tag:blogger.com,1999:blog-19092890.post1297783719690940562..comments2024-03-28T17:41:43.970+02:00Comments on Image Sensors World: BAE Fairchild Imaging sCMOS 2.0 FlyerVladimir Koifmanhttp://www.blogger.com/profile/01800020176563544699noreply@blogger.comBlogger21125tag:blogger.com,1999:blog-19092890.post-23194607567685762902014-12-29T03:30:20.321+02:002014-12-29T03:30:20.321+02:00Hamamatsu, on this Page: http://thelivingimage.ham...Hamamatsu, on this Page: http://thelivingimage.hamamatsu.com/resources/will-able-differentiate-1500-molecules-1475-whats-dynamic-range/ says "Using the Orca Flash 4.0 as an example, we divide the full well capacity of 30,000 electrons by the read noise at slow scan, which is 0.9 electrons, to get a dynamic range of 33,000:1, or 90.37 dB." (dB corrected by me)<br /><br />On Page 8 Andor explains this technique: http://www.andor.com/pdfs/literature/Andor_sCMOS_Brochure.pdf <br /><br />"<br />The Andor Neo and Zyla cameras are designed to make use of the innovative dual<br />‘column-level’ amplifier design of the sensors.<br /><br />Traditionally, sensors require that the user must select up-front between high or low amplifier gain (i.e. sensitivity) settings, depending on whether they want to optimize for low noise or maximum well depth. The dual amplifier architecture of the sCMOS sensor circumvents this need, in that signal can be sampled simultaneously by both high and low gain amplifiers. As such, the lowest noise of the chip can be harnessed alongside the maximum well depth, affording widest possible dynamic range of up to 33,000:1. <br />".<br /><br />The transition point (between the choice of low or high gain on each pixel) is calculated by an FPGA and the two imaged are merged.<br /><br />An even longer Technical Note appears on Page 19 of Andor's PDF, along with a Diagram of the Circuit. That gives a more complete answer.<br />Robhttps://www.blogger.com/profile/08222535289392471857noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-86751298198908529272014-02-14T07:45:38.337+02:002014-02-14T07:45:38.337+02:00This seems very similar to the version 2 scmos sen...This seems very similar to the version 2 scmos sensor that Hamamatsu released a year ago. Could it be the same sensor?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-45183234901336406002014-02-11T00:37:16.591+02:002014-02-11T00:37:16.591+02:00it's quite optimal for optical system where yo...it's quite optimal for optical system where you have an fully corrected imaging circle of given size and you want to get max of that circle.Piotr Malińskihttps://www.blogger.com/profile/00054903846537875611noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-18384924582603039592014-02-10T20:46:21.491+02:002014-02-10T20:46:21.491+02:00DR is probably for the pixel, so PRNU is not inclu...DR is probably for the pixel, so PRNU is not included.<br /><br />2e noise is for 100fps. The noise at 30fps may be 1e and then DR=87dB. This is possible as the dark current related shot noise for lower fps is only slightly higher: sqrt(25e/30).<br /><br />This is a polished good old 4T pixel sensor.DMnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-58509545657641608902014-02-10T12:55:07.214+02:002014-02-10T12:55:07.214+02:00ah ! I didn't notice that ! Thanks !
-yang niah ! I didn't notice that ! Thanks !<br />-yang niAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-68567993326073198612014-02-10T07:18:40.004+02:002014-02-10T07:18:40.004+02:00I just do not see any other need for this negative...I just do not see any other need for this negative supply in their spec.Vladimir Koifmanhttps://www.blogger.com/profile/01800020176563544699noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-17868309484277717112014-02-09T20:46:56.744+02:002014-02-09T20:46:56.744+02:00Is there any reason for the square resolution? Is ...Is there any reason for the square resolution? Is this common in scientific cameras? Thanks!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-77526313170358986722014-02-09T20:34:10.986+02:002014-02-09T20:34:10.986+02:00Vladimir, how can you guess the TX voltage is -0.4...Vladimir, how can you guess the TX voltage is -0.4V during integration period ? <br />Thanks !<br /><br />-yang niAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-74872084674666885722014-02-09T11:18:26.966+02:002014-02-09T11:18:26.966+02:00Well, no idea then. As I said, I'm not an expe...Well, no idea then. As I said, I'm not an expert. I was just drinking their kool-aid and passing it onAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-33624731455383014272014-02-09T07:43:42.958+02:002014-02-09T07:43:42.958+02:00You are correct, if there is no PRNU correction yo...You are correct, if there is no PRNU correction you are limited at ~30dB.Herve Hornungnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-84287982531857198512014-02-09T04:10:36.525+02:002014-02-09T04:10:36.525+02:00No price? :)No price? :)Piotr Malińskihttps://www.blogger.com/profile/00054903846537875611noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-33204780294423454042014-02-08T12:24:19.140+02:002014-02-08T12:24:19.140+02:00a couple of definition issues:
-22bit ADC is a wr...a couple of definition issues: <br />-22bit ADC is a wrong notation. They could claim max 16bit (11bit adc + 5bits from the gain of 32 of the gain amplifier). <br />-3% PRNU seems high. That limits the max SNR to 30dB(?). Please correct me if I'm wrongAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-60009117947761503732014-02-08T12:19:58.468+02:002014-02-08T12:19:58.468+02:00DR=20log(30000e/2e)=83.5dB. You can have 32 bit AD...DR=20log(30000e/2e)=83.5dB. You can have 32 bit ADC, whatever..it would not change that, unless you use multiple exposures, LOFIC, or logarithmic (3T kind) pixel which doesn't seem to be the case here..Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-52504935627230886512014-02-08T11:22:24.566+02:002014-02-08T11:22:24.566+02:00They have an equivalent of two 11b ADCs with diffe...They have an equivalent of two 11b ADCs with different gains combined. This gives them a higher DR.Vladimir Koifmanhttps://www.blogger.com/profile/01800020176563544699noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-4133837598296448892014-02-08T11:20:04.862+02:002014-02-08T11:20:04.862+02:00So, for the high gain channel, the noise is 2e-, l...So, for the high gain channel, the noise is 2e-, limited by the pixel, no matter how low the readout channel noise is. And for the low gain channel, the full well is 30,000e-, again limited by the pixel, even if the readout could manage a much larger signal. How do we get the 87db DR then?Vladimir Koifmanhttps://www.blogger.com/profile/01800020176563544699noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-28352034277556494712014-02-08T02:34:36.938+02:002014-02-08T02:34:36.938+02:00With 11bits in the ADC is not possible to codify a...With 11bits in the ADC is not possible to codify a dynamic range higher than 77dB. Only if some non linear transform is used can be codifiedRafahttps://www.blogger.com/profile/07727025214847423336noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-51508960491700249392014-02-08T01:33:59.160+02:002014-02-08T01:33:59.160+02:00One channel applies a bigger gain without increasi...One channel applies a bigger gain without increasing the read-out noise, then the camera maker has to mix the signal from the two channels to get an image with detail both in the highlights and the shadows.<br /><br />The sCMOS v1 had a full well capacity of 30000 e-, read-out noise of 2e- at 30fps and 3e- at 100 fps, and it did reach (as seen on the BlackMagic cameras) the 88dB of DR that they claimed. For the record, QEmax 60%.<br /><br />Some info on the sCMOS v1:<br />http://www.andor.com/learning-academy/dual-amplifier-dynamic-range-scmos-dynamic-range<br />http://www.pco.de/whats-new/high-dynamic-range-imaging-with-scmos-scientific-cmos/<br />http://www.pco.de/fileadmin/user_upload/db/download/scmos_white_paper_2mb_01.pdfAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-22633638667176453522014-02-08T01:02:59.213+02:002014-02-08T01:02:59.213+02:003-4fF, if I have to guess.3-4fF, if I have to guess.Vladimir Koifmanhttps://www.blogger.com/profile/01800020176563544699noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-49885908767121113912014-02-08T00:57:17.360+02:002014-02-08T00:57:17.360+02:00Vladimir, what could be the capacitance of the FD ...Vladimir, what could be the capacitance of the FD in the pixel according to your feeling please ? <br /><br />-yang niAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-77798254822959971452014-02-07T23:20:23.353+02:002014-02-07T23:20:23.353+02:00Whatever the parallel readout paths are, their com...Whatever the parallel readout paths are, their combined DR can't exceed that of pixel.Vladimir Koifmanhttps://www.blogger.com/profile/01800020176563544699noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-26681563754503335412014-02-07T22:59:31.826+02:002014-02-07T22:59:31.826+02:00I think 87dB is what you get by combining the outp...I think 87dB is what you get by combining the output from the two parallel readout circuits.Anonymousnoreply@blogger.com