The acceleration of product life cycles and the multiplication of vision-use cases leave vision system makers with no choice but to invest less time and money in new developments and focus on their added value.
In September 2022, Teledyne e2v released Optimom 2M, the first in a range of MIPI CSI-2 modules, to address this very challenge. The module combines the latest innovations in imaging and optics into one turnkey imaging solution by mounting a proprietary image sensor onto a board with a fixed lens and optional Multi Focus lens technology. But what are those innovations and how do they work? What benefits do they bring to vision-based systems?
Image Sensor Innovations
The Optimom 2M module features Topaz 2M, a 2-Megapixel global shutter CMOS image sensor that combines multiple innovations from pixel structure to packaging, right through to the chip design itself.
In a world where pure product performance would be the one and only driver in product development, vision system makers would select the largest possible pixel to maximize the sensitivity and saturation capacity of the device. However, in the real world, where money, footprint and power consumption all play a role, vision system makers must balance their desire to maximize the optical performance of the system within size and cost constraints by looking for image sensors with optimum electro-optical performance, that can still fit into a certain optical format.
Depending on the target optical format, the maximum acceptable pixel size can become a technological challenge. Also, going from one optical format to a smaller one (e.g. from 1.1-inch to 1-inch) often implies a significant reduction of pixel pitch as highlighted in Figure 2.
Topaz 2M features the world’s smallest global shutter pixel, which allows it to be matched with compact and cost-effective 1/3-inch lenses while still maximizing sensitivity and signal-to-noise ratio. This pixel, developed by the TowerJazz foundry using their 65nm technology, enables it to perform global shutter operation in a small 2.5μm square size by leveraging the concept of shared pixel structure. In the case of the Topaz 2M sensor, an 8T-shared pixel structure has been adopted, eight transistors being shared by two pixels in diagonal, therefore combining the advanced features from 6T pixel structures such as in-pixel reduction (a.k.a. CDS or Correlated Double Sampling) and the improved sensitivity of 4T structures with only four transistors occupying the surface of each pixel.
A Well-Thought-Out Optical Stack
On top of this structure, the Topaz 2M sensor and the Optimom 2M module benefit from improved sensitivity due to a disruptive optical stack structure on top of the pixel. The pixel optimizes pixel pitch with a gapless top lens to avoid light loss and unwanted reflections, but the real invention lies in the so-called “dual light-pipe” architecture that directly guides light onto the photodiode through micro-optical fibers created in the optical stack of the sensor, which plays with materials of different reflective indexes.
The image shown in Figure 3 presents a cross-view of the optical stack that is embedded into the products.
Apart from optimizing pixel size and optical structure, image sensors can now also benefit from advances in packaging technology to reduce sensor cost, weight, and footprint. For a few years, wafer-level packaging technologies have been booming in the market, especially for consumer applications such as mobile, automotive, or wearables.
While Ceramic Land Grid Array (CLGA) packages have been used in the industry for many years now, the recent technological advances in reducing pixel size have opened the door to wafer-level packages, even for higher-end image sensors meant for industrial inspection, logistics or robotics. CLGA packages require an individual packaging of the die into a ceramic structure, with spaced lands at the back for connecting to the sensor board, whereas wafer-level packages are produced in batches of wafers.
In the case of fan-out wafer-level packages, the silicon wafers are diced into individual sensor dies that are all embedded into a remolded glass substrate wafer, which is then cut into individually packaged sensors. The process and package size optimizations go one step further with another category of wafer-level packaging: chip scale packaging, in which the silicon wafer is directly packaged into the material without the intermediate step of molding a glass substrate around it. This leads to ever smaller and more compact image sensors. For both categories of wafer-level packaging, the back connection of the image sensor to the board is ensured by balls that provide higher density connections, an excellent solution to the challenge of producing miniature and lightweight imaging solutions for embedded systems such as drones or automated guided vehicles.
The recent combination of these pixel, sensor structure, and packaging innovations has enabled a new generation of image sensors with footprints that have reduced by a factor of four over only five years, as highlighted with the timeline and examples shown in Figure 4.
Apart from the packaging technology, the design of the sensor die to be packaged can also have an impact on the size of the final system. One of the key tricks available for image sensor manufacturers is to minimize the final system housing by matching the package center with the optical center to the exact same position. The impact of a mismatch between optical and package centers, as still observed in some image sensors today, is illustrated in Figure 5.
A Novel Technology
While the downscale in pixel pitch has a positive impact on the cost and size of image sensors, it has been quite detrimental to the versatility of optical systems, especially depth of field.
Depth of field, which can be defined as the difference between the closest and furthest distance at which an object can be captured with sufficient levels of sharpness, reduces as the pixel size shrinks and tolerance to out-of-focus images becomes smaller. For applications that require objects to be captured at various working distances (such as for parcel tracking in logistics centers), it becomes usual for system makers to look for close aperture optics (F/7.0 or F/8.0 typically) to maintain a sufficient depth of field despite pixel size reduction.
Unfortunately, closing the aperture comes at the expense of light sensitivity, as less light passes through the lens to be captured by the image sensor. Therefore, the challenge for focus adjustment technologies is to now enable a wide depth of field while maintaining high sensitivity of the vision system. This is precisely the issue solved by the Multi-Focus lens technology developed in the Optimom 2M optical module, which combines a wide F/4.0 aperture with broad working distances from 10 cm to infinity.
This proprietary lens stack technology reaches these performances by modifying the external shape of the lens to adjust the focus. The control of the lens shape is ensured electronically by means of I 2C protocol signals that are directly managed through the standard FFC/ FPC connector at the back of the module board. This connector handles MIPI CSI-2 data output, clock management, as well as image sensor and Multi-Focus lens control through I2C. This concept enables the Multi-Focus to benefit from multiple advantages compared to other focus adjustment technologies, such as fast response time < 1ms and resistance to electromagnetic effects.
The Optimom 2M optical module achieves state-of-the-art electro-optical performances and high versatility by leveraging multiple innovations. The embedded image sensor combines innovations in pixel structure, optical stack, and die packaging to enable a tiny and lightweight design able to match affordable S-mount lenses while maintaining a high level of sensitivity. The optional integrated Multi-Focus lens relies on a new focus adjustment technology, which enables the combination of a broad working distance, high sensitivity and fast response time.
This article was written by Marie-Charlotte Leclerc, Product Manager, Teledyne e2v (Grenoble, France). For more information, visit here .