Conventional CMOS image sensors equipped with global shutter function temporarily store charge signals in the memory area located next to the photodiode to resolve image distortion (focal plane distortion) caused by the time shift due to the row-by-row readout. In FSI CMOS sensors, there is a wiring layer on the silicon substrate forming the photodiode, and with such a structure, the benefit is that it is easy to form a light shield for protecting the charge signal temporarily stored in the memory area from leaked light. For this reason, conventional CMOS image sensors with global shutter function have adopted a FSI pixel structure. However, the wiring on top of the photodiode hinders the incident light, which creates an issue when attempting to miniaturize the pixels.
In response to this, Sony has developed a proprietary pixel structure that achieves the global shutter function on a BSI structure that has superior sensitivity characteristics. Normally, when pixels are miniaturized, the sensitivity and saturation characteristics deteriorate, but the new Sony technology enables a reduction in pixel size to 2.74 μm while maintaining performance of those characteristic, thereby achieving about 1.7 times higher resolution than conventional FSI CMOS sensors. Moreover, thanks to the high degree of freedom of the wiring layout of BSI pixel structures, a high speed of about 2.4 times that of conventional sensor can be achieved. In addition, the sensor’s stacked structure makes it possible to mount various signal processing circuits, whereby it is possible to realize smart functions such as signal processing only for the necessary part of the measurement and inspection images in a smaller size compared to conventional sensors.
Going forward, Sony will work to develop products equipped with this stacked CMOS sensor employing its proprietary global shutter function with BSI pixel structure for various industrial applications and intelligent transportation systems, including development of derivatives for signal processing circuits to be mounted. Sony plans to start shipping sample units in the summer of 2019 or later.
This is the first announcement of a back-illuminated, stacked sensor in the Pregius ("premium global shutter") product line. This product line target machine vision, intelligent traffic systems, and global shutter video.
ReplyDeleteThe 2.74 µm pixel size is characteristic of the 4th generation Pregius sensors, which were already expected to be back-illuminated. Now it seems that the confirmation of a stacked machine vision sensor has been announced by Sony for the first time. However, there were already "future" directions indicated by Sony at our FRAMOS Tech Days event in early February in San Francisco.
Full disclosure: My employer, the FRAMOS Group, is distributor of Sony industrial image sensors in Europe and North America.
ISW blog becomes more and more a marketing-only platform, such comments should be blocked the next time...
DeleteI am sensitive to that concern. Hence the "full disclosure".
DeleteOn the other hand, perhaps we should cut off Anonymous comments? ;-)
Actually, I block a lot of spam comments well before you see them. Most of them are various marketing and commercial messages. The majority of these spammers have registered names at Blogger. Blocking of anonymous does not help
DeleteI think this comment is too marketing purpose and should be deleted from this article.
DeleteBeside that, we are looking forward to the 1st sensor to be released in this family from Sony! ^^
RIP to all other global shutter design houses/foundries. This is market killer, 5 years ahead of competition
ReplyDeletesometimes sensors are good enough for a certain application and you dont need to have "5 years ahead" technology. i personally prefer the solution with a price/performance match for my application over "the best". Image sensors for machine vision (this is one of the main target markets for this family) are remarkably often "good enough". I think people overestimate the datasheet values when they prefer a x e- dark current sensor over a worse y e- sensor - in most cases this value is irrelevant for machine vision applications when there is enough light and short exposure times. Price is a concern above bare technical specs. Why should I pay for the crazy interface speed of this sensor if I connect a camera over a cable? even for 'highspeed' interfaces like Coaxpress, the G2 is fast enough. if Sony puts the price tag too high, this will leave a lot of market for other vendors (e.g. Gmax or Emerald sensors competing for similar markets, and this sensors are also quite good) - at least thats what i think. In my application the Pregius S does not offer that much more 'real advantage' over the gen 2 that would justify to replace the gen2 cameras. One main argument maybe the larger acceptance angle... but without doubt, the Pregius S is technologically a remarkable, impressive image sensor generation that will have success.
DeleteThis comment has been removed by the author.
ReplyDeleteBeautiful technology (= understatement), the pixel with the cat flap (as I call it) to move the charges from the PD to the memory node. This technology was also presented at IEDM'18. Amazing what you can do with the DTI filled with metal.
ReplyDeleteWonder dark current level on MEM node ?
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteCan anyone comment on the shutter efficiency of this sensor? For our applications, we typically look for over 1/40,000.
ReplyDeleteI think that this is like super-heterodyne receivor in RF, after the invention of this architecture by Armstrong, RF circuits performance improvement becomes balistically predictable. So the CMOS image sensor should be in the truck too !
ReplyDelete