Advanced Imaging

AdvancedImagingPro.com

   

Advanced Imaging Magazine

Updated: January 12th, 2011 10:01 AM CDT

CMOS Sensors for High-Speed Holographic Data Storage

The vastly increasing amount of data, along with lower material costs, has made holographic storage a valid option for high-speed data access
Figures 1A and 1B:
© Cypress Semiconductor
Figures 1A and 1B: A laser provides a light beam, which is split in a reference beam and a signal beam. The signal beam is merged with the data pattern and contains the data. The reference beam is used to interfere with the signal beam and write the data to the photosensitive material. The interference pattern of both beams is stored in the photosensitive material by changing the physical properties of the material.
Figure 2
© Cypress Semiconductor
Figure 2: Typical architecture of these high-speed imagers contain just one clock input, as well as a few power supply and synchronization pins. All other signals, for readout as well as exposure, are generated on-chip.
Figure 3
© Cypress Semiconductor
Figure 3: The Cypress LUPA-3000 is a 3Mpixel (1696 x 1710) image sensor operating at 485 fps.
Figure 4
© Cypress Semiconductor
Figure 4: The LUPA-400 sensor has two parallel LVDS output channels each running at 540MHz (DDR).
Advertisement

By Tom Walshap, Cypress Semiconductor

Over the last decade, the amount of information to be processed in image-capture systems has increased dramatically. This leads to an increase of storage capacity as well, while there is the continuing need for reducing device size and cost. The devices presented in this article show an excellent way of realizing both aspects offering a high degree of usability.

High-speed CMOS image sensors produced by Cypress Semiconductor (Mechelen, Belgium) and specifically laid out for motion capture at high frame rates can be applied in holographic data storage systems. Their resolution ranges from 0.4 up to 3Mp and they run at 485 full frames per second. The sensor architecture is based on progressive scan and the outputs deliver digital serial LVDS. Operational speed ranges from 206Mbps DDR up to 540 Mbps DDR, thereby realizing a throughput rate of 0.2 to 1.45Gpix/sec. Image quality is at least 8 bit. The target application requires a 6T snapshot pixel design with high sensitivity, where sensitivity greatly depends on pixel size. The devices are realized in a 0.25µm process.

Data storage and data access usually are done by means of magnetic and optical devices. However, this has the disadvantage of reading from and writing to memory just single bits. Holographic data storage uses a volumetric approach. It therefore can access substantially more data units per operation. Holographic storage has existed for a long time, but the vastly increasing amount of data, along with lower material costs, has made it a valid option for high-speed data access. High-speed image sensors can read the data from the storage element—processing multiple bits simultaneously.

THE PRINCIPLE

Basically, holographic data storage has a write and a read mode. Its functionality is based on a photosensitive material using the optical principle of interference of light beams. Figures 1A and 1B show an example of writing to and reading from the media. A laser provides a light beam, which is split in a reference beam and a signal beam. The signal beam is merged with the data pattern and contains the data. The reference beam is used to interfere with the signal beam and write the data to the photosensitive material. The interference pattern of both beams is stored in the photosensitive material by changing the physical properties of the material.

Readout is done by using the reference beam in the photosensitive material exactly the same way as in the write operation. The beam is diffracted onto the photosensitive material by the content (its physical properties). The diffracted light beam is an exact recreation of the signal beam where the data is stored. The output is captured by an image sensor and the stored data are digitized.

The maximum amount of data processed this way depends on the beams' wavelength. Holograms theoretically can store one bit per cubic block the size of the wavelength of the writing light. At a wavelength of 625nm, this results in a theoretical storage capacity of 4.1 terabits per cubic centimeter. In practice however, storage capacity is lower due to, for example, physical limitations in the system and error correction measures.

SENSOR SPECIFICATIONS

There is growing demand in the marketplace for high-speed image sensors that are small and easy to implement. These days, high-speed imagers are used more and more in consumer applications such as scanning and vision systems or, as in this case, for holographic data storage.

All these applications require large parts of the system's functionality to be located on the image sensor. Consequently, the ADCs, timing generators, image processors and additional output stages are implemented on the chip. In these imagers, the level of feature implementation is as important as sensitivity and processing speed. Most of these imagers are still built to customers' requests, and the proper selection of their functional features will help simplify the design of high-speed custom cameras.

Figure 2 shows the typical architecture. They typically contain just one clock input, as well as a few power supply and synchronization pins. All other signals, for readout as well as exposure, are generated on-chip.

The LUPA-3000 is a 3Mpixel (1696 x 1710) image sensor operating at 485 fps (Figure 3). It has a square 8µm pixel pitch, which results in a 13.68mm x 13.68mm optically active area. The pixels are laid out in 0.25µm technology and equipped with microlenses for good light sensitivity.

The sensor has 32 parallel LVDS output channels each running at 206MHz (DDR). It is equipped with 8-bit ADCs to maximize data throughput—at the expense of SNR. The sensor was designed specifically for applications that require very high data throughput rates (13.3Gbit/s). Lupa-3000 is packaged in a 168-pin BGA.

The LUPA-400, is a 0.4Mpixel (640 x 640) image sensor running at 485 fps (Figure 4). LUPA-400 has a square 4.6µm pixel pitch layout, which results in a 2.944mm x 2.944mm optically active area. As in the LUPA-3000, the pixels are done in 0.25µm technology and are equipped with microlenses for good sensitivity.

The sensor has two parallel LVDS output channels each running at 540MHz (DDR). It is equipped with 8-bit ADCs to maximize data throughput at the expense of SNR. It was designed for applications that require a very high data throughput (2.16Gbit/s), good light sensitivity and simplified camera design.

Cypress Semiconductor designed these two sensors specifically for holographic data storage applications. They can digitize large amounts of data in parallel and therefore boost data throughput.

Tom Walschap studied electronic engineering and has been with Cypress for eight years, including seven as a designer. He can be reached at tom.walschap@
cypress.com
.



Subscribe to our RSS Feeds