流和機(jī)械振動。吸力的力量，使滑塊跳轉(zhuǎn)對磁盤是由于負(fù)面壓力變壓吸附造成負(fù)面的滑塊相對于硬盤的表面。相對港務(wù)集團(tuán)通常是消極的，而暫停的坡道上雖然絕對港務(wù)集團(tuán)可能是積極的。作為全國暫停行動的坡道，相對港務(wù)集團(tuán)不斷變化最終達(dá)到絕對港務(wù)集團(tuán)價值立即在裝貨前。在時間的相對港務(wù)集團(tuán)是否定的，消極的壓力將嘗試?yán)瑝K對磁盤。如果總和彎曲剛度和預(yù)緊力之間的柔性和負(fù)載圓頂不到這種消極力量施加的滑塊，滑塊將走向磁盤的速度高于預(yù)期的速度分離彎曲的負(fù)載圓頂。此外，由于負(fù)載圓頂是分開彎曲，圖中可以看出，沒有預(yù)裝的滑桿推動滑塊實(shí)現(xiàn)磁盤。隨著之間的差距負(fù)載圓頂和彎曲關(guān)閉和預(yù)緊暫停從坡道滑塊，滑塊終于被推到名義飛行高度所示的最后略有增加電容和降低高度所顯示的 LDV測量 （圖.4 ， 5 ， 7 ， 8 ） 。 4摘要和結(jié)論最近文章加載/卸載主要處理過程，因?yàn)樾遁d的卸載動態(tài)負(fù)壓滑塊揭示了一個有趣的行為不同，加載過程。然而，更多 注重細(xì)節(jié)是需要加載過程比卸貨過程中，由于親造成磁盤損害要大得多前進(jìn)程在比后者。在本文中，我們表明，一個小偏差在設(shè)計點(diǎn)的預(yù)負(fù)荷之間的穹頂和彎曲可能會導(dǎo)致不良載入中 進(jìn)程造成不良一些滑桿/磁盤接觸。 我們發(fā)現(xiàn)，如果預(yù)之間的負(fù)載圓頂和彎曲太低，滑塊可以擺動和接觸磁盤多次即使滑桿是幾微米遠(yuǎn)離磁盤。此外，我們表明，滑塊也可以推倒對磁盤完全分開的負(fù)載圓頂從彎曲大會。這樣的結(jié)果是滑塊接觸磁盤上失控速度也可能導(dǎo)致硬盤損壞。14分離時發(fā)生暫停仍然在坡道，因此沒有預(yù)裝以下的滑塊分離。這種缺乏預(yù)使滑塊飛行，飛行高度高，直到之間的差距彎曲和負(fù)載圓頂關(guān)閉。因此，謹(jǐn)慎的設(shè)計暫停大會必須確保該組合彎曲剛度和預(yù)緊力之間的負(fù)載圓頂和暫停將是很大的，足以承載負(fù)面壓力保持負(fù)載圓頂重視暫停任何時候，以制止在裝貨前滑塊振蕩。15參考資料Bogy DB, Zeng QH (2000) Design and operating conditions for reliable load/unload systems. Tribol Int 33(5–6):357–366Hua W, Liu B, Sheng G, Li J (2001) Further studies of unload process with a 9D model. IEEE Trans Magn 37(4):1855–1858Liu B, Zhu LY (2001) Experimental study on head disk interaction in ramp loading process. IEEE Trans Magn 37(4):1809–1813Suk M, Gillis D (1998) Effect of slider burnish on disk damage during dynamic load/unload. ASME J Tribol 120(2):332–338Suk M, Ruiz O, Gillis D (2004) Load/unload systems with multiple flying heights (presented at the 2002 ASME/STLE international tribology conference, Cancu n, Mexico). ASME J Tribol 126(2):367–371Zeng QH, Bogy DB (2000) Effects of certain design parameters on load/unload performance. IEEE Trans Magn 36(1): 140–147