Samsung presented a poster paper "The Effects of Nano-Sized Defects on Dark Current of High End CMOS Image Sensor" on May 2008 Electrochemical Society Meeting. The main claim is that defects in epi layer cause a significant increase in hot pixels. The article compares few epi wafer vendors and shows dark current difference between them.
Another Samsung paper presented on 2008 Symposium on Design, Test, Integration and Packaging of MEMS/MOEMS, Nice, France describes WLP process flow entirely developed by Samsung. The claim is that the new process is cheaper and better than the available alternatives, such as Shellcase by Tessera.
Monday, October 06, 2008
Saturday, October 04, 2008
DXO Talks About ISP Challenges
DxO's Nicolas Touchard talks about ISP challenges in his presentation on CPIQ forum. The main problem is to keep up with pixel shrink effects, despite a long ISP design cycle. So the modern ISPs are optimized to compensate pixel imperfections that were 2-3 years ago.
I don't completely agree with this view. There are ISP design teams with good connections with sensor vendors and there are others with bad or no connections. The later ones learn pixel effects from literature and naturally have 2-3 years lag. The former ones have quite good understanding of small pixels and their ISP account for all modern pixel effects.
I don't completely agree with this view. There are ISP design teams with good connections with sensor vendors and there are others with bad or no connections. The later ones learn pixel effects from literature and naturally have 2-3 years lag. The former ones have quite good understanding of small pixels and their ISP account for all modern pixel effects.
Friday, October 03, 2008
Weekly Patent Review
This week the most interesting application comes from Canon. US20080231859 application is supposed to solve most of the small pixel problems, such as reduced sensitivity and low SNR. It achieves the improvement by changing a Bayer RGB filter for a "spectroscopic element formed by means of an optical microresonator".
The bad thing is that I was unable to understand how these resonating cavities made and how they manage to effectively generate photocharge in smaller than wavelength pixels. I've read the description back and forth and I'm still puzzled. Anyway, the promise of this idea is great and if Canon knows how to make it, this would be a huge boost in image quality of small pixels.
The bad thing is that I was unable to understand how these resonating cavities made and how they manage to effectively generate photocharge in smaller than wavelength pixels. I've read the description back and forth and I'm still puzzled. Anyway, the promise of this idea is great and if Canon knows how to make it, this would be a huge boost in image quality of small pixels.
Thursday, October 02, 2008
Aptina Becomes Independent Separate on Friday, Oct. 3
Seeking Alpha's Micron earnings call transcript has few words about Aptina's next stage in life:
Daniel Amir - Lazard Capital Markets:
Just a question on the image sensor business. Can you comment a bit on the visibility there, what you’re seeing there in terms of growth for the next couple of quarter, if at all? And kind of what the strategy is now for the segment?
Steven R. Appleton, CEO:
I wouldn’t say there’s a lot of growth. I think it looks pretty stable. I would say not a lot of growth, not a lot of decline. There are some incremental improvements that we’re making. If you would have noticed, for image sensor companies, they’ve made somewhat of a comeback and I think have gained some market share in the last quarter or two.
But, hey, look, I think that business, a lot of it is in the mobile stage and the technology stage, if you will, around PDAs and digital cameras and PC cams, etc. so I think we have some similar challenges to what other segments do, but it still looks like they’re on a pretty good path.
In terms of what we’re intending to do in that space, there hasn’t been any change of plans with respect to what’s happening with that Aptina. In fact, on this Friday they will start operating as a separate company from Micron, although it will be a wholly-owned subsidiary they will have their own systems and so forth, and we’re still moving down the path as journey with a partner to likely take that company and try to work, but yet have Micron continued to manufacture the product, at least in the foreseeable future.
Daniel Amir - Lazard Capital Markets:
Just a question on the image sensor business. Can you comment a bit on the visibility there, what you’re seeing there in terms of growth for the next couple of quarter, if at all? And kind of what the strategy is now for the segment?
Steven R. Appleton, CEO:
I wouldn’t say there’s a lot of growth. I think it looks pretty stable. I would say not a lot of growth, not a lot of decline. There are some incremental improvements that we’re making. If you would have noticed, for image sensor companies, they’ve made somewhat of a comeback and I think have gained some market share in the last quarter or two.
But, hey, look, I think that business, a lot of it is in the mobile stage and the technology stage, if you will, around PDAs and digital cameras and PC cams, etc. so I think we have some similar challenges to what other segments do, but it still looks like they’re on a pretty good path.
In terms of what we’re intending to do in that space, there hasn’t been any change of plans with respect to what’s happening with that Aptina. In fact, on this Friday they will start operating as a separate company from Micron, although it will be a wholly-owned subsidiary they will have their own systems and so forth, and we’re still moving down the path as journey with a partner to likely take that company and try to work, but yet have Micron continued to manufacture the product, at least in the foreseeable future.
Wednesday, October 01, 2008
Micron Quarterly Results
Yahoo: Micron reported its quarterly results today. Sales of CMOS image sensors in the fourth quarter of fiscal 2008 increased slightly compared to the third quarter and represented 12 percent of the company’s total sales in the fourth quarter. The company’s gross margin on sales of imaging products during the fourth quarter decreased to 29 percent compared to 35 percent in the third quarter, primarily due to lower average selling prices.
Samsung Techwin on Image Sensors
DPReview published an interview with Choong-hyun Hwang, Vice President of the Digital Camera Business of Samsung Techwin Strategy Marketing Team. There are few interesting bits on DSC image sensors from Samsung Techwin point of view:
DPReview: On your compact cameras do you use your own sensors?
Samsung: No, we don't.
DPR: Would there be any advantage to developing your own compact sensors?
S: In terms of efficiency we're better at the moment outsourcing. We currently buy sensors from a range of suppliers.
DPR: Do you make any compact camera sensors?
S: No, at the moment we don't.
DPR: Do you think that CMOS is likely to eventually take over from CCD sensors in compacts?
S: Ah you are asking difficult questions! I think that many people - everyone - expects CMOS sensors to be applied to more compact cameras. Sony has a high speed CMOS sensor in development, and the speed is remarkable.
DPR: Is it less expensive to manufacture CMOS sensors?
S: Oh no, it's much more expensive. The Sony and Canon high speed CMOS sensors are very expensive to manufacture. The standard CMOS sensors used in mobile phones are very small and are inexpensive, but not those for compacts cameras. Canon has been making a lot of effort to replace CCD with CMOS, and they have not yet been successful because the performance and efficiency aren't good enough yet.
DPReview: On your compact cameras do you use your own sensors?
Samsung: No, we don't.
DPR: Would there be any advantage to developing your own compact sensors?
S: In terms of efficiency we're better at the moment outsourcing. We currently buy sensors from a range of suppliers.
DPR: Do you make any compact camera sensors?
S: No, at the moment we don't.
DPR: Do you think that CMOS is likely to eventually take over from CCD sensors in compacts?
S: Ah you are asking difficult questions! I think that many people - everyone - expects CMOS sensors to be applied to more compact cameras. Sony has a high speed CMOS sensor in development, and the speed is remarkable.
DPR: Is it less expensive to manufacture CMOS sensors?
S: Oh no, it's much more expensive. The Sony and Canon high speed CMOS sensors are very expensive to manufacture. The standard CMOS sensors used in mobile phones are very small and are inexpensive, but not those for compacts cameras. Canon has been making a lot of effort to replace CCD with CMOS, and they have not yet been successful because the performance and efficiency aren't good enough yet.
SiliconHive Partners with Acutelogic
Video/Imaging Design Line: A partnership of Silicon Hive and Acutelogic spawned a new family of HiveGo programmable camera imaging subsystems (CSS) for SoC manufacturers in the mobile phone and consumer electronics space.
The three initial products offered as licensable IP are:
The three initial products offered as licensable IP are:
- HiveGo CSS 3016J which targets DSC, DVC and UMPC cameras, incorporates Silicon Hive's 3rd generation HiveFlex ISP 2300 scalable SIMD processor, a control processor, and accelerators for filters, scaling, smooth digital zoom, and distortion correction. The hardware configuration yields ultra-high quality still capture at up to 260 Mega Pixels per second, or 16MP at 15 frames per second (fps) 'on the fly,' or capturing and processing HD video streams at 1080p 60fps. Camera input formats such as SMIA-CCP2 and MIPI-CS12 are supported. Robust GPIO connections for camera systems controls such as lens, focus, and flash are included.
- HiveGo CSS 3012J targets high quality imaging demands of mobile Internet devices (MIDs). The mid product performs still capture at up to 200 Megal Pixels/sec or 12 MP at 15 fps.
- HiveGo CSS 3108J targets high-volume smart phones, performing still capture at up to 130 Mega Pixels/sec or 8 MP images at 15 fps.
Catching-Up with Tessera News
Yahoo: Advanced Wafer Level Packaging Inc. (AWLP) has licensed the full range of SHELLCASE image sensor packaging technologies from Tessera, including the SHELLCASE MVP solution, which is one of the industry’s first Through Silicon Via (TSV) solutions.
AWLP is an early-stage start-up company based in Seoul, Korea. The new outsource packaging company will be dedicated to providing SHELLCASE packaging services to semiconductor manufacturing companies. AWLP was established in 2007 as a foundry service provider dedicated to Tessera WLCSP technology. AWLP’s facility is now being built in Kwangju City and will begin production by the end of 1Q09.
Yahoo: Fujitsu Microelectronics' Milbeaut chipsets incorporate the FotoNation FaceTracker image enhancement solution from Tessera. The FotoNation FaceTracker solution tracks faces in the camera viewfinder, detecting more than 10 faces in one image in as little as 0.1 seconds. It continues to track them moving throughout the image, even in rotation or profile. The camera uses this “face data” to adjust image capture settings such as exposure and focus, which results in human subjects that are well-lit and well-focused, even when backlit or off-center. The technology also includes face orientation data as a simple solution for auto-rotation without additional hardware.
AWLP is an early-stage start-up company based in Seoul, Korea. The new outsource packaging company will be dedicated to providing SHELLCASE packaging services to semiconductor manufacturing companies. AWLP was established in 2007 as a foundry service provider dedicated to Tessera WLCSP technology. AWLP’s facility is now being built in Kwangju City and will begin production by the end of 1Q09.
Yahoo: Fujitsu Microelectronics' Milbeaut chipsets incorporate the FotoNation FaceTracker image enhancement solution from Tessera. The FotoNation FaceTracker solution tracks faces in the camera viewfinder, detecting more than 10 faces in one image in as little as 0.1 seconds. It continues to track them moving throughout the image, even in rotation or profile. The camera uses this “face data” to adjust image capture settings such as exposure and focus, which results in human subjects that are well-lit and well-focused, even when backlit or off-center. The technology also includes face orientation data as a simple solution for auto-rotation without additional hardware.
25M fps Bar Code Reader
PhysOrg: The new imaging technique reported by UCLA postdoctoral fellow Keisuke Goda, graduate researcher Kevin K. Tsia and electrical engineering professor Bahram Jalali uses a phenomenon known as amplified dispersive Fourier transform to read bar codes at a frame rate of 25 MHz — about a 1,000 times faster than current technology.
The new technology, dubbed the CWEETS Scanner (chirped wavelength electronic encoded time domain sampling), first maps the one-dimensional bar code image onto the spectrum of an ultrashort laser pulse and then maps that into an amplitude-modulated waveform that is captured with a single optical-to-electrical converter. This is in stark contrast to typical camera-based bar code readers, which require many optical-to-electrical converters — in other words, an array of pixels — to capture the image. The new imager requires only a single pixel and is free of mechanically moving parts.
To view an animated film illustrating the concept of amplified dispersive Fourier-transform imaging, see http://goda.bol.ucla.edu/barcode (Windows Media).
Now the only remaining thing is to find an application which requires to read barcodes at 25M fps speed.
The new technology, dubbed the CWEETS Scanner (chirped wavelength electronic encoded time domain sampling), first maps the one-dimensional bar code image onto the spectrum of an ultrashort laser pulse and then maps that into an amplitude-modulated waveform that is captured with a single optical-to-electrical converter. This is in stark contrast to typical camera-based bar code readers, which require many optical-to-electrical converters — in other words, an array of pixels — to capture the image. The new imager requires only a single pixel and is free of mechanically moving parts.
To view an animated film illustrating the concept of amplified dispersive Fourier-transform imaging, see http://goda.bol.ucla.edu/barcode (Windows Media).
Now the only remaining thing is to find an application which requires to read barcodes at 25M fps speed.
Toshiba Teli Compares CCD with CMOS
Test & Measurement World: Toshiba Teli, the manufacturer of cameras for machine visions shares its view on CCD and CMOS sensors strengths and weaknesses:
"Today, the major difference in usage between CMOS and CCD cameras for machine vision occurs in situations where light levels are very low. In those cases, you still need CCD sensor technology because it is more sensitive to light: The minimum sensitivity for CCD cameras is 1 lux, but for CMOS it’s 5 to 10 lux.
But if there’s sufficient illumination during inspection, CMOS camera performance is excellent. And that’s the case with the majority of machine-vision applications, since they are usually well lighted. Also, some applications that need a higher near-infrared response, like about 900 to 1000 nm, will require CCD cameras, because those cameras are more sensitive at a higher spectrum.
CMOS sensors are less expensive than their CCD counterparts. So, for some customers, if CMOS technology meets their sensitivity needs, that’s what they will use. But at least as important is the ability of CMOS cameras to increase frame rates by letting users choose a smaller, lower-resolution area within the image."
"Today, the major difference in usage between CMOS and CCD cameras for machine vision occurs in situations where light levels are very low. In those cases, you still need CCD sensor technology because it is more sensitive to light: The minimum sensitivity for CCD cameras is 1 lux, but for CMOS it’s 5 to 10 lux.
But if there’s sufficient illumination during inspection, CMOS camera performance is excellent. And that’s the case with the majority of machine-vision applications, since they are usually well lighted. Also, some applications that need a higher near-infrared response, like about 900 to 1000 nm, will require CCD cameras, because those cameras are more sensitive at a higher spectrum.
CMOS sensors are less expensive than their CCD counterparts. So, for some customers, if CMOS technology meets their sensitivity needs, that’s what they will use. But at least as important is the ability of CMOS cameras to increase frame rates by letting users choose a smaller, lower-resolution area within the image."
Subscribe to:
Posts (Atom)