Cypress Partners With InPhase

Yahoo: Cypress announced that it has partnered with InPhase Technologies in holographic data storage, to supply a CMOS image sensor solution for InPhase's Tapestry holographic data storage system.

Cypress and InPhase have collaborated since 2005 to develop a sensitive and fast CMOS image sensor to meet the high-speed reading of data recorded by InPhase's Tapestry drive. The Cypress CMOS imager features a digital interface and is sized with a pixel count of 1696 X 1710 and a pixel pitch of 8um. The imager enables data to be read at 500fps.

Holographic storage delivers high capacity by recording data throughout the volume of the recording material, and not just on the surface. With each flash of the laser, a data page of approximately 1 million bits is recorded. Each data page is located at a unique address within the material and several hundred pages of data, each with their own unique address, are recorded in the same location of the medium. Each of these collections of data pages is then referred to as a "book." This new recording technique enables more holograms to be stored in the same volume of material by overlapping not only pages, but also books of data. This dramatically increases the storage density.

OmniVision and Scalado Collaborate

Yahoo: OmniVision added a new hardware feature to its new 1.75um 3MP OV3640 CameraChip enabling cameras to take advantage of Scalado's SpeedView technology.

"Long capture and shot-to-shot times have been a constant complaint ever since the megapixel race began," said Sami Niemi, CTO for Scalado. "The WYSIWYG Capturing technology makes it possible to really capture the moment, which is the true essence of any camera phone. Add to that instant photo file maneuvering so users can find and share pictures quickly and easily, and we have addressed another major complaint cited by the majority of camera phone users."

Omnivision also announced volume shipments of OV6680 Square Graphics Array (SGA) sensor with its unique 400 x 400 array.

Micron CIS Business Improving

Micron has announced its Q2 results yesterday. SeekingAlpha conference call transcript reveals few details on image sensor business state.

Mike Sadler, VP of Worldwide Sales says:

We are seeing an improving demand for image sensors as the mobile phone supply chain has recovered from an inventory overhang dating back to last year.

Our customers in the mobile phone arena appear to have worked through inventories, and are now back to consuming Micron image senators to meet current sell-through demand. Our imaging business saw only a moderate revenue decline quarter-over-quarter, after substantially negative growth associated with the inventory correction in fiscal Q2.

The entire portfolio of sensors utilizing our advanced 1.7-micron pixel, ranging from VGA through 8-megapixel density, are being embraced by our customers. We have stacked up a number of design wins that are in the early stage of production, or moving to production throughout the balance of the year.

Peter Trigarszky- Citigroup

That is helpful. And then on image sensors, if you could talk about, I know last quarter you had lowered your loadings going into this quarter. Could you talk about how inventory could be worked down there, or how you see that potentially progressing? Also in VGA, what are your plans for competing in that space from a cost perspective? And maybe give us some sense of low end versus high end, how that mix is progressing for you?

Mike Sadler

Sure, sure. I mentioned that we brought our finished goods inventory down on image sensors. We are at about a quarters worth of finished goods inventory. We had crept up slightly higher than that as we exited fiscal Q2. We will take inventories down again in the current quarter.

So we are getting back to where we consider to be a comfortable level of finished goods inventory on image sensors. We are continuing to take more market share in the high end, our 5-megapixel sensor has been a hit in the marketplace on a couple of high end phones, and the volumes continue to increase with that particular chip.

On the low end of the market, the VGA chip, we have introduced a 1/11" VGA chip, or are in process of introducing it to the market currently, and that is basically going to be our low end cost-aggressive solution, if you will, to trying to attack the low end of the market.

Daniel Amir - WR Hambrecht & Co.

Can you comment what is your image sensor mix here on 1-megapixel and above and VGA?

Mike Sadler

Sure. Let me just do some quick math here. In the quarter that we just completed, at 1-megapixel and above, well, well over half. Probably closer to two-thirds to three-quarters of total unit shipments were 1-megapixel and above.

Eric Rubel - Miller Tabak Roberts

Could you break out the segment revenue for image sensor, and talk about the segment gross margin?

Bill Stover, CFO

The revenues for the quarter came in just above 10% of the total, and we have not been giving specific margin information.

Eric Rubel - Miller Tabak Roberts

Okay. Then if I can ask on, back in September in New York at the Analyst Day, you guys updated us on process at, I believe, 0.11 pixel size 1.75, and a wafer size of about 200-millimeter. Can you update us on how you see that playing out for fiscal 2008 for process, pixel size, and wafer? And I have one more follow-up.

Bill Stover

Well, at the leading edge, a lot of our imagers now are transitioning to the 1.75 Micron pixel technology. We previously announced the 1.4 Micron pixel but that will be a bigger part of our production, a little further out in the future. And was the other question on 200-millimeter versus 300-millimeter wafer capacity?

Eric Rubel - Miller Tabak Roberts

Yes.

Bill Stover

300-millimeter wafer capacity obviously continues to increase as a percent of the total as we ramp out Lehigh, and complete the conversion at tech. I think the way to think of it is in terms of…

Eric Rubel - Miller Tabak Roberts

I am sorry to interrupt, but I am speaking specifically with image sensor, for the image sensor segment, that all of those products were being done on 200-millimeter. Any change on that?

Steve Appleton

That is right. We continue to manufacture the image sensor on 200-millimeter. There has always been discussion around the industry thinking about running 300-millimeter, in fact I think one of our competitors has done so, I think they are primarily doing that, because they have old 300-millimeter equipment they can't really use anymore for memory.

In our particular case, when we look at the cost crossover on that, we don't think it will be cost effective for quite some time. The reason is really a pretty simple one. If you want to go out and buy a piece of a 300-millimeter equipment today then you're going to buy a piece of equipment that can run 50, 60-nanometer type capability, and frankly none of us are running our image sensors on processors that are down at 50 or 60-nanometers, nobody does.

So you are paying for the tool, but you're not utilizing it for the capability. In our particular case, I think as many of you know, we run a model where it's on a tool set that as an X minus 1 or X minus 2 generation, and that tool set has already used its life so to speak for us on the memory front, so it's very, very cost effective for us to use it, because in large part it has been depreciated, and it is a tool set that we would have sold for lower cents on the dollar, then actually get use out of it. So, from our perspective, it’s a great model that works for us. And that’s why we continue to do the 200-millimeter.

As a follow-on note, there clearly will be a time when we are in that situation for 300 millimeter equipment, and as that occurs, we’ll look at the cost effectiveness of taking 300-millimeter equipment that was running memory, and then transferring it over and running image sensors on it, much like we do today on the 200-millimeter, because we think that’s really the way to get the most cost effective part out.

NEC Develops 8MP ISP for Camera Phones

Tech-On: NEC announced that it has started to ship samples of CE131, an ISP for use in phonecams that can process image data of up to 8MP. Volume production is schedule for October 2007. The company reportedly expects the monthly production of 1M units.

Color Filter Array Ideas Paper

In lieu of all the buzz around the "new" Kodak color filter technology, here is a good article by John Savard summarizing the history and different ideas on color filters design.

There used to be a great Fillfactory paper on the same subject, but I can not find it on-line anymore.

Update: As someone mentions in comments, the old Fillfactory article is archived on Wayback Machine: RGB FAQ. Thanks for the link!

Omnivision Announces Its First 1.7um Pixel Product

Yahoo: Omnivision launched the OV3640, the first fully integrated 3MP sensor in a 1/4-inch format, based on its new 1.75um pixel. The sensor fits the standard 8 x 8 mm sockets. It features 2-lane MIPI interface. The new sensor incorporates ISP and onboard JPEG compression.

OV3640 also incorporates advanced image stabilization functionality. It's not clear how it works. The PR says only that the "sensor will automatically sense the slightest camera shake and activate the image stabilization function to help prevent image blur and produce sharper images." To me this sounds like just boosting ISO upon camera shake detection.

The sensor also has integrated AF control for 8.5 x 8.5 mm AF camera modules. Talking about spec, there are two numbers announced: improved dynamic range of up to 65dB and ultra low dark current of 30 e/sec at unspecified temperature. In fact, this is the second time that Omnivision states 1.75um pixel dark current at unknown temperature. This brings a suspicion that it's measured at room temperature. If so, the number is not that good.

Anyway, Omnivision joins a small club of companies owning 1.75um pixel technology. I hope it will close the performance gap with Samsung and Micron in this pixel generation.

Kodak Develops New Color Filter Technology?

Yahoo: Kodak said it has developed a color-filter technology that at least doubles the sensitivity to light of the image sensor in every digital camera.

"We're talking about a 2-to-4-times improvement in (light) sensitivity" - said Chris McNiffe, general manager of the photography company's image sensor business.

Reuters: Kodak's new proprietary technology adds "clear" pixels to the red, green, and blue elements that form the image sensor array, collecting a higher proportion of the light striking the sensor.

Manufacturing customers interested in the design will likely get a chance to sample it in early 2008, but McNiffe was unsure when devices using the technology would be in stores. The technology could be used at first in devices such as cell phones and eventually products made for industrial and scientific imaging.

IDC's Chute said Kodak would probably use the technology for its own cameras, hoping to gain a competitive edge. "The potential (for its success) is always there, but it's a wait-and-see thing," he said.

CNET too published an extended article with many pictures on the Kodak color filter technology.

I wonder how Kodak can claim it's new, if the idea is known for decades already. Hopefully Kodak found a way to overcome the huge crosstalk coming from the "white" pixel.

Update: EETimes, PC World, DPReview, Imaging Resource and San Jose Mercury News each published its own version of the Kodak announcement. Imaging Resource's version is the most comprehensive.

Strategies Unlimited Market Report

Solid-State Technology: The image sensor market is projected to grow 14% in 2007, following 30% growth in 2006 to roughly $6 billion, and will slow through the next several years, according to a report from Strategies Unlimited.

Micron is now the top supplier of CMOS sensors, with strong gains also enjoyed by STMicroelectronics and Samsung. Top five suppliers still hold about two-thirds of overall image sensor market share, with Sony and Micron each approaching $1 billion in annual revenues. 50 or so image sensor suppliers are about twice what the market supported a decade ago, though that number hasn't changed in the past few years -- suggesting that for every Atmel and ESS Technology that gets out of the image sensor business, there's a Planet82 or a ProMOS Technologies that comes in to take their place.

ST Licenses Varioptic Lens

Varioptic, the liquid lens company, announced the signature of a licensing agreement with STMicroelectronics. Under the terms of the agreement, STMicroelectronics will use Varioptic’s liquid lens technology to add auto-focus capability to a family of camera phone modules. Additionally, ST will also be able to manufacture liquid lenses in its own facilities. Varioptic’s liquid lens products include the current Arctic series (Arctic 320 and 416) as well as new thinner versions to be launched in the coming months.

Eric Aussedat, General Manager of ST’s Imaging Division, commented: “By combining Varioptic’s latest liquid lenses with our industry-leading 1.75-micron sensor technology, we believe we can break current performance-cost frontiers and dramatically increase the adoption of autofocus in phones.”

Tessera Wafer Level Camera

After many years in development Tessera finally announced a new wafer-level camera technology OptiML WLC (also known as OptiuL WLC).

Using OptiML WLC technology, thousands of lenses are manufactured simultaneously on a wafer, and then bonded at the wafer level to create the optical element of the camera. The result is simplified assembly, up to 30% cost savings for the optical component of the camera module and up to 50% size reductions over conventional camera phone modules. Tessera's recently acquired Eyesquad technology can be easily integrated into the OptiML WLC solution, providing advanced auto-focus and digital optical zoom without the use of moving parts.

Update: EETimes too published the announcemnet. Tessera claims the "Z" dimension for the OptiML WLC's VGA implementation comes in at 2.5 mm. Eyesquad's digital focus technology becomes crucial when wafer-level cameras need 2-megapixel resolution or better.

The first working-prototype module made to custom specs using OptiML WLC will emerge at the end of next quarter. Wafer-level cameras will be incorporated in handsets by the end of 2008.

Sharp's 5mm Thick Module with AF

Tech-On: Sharp Corp. launched a 1/4-inch 2MP CMOS camera module, the "LZ0P39DS," which features automatic focus. The module measures 8.5 x 8.5 mm with a thickness of 5.0 mm.

The module is compliant with UXGA UYVY-8b video at 15 fps. The aperture of the lens is F2.8/H:53. Volume production will be started in July 2007 at 300,000 modules per month.

Machine Vision Sensors Trends

Vision Systems Design: Martin Wäny of Awaiba published a review of trends in machine vision CCD and CMOS sensors. The review covers some basics, but it's quite incomplete. Many important developments are omitted. It's an interesting read for general public, anyway.

Sony to Spend $492M to Expand CIS Capacity

International Herald Tribune: Sony Corp. said Wednesday it will invest 60 billion yen (US$492 million; €363.4 million) to ramp up production of image sensors used in digital cameras.

The expansion at Sony's Kumamoto image sensor factory in southern Japan will span the next three years, the company said in a statement.

Tokyo-based Sony did not say by how much output would be boosted, but the Nikkei business newspaper said the investment will increase capacity by 20 percent.

The Kumamoto facility makes CMOS sensors and micro-display devices that are used in hot-selling camera-equipped mobile phones and digital cameras. Sony is increasing its production capacity to match an expecting surge in Japan.

Update: ZDNet too reports that with $494.3 million investment, Sony will boost its output capacity of CMOS censor chips. Analysts say Sony may face an uphill battle in its planned expansion in its sensor chip operations.

Since Sony is relatively new to the CMOS market, it will likely need to offer its products at aggressive prices to compete with established makers such as Micron Technology and Toshiba, hurting its profitability, said Yoshihisa Toyosaki, president of research firm j-STAR Global.

Update June 7: EDN has an expanded article on the same matter. Sony said it is positioning image sensors as a key focus area of its semiconductor business.

Sony Corp. announced that it would invest approximately $493.1 million (60 billion Japanese yen) in Sony Semiconductor Kyushu Corp.'s Kumamoto Technology Center (Kumamoto TEC) Fab 2 facility in Japan. The money will be used to extend clean room facilities by 5,000-sq. meters and reinforce image sensor fabrication capacity. The mass production have already commenced at Fab 2's existing 5,000-sq. meter facility in the spring of this year.

Update June 7: Tech-On gives even more details on Sony plans. Currently, only Sony Semiconductor Kyushu manufactures Sony's CMOS sensors on mass production lines, with the capacity of three million units per month. The extended clean room will mainly be used to increase the production of CMOS sensors for camera cell phones and digital cameras.

Image Sensor Workshop 2007 Program

Image Sensor Workshop 2007 takes off on June 6. There is a lot of interesting papers in the advanced program.

ST team presents a back-illuminated 1.45um pixel based on 1T5 (?) architecture:

A 3 Mega-Pixel Back-illuminated Image Sensor in 1T5 Architecture with 1.45 mm Pixel Pitch.
Jens Prima, Francois Roy, Perceval Coudrain, Xavier Gagnard, Norbert Moussy, Benoit Giffard and Pierre Gidon

Samsung presents its work on 1.4um pixel:

Optoelectronic Investigation for High Performance 1.4 mm pixel CMOS Image Sensors
Seong-Ho Cho, Gibum Kim, Hyunpil Noh, Chang-Rok Moon, Kangbok Lee, Kwangok Koh and Duckhyung Lee

Other interesting papers include:

Highlight Scene FPN on Shared Pixels and a Reduction Technique
Takashi Watanabe, Katsuji Kimura, Masamitsu Taki, Mitsuru Homma, Shoko Daikoku, and Tetsuya Fujimoto, and Kiyotoshi Misawa

1/4 Inch 2Mega CMOS Image Sensor Fabrication
Se Jung Oh, Jae Young Lim, Chan Ki Kim, An Sik Choi, Do Young Lee

Implementing a CMOS Image Sensor Noise Performance Model
Keith Findlater, Ryan Gow, David Renshaw, Lindsay Grant, John Hart, Stuart McLeod, Robert Nico

A 2.3e- Read Noise 1.3Mpixel CMOS Image Sensor with Per-Column Amplifier
Kwang-Bo Cho, Chiajen Lee, Siri Eikedal, Hai Yan, Taehee Cho, Tien-Min Miao, Jason Song, Christopher Zeleznik, Alexander Mokhnatyuk, and Sandor Barna

Continuous Time Column Parallel Readout for CMOS Image Sensor
G.G.Storm, M.D. Purcell, R.E. Elliott, L.A. Grant T.Lule, F. Mutze, and M.Sommer

Analysis of Source Follower Random Telegraph Signal Using nMOS and pMOS Array TEG
K. Abe, S. Sugawa, R. Kuroda, S. Watabe, N. Miyamoto, A. Teramoto, T. Ohmi, Y. Kamata and K. Shibusawa

Consideration of Dark Current Generation at The Transfer Channel Region in the Solid State Image Sensor
Yoshihisa Kunimi and Bedabrata Pain

1.4 Gpix/s, 2.9Mpix CMOS Image Sensor for Readout of Holographic Data Memory
Roel Aerts, Steve Domer, Judy Fong, Kellie Vanda, Ray Rubacha, Brannon Harris and Tomas Geurts

A High Speed High Resolution Digital CMOS Sensor
A. Krymski

Dynamic Range Extension of CMOS Imager With Linear Response by Hybrid Use of Active and Passive Pixel Readouts
Keilchiro Kagawa, Yudai Adachi, Yugo Nose, Kumiyuki Tam, Atsushi Wada, Masahimo Numoshita and Jun Ohta

Active Pixel Sensor Arrays in 90/65 nm CMOS-Technologies with Vertically Stacked Photodiodes
S. Henker, C. Mayr, J-U. Shlusler, R. Schuffny, U. Ramacher and A. Heitmann

A Wide Dynamic Range CMOS Image Sensor with Resistance to High Temperatures
Koichi Mizobuchi, Satoru Adachi, Tomokazu Yamashita, Seiichiro Okamura, Hiromichi Oshikubo, Nana Akahane and Shigetoshi Sugawa

A 121.8dB Dynamic Range CMOS Image Sensor using Pixel-Variation-Free Midpoint Potential Drive and Overlapping Multiple Exposures
Yusuke Oike, Atsushi Toda, Tadayuki Taura, Akihiko Kato, Hiroki Sato, Masanori Kasai and Tadakuni Narabu

A High Dynamic Range Digital Linlog CMOS Image Sensor Architecture Based on Event Readout of Pixels and Suitable for Low Voltage Operation
Alexandre Guilvard, Pierre Magnan, Josep Segura, and Philippe Martin-Gonthier

Color filter array with sparse color sampling crosses for mobile phone image sensors
Gang Luo

Crosstalk, Color Tint and Shading Correction for Small Pixel Size Image Sensor
Alexander Getman, Timofei Uvarov, YongIn Han, Bumsuk Kim, JungChak Ahn, YongHee Lee

Capacity and Image Lag Simulator as a Development Tool of CMOS Image Sensor
Kee-Hyun Paik, Jongcheol Shin, Seok-Ha Lee, Chang-Rok Moon, Chang-Hyo Koo, Keun-Ho Lee, Duckhyung Lee,Young-Kwan Park and Moon-Hyun Yoo

3-D Wave Optical Simulation of Light Wave-guide Structures by LBEM
Hideki Mutoh and Shigetoshi Sugawa

Wafer-Level Thinned CMOS Imagers Implemented in a Bulk-CMOS Technology
Bedabrata Pain, Chao Sun, Paul Vo, Bruce Hancock, Thomas Cunningham, Chris Wrigley, Risaku Toda, Victor White, Amrita Banerjee and Durgmadhab Misra

A monolithic 111-M Pixel High Speed, High Resolution CCD
Richard Bredthauer, Greg Bredthauer, and Kasey Boggs

Pixel-Level A/D Conversion: Comparison of Two Charge Packets Counting Techniques
Arnaud Peizerat, Marc Arques, Patrick Villard, Jean-Luc Martin

High Sensitivity of Dielectric films Structure for Advanced CMOS Image Sensor Technology
Chung-Wei Chang, Shou-Gwo Wuu, Dun-Nian Yaung, Chien-Hsien Tseng, Han-Chi Liu, David Yen, Yi-Jiun Lin, Chun-Ming Su, Chun-Yao Ko, C.Y. Yu1, C.H. Lo, F.J.Hsiu, C.S.Tsai, Chung S. Wang, Mingo Liu, Chrong-Jung Lin, Ya-Chin King

Modeling of the Temporal Pixel to Pixel Noise of CMOS Image Sensors
Cédric Leyris, François Roy, Didier Hérault, and Mathieu Marin

Characterization of the Buried Channel n-MOST Source Followers in CMOS Image Sensors
Xinyang Wang, Padmakumar R. Rao, and Albert J.P. Theuwissen

Optimization of Random Telegraph Noise Non Uniformity in a CMOS Pixel with a pinned-photodiode
Assaf Lahav, Dmitry Veinger and Amos Fenigstein

Column Parallel Signal Processing Techniques for Reducing Thermal and Random Telegraph Noises in CMOS Image Sensors
Shoji Kawahito and Nobuhiro Kawai

Twinkling Pixels: Random Telegraph Signals at Reset Gate Edge
Bedabrata Pain, Thomas J. Cunningham, Bruce Hancock, Chao Sun, and Chris Wrigley

Evaluation of a Small Negative Transfer Gate Bias on the Performance of 4T CMOS Image Sensor Pixels.
Hyungjun Han, Hongjoo Park, Parker Altice, Woonil Choi, Younsub Lim, Sangjoo Lee, Seok Kang, Jaeyeong.kim, Smith Yoon and Jerry Hynecek

Some Thoughts on Diffusion Dark Current
M.M. Blouke

Gr Gb difference in 3M CMOS Image Sensor with 1.75mm pixel
Bumsuk Kim, Yoonho Jang, Kyoungsik Moon, Eun-Gyu Lee, Alexander Getman, JungChak Ahn and YongHee Lee

CMOS Color Image Sensor with Overlaid Organic Photoconductive Layers Having Narrow Absorption Band
Shunji Takada, Mikio Ihama and Masafumi Inuiya

52 Mega-pixel APS-H-size CMOS Image Sensor for Super High Resolution Image Capturing
Masaaki Iwane, Takashi Matsuda, Takashi Sugai, Koichi Tazoe, Takashi Okagawa, Toshiaki Ono, Takanori Watanabe, Katsuhisa Ogawa, Hidekazu Takahashi and Shunsuke Inoue

Variability Limits the Advantage of Photo Diode’s Zero Bias Operation
Bart Dierickx

Stratified Photodiode a New Concept for Small Size High Performance CMOS Image Sensor Pixels.
Younsub Lim, Kyunglak Lee, Heejeong Hong, Jaeyeong Kim, Seunghoon Sa, Juil Lee, Daebyung.Kim and Jerry Hynecek

Super Small, Sub 2µm Pixels For Novel CMOS Image Sensors
G. Agranov, R. Mauritzson; S. Barna, J. Ladd, J. Jiang, A. Dokoutchaev and X. Fan

Improved Design of 1T Charge-Modulation Pixel Structure for Small-Size and Low-Dark-Current Achievements
Arnaud Tournier, François Roy, Guo-Neng Lu and Benoît Deschamps

Orthogonal Transfer Arrays for Wide-Field Adaptive Imaging
Barry E. Burke, Michael J. Cooper, Peter E. Doherty, Andrew H. Loomis, and Douglas J. Young, John L. Tonry and Peter Onaka

Overall, it's going to be a very interesting event.

50MP Canon

Engadget: Canon has built a 50MP CMOS sensor, which is reportedly almost twice the resolution of its nearest competition, and is prepping it as a sort of large format surveillance camera for monitoring large, busy areas such as parking lots and theme parks, along with detailed work like factory part inspections. Despite the sensor's clear industrial-end aims, Canon has managed to build its prototype at 19 x 28mm in size, the same dimensions of the sensors in its DSLR cameras. So far Canon hasn't announced release plans for this sensor.

Notebook Webcam Design Considerations

Cypress product manager Shone Tran publishes an article on webcam design in EETAsia. The claim is that no high resolution camera is needed in this application. Also, various integration trade-offs discussed.

Forza FORZ-HD Technology

PRWeb: Forza Silicon Corporation announced today the availability of its proprietary FORZ-HD CMOS imaging technology, which supports frame rates up to 60 fps, proven at 33.2 Mega-pixel resolution, and features 12-bit on-chip ADCs. FORZ-HD technology is targeted for high resolution, high-speed, high dynamic range digital cameras used in professional photography or digital cinematography, and in medical, biotech, scientific, satellite and defense industries.

FORZ-HD imaging technology is used in CMOS image sensors fabricated by Forza foundry partner Tower Semiconductor using 0.18 micron process. FORZ-HD is compatible with CMOS Image Sensor processes used by Forza foundry partner IBM Microelectronics.

The datasheet indicates that ADC is successive approximation column parallel having 260Ksps. An example spec table states the following parameters:

Array Size 7832 x 4360
ADC Resolution 12-bit
ADC Conversion Rate 260 KS/sec
Frame Rate 60 fps
Conversion Gain 50 uV/e
Responsitivity 2 V/lux-sec
Signal/Noise Ratio 43 dB
Dynamic Range 60 dB
Digital Power Supply 1.8 V
Analog Power Supply 3.3 V

Analysts on Omnivision Report

Barrons: Sentiment on the Street is mixed.

Daniel Gelbtuch, an analyst at CIBC, maintains a Sector Perform rating on the stock, despite conceding that “valuation appears fetching.” He thinks the company is likely to lose share to competitors, including privately held SiliconFile at the high end of the market and privately held MagnaChip at the low end.

Also skeptical is Jefferies & Co.’s Adam Benjamin, who cites concerns that the company’s guidance “could prove aggressive due to its high turns ratio and our belief that recent order expediting could be clouding true demand.”

Tristan Gerra, an analyst at Robert W. Baird, adds that both Micron (MU) and Samsung continue to take market share in image sensors, and asserts that the company’s recovery “should be short-lived.” He repeated his Underperform rating on the stock.

Michael Masdea, an analyst at Credit Suisse, is more bullish, and maintains his Overweight rating. He writes in a note that the company should benefit as camera phone makers trend toward higher resolution image sensors, and many move to two sensors per phone to enable video conferencing.

Omnivision Q407 Conference Call

SeekingAlpha published a transcript of Omnivision Q4 2007 conference call. Clearly, Omnivision business is improving:

"Revenues were better than expected, particularly in 2-megapixel and VGA. More importantly, we saw acceleration in 2-megapixel orders for future quarters, demonstrating what we believe to be a fundamental shift in the mobile handset market to higher resolution sensors."

The gross margin was quite low at 22.3% compared to 24.9% last quarter. To improve it, Omnivision takes following measures:

"One, as a fab-less company, virtually all of our product costs are brought in and we are working very aggressively with all our vendors to reduce input costs and improve output yields;

Two, we are collaborating very closely with TSMC to develop new and lower cost technologies;

Three, as you may remember from last quarter, we entered into a foundry manufacturing agreement with PFC to develop new processes and new products using their 300-millimeter DRAM capacity;

Four, along with TSMC, we have continued to fund our joint venture with VisEra to expand their capacity and help prepare their facilities to handle 300-millimeter wafers;

Five, earlier this month, we also made an investment in the WLCSP, a chip scale packaging partner. This provides us with a second source of chip scale packaging capacity;

Six, we believe that TSMC’s acquisition of a controlling interest in XinTec, our original CSP provider, will bring a higher level of efficiency to XinTec’s production processes and lead to increased yields;

Seven, we are in the process of automating and upgrading our final testing facility by installing new test equipment, making our testing process more efficient."

"We are also very close to launching the OV3640, the first product based on the new architecture, a higher performance, 3.2-megapixel image sensor for high-end slim phones, which is capable of fitting into the all-important quarter-inch form factor."

"We are seeing a very good ramp in the PC notebook market, with design wins at four of the five leading notebook manufacturers. Our market share is very strong. We believe that we are well-positioned to take advantage of the acceleration in the attach rate for cameras in this market. Our OV7725 sensor offers superior image quality in all lighting conditions and we are winning designs at resolutions up to and including 2-megapixels.

From some of the market estimates we have seen, the number of notebooks shipping with a camera this year could exceed 20 million, which equates to an attach rate of about 25%. Adoption rates appear to be accelerating and PC notebook units could increase to over 30 million next year."

A question and answers session:

Q: Aaron Husock - Morgan Stanley

Thanks for taking my call. I guess to start, could you give us a little more detail on the mix in the quarter and where that’s going? I guess last quarter, you were around two-third VGA, 10% 2-megapixel and the rest 1.3. Where did that come out in the April quarter and where can that go in July?

A: Peter Leigh

As I indicated in my prepared remarks, VGA was up a little bit in the quarter, around 70% and the 1-megapixel and 2-megapixel shares were virtually unchanged. We do believe that 2-megapixel should increase as a percentage of revenues in the upcoming quarters.

Q: Aaron Husock - Morgan Stanley

One final question -- there’s been some talk in the market about a shortage of image sensors for the PC and notebook market in China. Can you just comment on that, what you are seeing there? Is there a shortage and do you guys have adequate supply?

A: Ray Cisneros

That’s an interesting comment but no, we don’t see that. Right now, we feel very strong about our position in the PC cam and laptop market, and so we can’t really confirm that and we don’t see it.

Q: Doug Freedman - American Technology Research

Is there a way that you can help us understand what the guiding factors are, how much we should look at ASPs being a contributing factor versus your ability to reduce cost?

A: Peter Leigh

I think the way you have to look at it is that historically in this market, prices have gone down at roughly 20% to 25% per year, and if you look at the last two quarters for OmniVision, actually we are at about a 20% decline. Historically, the company has been able to drive down its costs at about that rate. Our task now, of course, is to drive our costs down faster than that so that we can begin to see some margin expansion. We think we are on the right track to doing that and we hope to be able to deliver steady improvement over the next several quarters.

Q: Paul Coster - J.P. Morgan

The guidance that you have issued suggest that your market share in the next quarter, just on the back of the envelope here, in the handset market is potentially north of 25%. Do you agree and how do you think the competitive dynamics are shaping up in the space at the moment?

A: Ray Cisneros

Historically, we’ve mentioned that our market share is anywhere between 25% to 30%. Obviously it depends on which third party reference report you are going to look at, but yes, given our guidance obviously there should be some juggling of the numbers. It is only until the numbers are set and done do we ever get a good feel of the total market share as that, the overall number, the total available market continues to grow as well. But yes, we expect some favorable results.

Q: Harsh Kumar - Morgan Keegan

So are you suggesting that you might be doing market share in the higher end space, perhaps? That’s the only real explanation that I can think of why you would be outperforming the market significantly.

A: Ray Cisneros

It’s interesting. I should mention as well in the VGA space, as Peter has indicated, there was a slight increase in the VGA percent mix in Q4 and we see the same strong VGA position in Q1 as well. So we have a bi-fold surge here of 2-megapixel and VGA.

Image Sensor Market to Grow in 2007

Evertiq: After a lackluster start in 2007, the market for semiconductor image sensors is expected to strengthen this summer and finish the year with modest growth of 11% and record-high sales of $8.2 billion, based on IC Insights' analysis of current business conditions.

Worldwide sales of CCD and CMOS image sensors will increase at a cumulative average growth rate (CAGR) of 14% between 2006 and 2011, reaching annual revenues of $14.3 billion in five years (Figure 1), according to IC Insights' 2007 Optoelectronics, Sensor, and Discrete (O-S-D) Report.

During the first quarter of 2007, inventory corrections in camera cell phones and slowing growth rates in digital still cameras contributed to a 6% sequential decline in image sensor sales compared to 4Q06. However, total image sensor sales reached $1.8 billion in 1Q07, a 6.5% increase from 1Q06. IC Insights is predicting that image sensor sales will increase 10.7% to $8.2 billion in 2007, compared to $7.4 billion in 2006 when the market grew 17.4% from $6.3 billion in 2005.

Shipments of image sensors peaked in 3Q06, partly due to the buildup of camera phone inventories in Asia. After dealing with weakness in the first three months of this year, suppliers of CMOS image sensors are now reporting signs of stabilization and early recovery in some market segments as excess inventories of phones begin to disappear. More importantly, new orders for 1.3-megapixel and greater CMOS image sensors are finally gaining momentum as more consumers decide to buy cellular handsets with enhanced embedded cameras for higher resolution still photos and video recordings. New CMOS image sensor solutions with built-in auto focusing features also promise to increase average selling prices (ASPs) of camera-phone devices in the second half of 2007.

CMOS devices now account for more than 65% of overall image sensor sales. CCDs (charge-coupled devices) continue to lose share to low-power consuming CMOS technology, which enables more functions to be integrated on a single die and allows standard IC wafer fabs to make inexpensive image sensors. IC Insights forecasts 24% sales growth in CMOS image sensor sales to $5.5 billion in 2007, while CCD revenues are expected to decline 9% to nearly $2.6 billion.

After a couple years of phenomenal growth, Micron Technology took over the top spot in the CMOS image sensor market in 2006 with $845 million in sales, followed by STMicroelectronics with $540 million. OmniVision, previously the largest CMOS image sensor supplier, slipped to third place with $520 million in device sales last year, according to IC Insights' ranking in the 2007 O-S-D Report. Underscoring the difficulties faced by suppliers at the start of 2007, Micron's image sensor sales fell sequentially 36% to $156 million in its last fiscal quarter, ended March 1, compared to $244 million in the prior three-month period ended November 30, 2006. Micron blamed weakness in mobile handsets, increased competition, and a shift in the market mix toward lower value VGA-based camera phones.