Demand is increasing daily for large data storage systems that are useful for applications in spacecraft, space satellites, and space robots, which are all exposed to radiation-rich space environment. As candidates for use in space embedded systems, holographic storage systems are promising because they can easily provided the demanded large-storage capability. Particularly, holographic storage systems, which have no rotation mechanism, are demanded because they are virtually maintenance-free. Although a holographic memory itself is an extremely robust device even in a space radiation environment, its associated lasers and drive circuit devices are vulnerable. Such vulnerabilities sometimes engendered severe problems that prevent reading of all contents of the holographic memory, which is a turn-off failure mode of a laser array. This paper therefore presents a proposal for a recovery method for the turn-off failure mode of a laser array on a holographic storage system, and describes results of an experimental demonstration.
© 2011 OSA
Demand is increasing daily for large storage systems for use in spacecraft, space satellites, and space robots, which must all function in radiation-rich space environments . Holographic storage systems are excellent candidates for such embedded systems intended for use in space since three-dimensional holographic memories can easily provide large storage capacity. The density potential reaches V/λ3, where V denotes the recording volume and λ is the wavelength of the recording light source . For example, photorefractive crystal holographic memories LiNb03 can store 100 bits/μm2  and can achieve 100 Gbit/in2 . The area densities are superior to even the latest 32 nm process dynamic random access memory (DRAM) cell with area of 0.039 μm2 .
In addition, a holographic memory is a defect-tolerant device [6, 7]. Since each bit of information is generated from the entirety of a holographic memory, its sensitivity to material defects is extremely low. Therefore, although the degradation of holographic memory material is accelerated under a radiation-rich space environment, the deteriorated holographic memory can generate correct information reliably for long periods. In contrast, a silicon memory device is sensitive to high-energy charged particles because a single bit of information is stored on a single transistor circuit. Therefore, the stored information is lost if the transistor is degraded [8–10]. Consequently, in addition to their advantage of large storage, holographic memories have superior fault tolerance to that of current silicon memory devices in terms of protecting stored information.
On the other hand, optical discs such as Compact Disc Read Only Memory (CD-ROM), Digital Versatile Disc (DVD), and blu-ray discs have been developed successively. They are used widely in current computer systems [11, 12]. Disc type holographic memories such as the Holographic Versatile Disc (HVD), have also been developed as next-generation optical discs [13–16]. Nevertheless, since space systems must be maintenance-free, it is better that a mechanism that is rotated by a motor be removed. To date, maintenance-free holographic storage systems without a rotation mechanism have been developed. For example, an acousto-optic deflector addressing method , a twisted-nematic liquid crystal addressing method , a microelectromechanical systems (MEMS) addressing method , and a surface-emitting laser diode array (SELDA) addressing method  have been proposed. Among the studies which do not use a rotating mechanism, the fastest addressing method is the SELDA laser diode addressing method.
However, when using the SELDA or the laser diode addressing method, although a holographic memory itself is an extremely robust device in space radiation environments, the lasers and its drive circuit devices are vulnerable. Such devices can sometimes cause severe problems that prevent the reading of all contents of the holographic memory. This is a turn-off failure mode of a laser array. In a space system, such a crucial failure mode must be obviated through countermeasures of some type. However, to date, although the failure modes of laser themselves have been analyzed in many studies [21, 22], recovery methods for a driver melt failure mode have never been reported. This paper therefore presents a proposal of a recovery method of a turn-off failure mode of a laser array on a holographic storage system with SELDA or laser diodes, along with results obtained from an experimental demonstration.
2. Defect tolerance of a holographic storage system with a laser array addressing mechanism
2.1. Overall construction
A holographic storage system with a laser array addressing mechanism comprises a photodiode array, a holographic memory, and a laser diode array, as shown in Fig. 1 . On the earth, the single output of the holographic storage system is connected to a single processor system. In contrast, in a space radiation environment, triple-module redundancy (TMR) [23, 24] is always used to enable correction of a SEU on flip-flops, latches, and memory, and a single event transient (SET) on operations of logic circuits [8, 9, 25]. With that construction, the three independent outputs of the holographic storage system are connected to three processor systems. Therefore, three photodiode arrays are implemented. A laser array is mounted on the left side of the holographic memory for use in addressing the data in the holographic memory. One laser addresses one page data that indicates a two-dimensional binary image. Turning one laser on, one page data can be read out from the holographic memory. Finally, the page data is received on photodiode arrays and is sent to processor systems. Moreover, holographic memory system offers the important advantage of maintenance-free operation since the holographic memory system has no mechanical parts such as a rotation mechanism. The holographic memory system has another advantage: The probability of losing data is much lower than that of current silicon memory devices.
2.2. Holographic memory part
Holographic memories are well known to have high defect tolerance [6,7]. Since each bit of the contents can be generated from the entire holographic memory, damage to some fraction rarely affects its diffraction pattern or its content. Even if a holographic memory device includes small defect areas, holographic memories can record contents correctly and can then generate contents correctly. Such mechanisms can be regarded as those for which majority voting for each content bit is executed from an infinite number of diffraction beams. In a semiconductor memory, single-bit information is stored in a single-bit memory circuit. In contrast, in a holographic memory, a single bit of a reconfiguration context is stored in the entire holographic memory. Therefore, the holographic memory’s information is robust. In contrast, in the semiconductor memory, the defect of a transistor always erases information of a single bit or multiple bits. Earlier reports have described experiments demonstrating that a holographic memory is robust . In such experiments, even if 1000 impulse noises and 10% Gaussian noise were applied to a holographic memory, reading all contents was successful. Therefore, defects of a holographic memory device are beyond consideration.
2.3. Laser array part
In this system, a laser array is a fundamental component used in addressing numerous contents stored on a holographic memory. Each laser addresses one page data on the holographic memory. Although content information stored on a holographic memory is robust, if a laser becomes defective, then reading of the corresponding content might become impossible. Since lasers and their driver semiconductor circuits are important, but are space-radiation-vulnerable components, a main concern of the holographic storage system is the failure of vulnerable semiconductor components.
Turn-on failure mode
A laser might have a turn-on failure resulting from the breakdown of laser or its driver circuits. Such a situation dictates that a failed laser can not turn on. However, the recovery method is simple. The holographic memory system can store numerous contents. Many lasers are used for switching the contents. Therefore, by storing contents redundantly on a holographic memory in advance, once trouble occurs due to a laser source defects or its driver defects, the failure is recoverable using other redundant contents. The holographic memory system accommodates the turn-on failure mode for lasers.
Turn-off failure mode
Furthermore, a laser might be sent into a turn-off failure mode by melting of the driver transistors used for the laser. In this case, the failed laser remains constantly turned on. The turn-off failure mode is caused mainly by trouble that occurs with driver transistors. Although the probability of such a turn-off failure occurrence is extremely low, this trouble level is slightly higher than that of the turn-on defect mode. If one laser has a turn-off failure mode and is therefore turned on constantly, then the corresponding holographic memory information is superimposed constantly over other content that the system might be required to read. Both contents, where one is the desired content and the other is an unexpected failure of a content, are linearly superimposed on a photodiode array. Such effect shows an OR-operation of the two contents. Therefore, the correct content reading procedure cannot be executed by the undesired content. Therefore, the turn-off failure mode might render all content-reading procedures impossible. In a space system, the turn-off failure mode is categorized as a fatal event.
3. Recovery method for the turn-off failure mode
The easiest way to recover the turn-off failure mode is to implement a series power off circuit. In this case, if the laser driver circuit is melted down, then the main power for a laser array has only to be cut. However, since the power off circuit itself might be melted down simultaneously, a series transistor or a relay is necessary to hold the excessive power. It is not a sufficient solution.
This paper therefore presents a proposal of a recovery method for a turn-off failure mode of a laser array. A circuit diagram of an array of laser driver circuits is portrayed in Fig. 2. In one laser array, many identical arrays of laser driver circuits are implemented. Each driver array has a single common current limitation circuit because, in the laser array, only a single laser turns on in a normal case. In Fig. 2, although the current limitation circuit is shown as a simple register, an actual implementation might include a bias circuit and a temperature correction circuit. All lasers and their driver transistors are connected to the common current limitation circuit. Here, it is assumed that a turn-off failure occurs on a laser in a driver array. In this case, the laser is constantly turned on. Subsequently, under an entire holographic memory system, one page data corresponding to the failed laser is constantly irradiated onto a photodiode array so that reading another page data becomes impossible, as explained above. Invariably, such trouble results from circumstances in which an NPN driver transistor (for example X1) is melted somehow. Our proposed method to recover from the problem is a simple method that turns on all other lasers in the driver array. As presented in Fig. 2, a driver array has many lasers. Therefore, if all lasers are turned on, then the current and forward voltage of the lasers drops considerably. The light intensities are also decreased dramatically. Therefore, using this method, although the use of all lasers in the single driver array becomes impossible when a turn-off laser failure arises, the other driver arrays can support the addressing of page data since a laser array has many driver arrays. A suitable number N of lasers included in a driver array is estimated according to the following equation.
4. Experimental system
To demonstrate the turn-off failed laser recovery method, a 340 bit × 16 page simple holographic storage system was constructed. A block diagram and photographs of the experimental system are portrayed in Figs. 3 and 4. The holographic storage system was constructed using 16 infrared laser diodes (DL-7140-201S; Tottori Sanyo Electric Co., Ltd.), a liquid crystal spatial light modulator (LC–SLM) (L3P07X-31G00; Seiko Epson Corp.) as a holographic memory, and a CCD camera (XC-ST30; Sony Corp.). In an actual system, a volume-type high-resolution holographic memory must be implemented to achieve a large storage capacity [3, 4], here, but an LC–SLM was used as the two-dimensional holographic memory for this study because of its programmability. The LC–SLM panel has 1,024 × 768 pixels, each of 14 × 14 μm2. The LC–SLM, which is connected to an evaluation board (L3B07X-E10A; Seiko Epson Corp.), is a transmission-type spatial light modulator and a 90° twisted nematic device. The board’s video input is connected to the external display terminal of a personal computer. Programming for the LC–SLM is executed by displaying 16 holographic memory patterns with 256 gradation levels on the personal computer display. The display region on the LC–SLM is divided into 16 regions, each of which stores one page data, which includes 340 bits. One page data is addressed by the infrared laser diode and is read out by the CCD camera. The pixel size and resolution of the CCD camera are 6.35 μm× 7.4 μm and 752 × 485 pixels. To receive 340 bits, the resolution of the CCD camera is sufficient. The output of the CCD camera is also connected to the personal computer through a video capture board. The CCD camera was placed 70 mm distant from the LC–SLM. The distance between infrared laser diodes and LC–SLM is 100 mm.
4.1. Liquid-crystal holographic memory
Here, a liquid-crystal spatial light modulator is used as a holographic memory. The holographic memory pattern is determined as follows. Each laser source is assumed as a small dot with a diverging beam. The laser’s divergent reference wave propagates into the holographic plane. The holographic medium comprises rectangular pixels on the x1 – y1 holographic plane. The pixels are modulated as analog values. On the other hand, a content pattern is made up of rectangular pixels on the x2 – y2 object plane, which can be modulated as either on or off. The intensity distribution of a holographic medium is calculable using the following equations.Figure 5 shows 16 calculated fringe patterns corresponding to 16 page contents programmed onto different places of an LC–SLM. The number of pixels of each recording area including one page data are about 200 × 150. The interval between recording areas is about 50 pixels.
4.2. Laser array design
The laser array consists of four driver circuit arrays. Each driver circuit array serves four lasers. Therefore, in a complete arrangement, 16 lasers are implemented onto a jig at regular intervals of 10 mm, as shown in Fig. 6. The lasers are infrared laser diodes (DL-7140-201S; Tottori Sanyo Electric Co., Ltd.). The maximum power and wavelength of each laser are, respectively, 80 mW and 785 nm. A current limiting circuit is constructed using a 15 Ω register. Through the common register, 3.3 V is applied to the laser anode. The laser cathodes are connected to an NPN transistor (2SC3792; Sanyo Electronic Co., Ltd.). Addressing of the lasers is controlled by the transistors. Here, the laser power was designed as about 40 mW. At that time, the operating current was measured as about 86 mA (= Ityp). The threshold current is about 35 mA (= Ith). Therefore, following Eq. (1), at least, N must be equal to or greater than 3. In the laser array, Nspare can be regarded as equal to one so that each driver array is able to recover from the turn-off failure mode even if a single laser in the driver array is in turn-on failure mode. If two non-turn-on failed lasers turn on in addition to a turn-off failed laser, then the operating current of each laser becomes sufficiently smaller than the threshold current. At that time, the light intensity of a turn-off failed laser can be decreased dramatically. Furthermore, the remaining 12 lasers can address page data of the holographic memory.
5. Experimental results
Using the experimental system explained previously, the usefulness of the recovery method was confirmed. In this experiment, two page data were implemented as shown in Figs. 7(a) and 7(b). Figures 7(a) and 7(b) respectively portray the desired page data and unexpected page data in the first experiment and show desired page data and unexpected page data in the second experiment. Both holographic memory patterns were calculated using Eqs. 2 and 3. The holographic memory pattern presented in Fig. 7(a) was implemented onto the holographic regions corresponding to No. 1, No. 6, No. 7, No. 8, and No. 9 lasers. The holographic memory pattern presented in Fig. 7(b) was implemented onto the holographic regions corresponding to No. 2, No. 3, No. 4, and No. 5 lasers, as shown in Fig. 5. In the first experiment, the desired data shown in Fig. 7(a) was read by activating the No. 1 laser, as presented in Fig. 8(a). This reading operation is a normal case: The desired page data were read correctly by a CCD camera. At that time, the power of the No. 1 laser was measured as 39.8 mW, as shown in Table 1(a).
Here, it is assumed that driver transistors of the No. 2 laser were broken by the effect of high-energy charged particles such that the No. 2 laser was constantly turned on. In this case, when the desired page data corresponding to the No. 1 laser is read, the second unexpected page data corresponding to the No. 2 laser is superimposed on the first desired page data so that the desired page data cannot be read out correctly. Figure 8(b) shows the situation in which both the No. 1 laser for the desired page data and No. 2 laser for the unexpected page data turn on. In this case, the power of No. 2 laser was 35.3 mW, as presented in Table 1(b).
Here, a recovery operation was started. The No. 2, No. 3, No. 4, and No. 5 lasers are connected to a same current limitation circuit. If the No. 3 laser turns on, as shown in Fig. 8(c), then the power of the failed No. 2 laser and the No. 3 recovery laser can be reduced to 4.79 mW and 5.35 mW, respectively, as shown in Table 1(c). Moreover, if the No. 3 laser and No. 4 laser turn on, then the three lasers turn on in the same driver array. This is the designed recovery number of lasers. In this case, the respective powers of the failure No. 2 laser, No. 3 recovery laser, and No. 4 recovery laser can be reduced to 64.8, 67.5, and 82.5 μW. The first content can be read out correctly. Of course, when the No. 3 laser, No. 4 laser, and No. 5 laser turn on, then the power of the failed No. 2 laser, No. 3 recovery laser, No. 4 recovery laser, and No. 5 recovery laser can be reduced further, respectively, to 42.3, 44.1, 51.6, and 47.7 μW. In this case also, the first contents can be read out correctly. Here, the power difference between the No. 1 and No. 2 lasers in case (b) was less than the power difference between the No. 2 and the No. 3 in case (c), and less than the power difference between the No. 2, No. 3, and No. 4 in the subthreshold case (d), and so on. Such a large power difference is mainly caused by driving all the lasers in parallel using a single current- limitation circuit. Always, the power of a laser is proportional to its current. Therefore, a current limitation circuit can control a laser to a similar laser power almost entirely without depending on the laser’s forward voltage – forward current characteristic variation, although each laser invariably has a production variation. However, when the recovery lasers turn on, all lasers are driven by a single current limitation circuit. In this case, since all lasers’ forward voltages are fixed as the same one, the laser’s production variation is readily apparent. However, even if such variation appears, the power of all lasers is sufficiently reduced. Therefore, such difference can be regarded as beyond consideration.
In addition to the first experiment, a second recovery operation has been confirmed. In the second experiment, a laser corresponding to the desired data shown in Fig. 7(b) was assigned to the No. 2 laser, as presented in Fig. 9(a). In this normal reading operation, the power of the No. 2 laser was measured as 36.2 mW, as shown in Table 2(a). Here, as in the first experiment, it is assumed that driver transistors of the No. 6 laser were damaged by the effects of high-energy charged particles, and that it malfunctions such that the No. 6 laser is constantly turned on. In this case, when the desired page data corresponding to the No. 2 laser are read, the second unexpected page data corresponding to the No. 6 laser are superimposed on the first desired page data so that the desired page data cannot be read out correctly. Here, a recovery operation was started. In this case, the No. 6, No. 7, No. 8, and No. 9 lasers are turned on, which are connected to a same current limitation circuit. As the number of recovery lasers increases, the respective power levels of the failed No. 6 laser, No. 7 recovery laser, and No. 8 recovery laser, and No. 9 recovery laser were all decreased drastically. As in the first experiment, if two or more recovery lasers turn on in addition to the failed No. 6 laser, then the power of the failed No. 6 laser and the other recovery lasers No. 7, No. 8, and No. 9 can be decreased drastically as shown in Figs. 9(d) and 9(e).
Using this method, even if a turn-off failure occurs on a laser array, recovery is possible with the loss of only a few data pages.
This paper has presented a proposal for a method of recovery from a turn-off failure mode of a laser array on a holographic storage system with SELDA or laser diodes. In addition, the experiments described in this paper have demonstrated that a turn-on failure mode of a laser array is recoverable correctly by turning another laser on, with the loss of only a few page data. This failure recovery method is particularly useful for maintenance-free holographic storage systems designed for use in a space environment.
This research was supported by the Accelerating Utilization of University Intellectual Property Program.
References and links
1. H. Kuninaka and J. Kawaguchi, “Lessons learned from round trip of Hayabusa asteroid explorer in deep space,” IEEE Aerospace Conference , 1–8 (2011).
2. P. J. van Heerden, “Theory of optical information storage in solids,” Appl. Opt. 2, 393–400 (1963). [CrossRef]
3. A. Pu and D. Psaltis, “Holographic data storage with 100 bits/μm2 density,” Optical Data Storage Topical Meeting Conference Digest , 48–49 (1997). [CrossRef]
4. G. W. Burr, C. M. Jefferson, H. Coufal, C. Gollasch, M. Jurich, J. A. Hoffnagle, R. Macfarlane, and R. M. Shelby, “Volume holographic data storage at an areal density of 100 Gbit/in2,” Conference on Lasers and Electro-Optics , 188–189 (2000).
5. N. Butt, K. Mcstay, A. Cestero, H. Ho, W. Kong, S. Fang, R. Krishnan, B. Khan, A. Tessier, W. Davies, S. Lee, Y. Zhang, J. Johnson, S. Rombawa, R. Takalkar, A. Blauberg, K. V. Hawkins, J. Liu, S. Rosenblatt, P. Goyal, S. Gupta, J. Ervin, Z. Li, S. Galis, J. Barth, M. Yin, T. Weaver, J. H. Li, S. Narasimha, P. Parries, W. K. Henson, N. Robson, T. Kirihata, M. Chudzik, E. Maciejewski, P. Agnello, S. Stiffler, and S. S. Iyer, “A 0.039 μm2 High Performance eDRAM Cell based on 32nm High-K/Metal SOI Technology,” IEEE International Electron Devices Meeting , 27.5.1 – 27.5.4 (2010).
6. M. Toishi, A. Okamoto, S. Honma, and M. Bunsen, “Fault-tolerant holographic memory with two photrefractive crystals,” Pacific Rim Conference on Lasers and Electro-Optics , 2, 160–161 (2001).
7. H. J. Coufal, D. Psaltis, and G. T. Sincerbox, “Holographic Data Storage,” Springer-Verlag, 7 (2000).
8. S. Redant, R. Marec, L. Baguena, E. Liegeon, J. Soucarre, B. Van Thielen, G. Beeckman, P. Ribeiro, A. Fernandez-Leon, and B. Glass, “Radiation Test Results on First Silicon in the Design Against Radiation Effects (DARE) Library,” IEEE Trans. Nucl. Sci. 52(5), 1550–1554 (2005). [CrossRef]
9. A. Makihara, Y. Sakaide, Y. Tsuchiya, T. Arimitsu, H. Asai, Y. Iide, H. Shindou, S. Kuboyama, and S. Matsuda, “Single-Event Effects in 0.18 um CMOS Commercial Processes,” IEEE Trans. Nucl. Sci. 50(6), 2135–2138 (2003). [CrossRef]
10. J. D. Black, D. R. Ball, W. H. Robinson, D. M. Fleetwood, R. D. Schrimpf, R. A. Reed, D. A. Black, K. M. Warren, A. D. Tipton, P. E. Dodd, N. F. Haddad, M. A. Xapsos, H. S. Kim, and M. Friendlich, “Characterizing SRAM Single Event Upset in Terms of Single and Multiple Node Charge Collection,” IEEE Trans. Nucl. Sci.55, 2943–2947 (2008). [CrossRef]
11. M. Tsinberg and C. Eggers, “DVD technology,” International Conference on Image Processing , 1, 2 (1998).
12. J. Hellmig, A. Mijiritskii, H. J. Borg, P. Vromans, and K. Musialkova, “Dual-layer Blu-ray Disc based on fast-growth phase-change materials,” International Symposium on Optical Memory and Optical Data Storage Topical Meeting , 407–409 (2002). [CrossRef]
13. P. J. Marchand and P. Ambs, “Developing a parallel-readout optical-disk system,” IEEE Micro 14, 20–27 (1994). [CrossRef]
14. E. Chuang, L. Wenhai, J. P. Drolet, and D. Psaltis, “Holographic random access memory (HRAM),” Proc. of the IEEE 87, 1931–1940 (1999). [CrossRef]
15. S. S. Orlov, E. Bjornson, W. Phillips, Y. Takashima, X. Li, and L. Hesselink, “High transfer rate (1 Gbit/sec) high-capacity holographic disk digital data storage system,” Conference on Lasers and Electro-Optics , 190–191 (2000).
16. H. Horimai and X. Tan, “Holographic Information Storage System: Today and Future,” IEEE Trans. Magn. 43, 943–947 (2007). [CrossRef]
17. J. Hong, J. Ma, T. Chang, I. McMichael, W. Christian, and D. Pletcher, “Holographic memory for fast data access,” IEEE Lasers and Electro-Optics Society Annual Meeting , 1, 160–161 (1996).
18. E. Chuang, J. J. Drolet, G. Barbastathis, and D. Psaltis, “Compact lens-less holographic memory,” Optical Data Storage Topical Meeting Conference Digest , 50–51 (1997). [CrossRef]
19. F. B. McCormick, “Optical MEMS potentials in optical storage,” IEEE/LEOS Summer Topical Meetings , II/5–II/6 (1998).
20. E. G. PAEK, “Microlaser Arrays for Optical Information Processing,” Optics and Photonics News , 4, 16–23 (1993). [CrossRef]
21. H. J. Yoon, N. J. Chung, M. H. Choi, I. S. Park, and J. Jeong, “Estimation of system reliability for uncooled optical transmitters using system reliability function,” J. Lightwave Tech. 17, 1067–1071 (1999). [CrossRef]
22. J. W. Tomm, A. Barwolff, R. Puchert, A. Jaeger, C. Lienau, and T. Elsaesser, “Heating of high-power laser diode arrays: from temperature data to power management and failure mechanisms,” Conference on Lasers and Electro-Optics , 240–241 (1998).
23. C. E. Stroud, “Reliability of Majority Voting Based VLSI Fault-Tolerant Circuits,” IEEE Trans. VLSI Syst. 2(4), 516–521 (1994). [CrossRef]
24. M. Radu, D. Pitica, and C. Posteuca, “Reliability and failure analysis of voting circuits in hardware redundant design,” International Symposium on Electronic Materials and Packaging , 421–423 (2000).
25. Yan Lin and Lei He, “Device and architecture concurrent optimization for FPGA transient soft error rate,” International Conference on Computer-Aided Design , 194–198 (2007). [CrossRef]