Marc
no wonder that you have not seen a CMOS that made you happy yet - because there is none made (of the matching size) !
and Bob- thanks for the link- I knew the document , but there again this text part:
" Both CCDs and CMOS imagers can offer excellent imaging performance when designed properly"
and the last sentence say it all:
"CCDs and CMOS will remain complementary. The choice continues to depend on the application and the vendor more than the technology. Teledyne DALSA's approach is "technology-neutral": we are one of the few vendors able to offer real solutions with both CCDs and CMOS."
So the question remains !
Regards
Stefan
It depends on what you mean by excellent and if you care about your images.
Comparing an interline cmos sensor at 11 frames per second is leaning on perhaps its weakest link.
Dual amplifier approaches to attempt to gain wider dr and lower saturation are also applied to traditional ccd applications in some cases.
While that is pretty good it is not a good comparison to an interline ccd at 1 frame per second where the thermal issues are not as severe. CCD's consumption per "read" is somewhere in the order of 100 times larger than cmos.
It reads something like what you see on restaurant menus our "excellent steak" for example, after all they sell both kinds.
CMOS can certainly get you high readout speeds and more circuitry on-chip can yield more function however at an increased design cost and noise level. Let me clarify what I mean by noise. It is the variation of the least significant bits or in signal terms, the available signal in terms of bits of precision. Lots of times, you will see sensors quoted at n-bits of readout per pixel, but analysis demonstrated that maybe only the high order n-2 or 3 are actually any good.
There is certainly a place for cmos and one factor that might drag me there kicking and screaming might simply be the lack of ccd availability once its fabs shut down. For a given physical sensor size you get smoother transitions and gradients with ccd and that is mainly due to its larger well size. Once cmos gets there it still needs to deal with clocking noise which seems evident on many sensors I have examined. This is the most annoying sort.
So yes I believe (don't know for sure) that there will be improvements to a certain extent but we are nowhere near a crisis point and nowhere near a crisp demonstration of fact that cmos is the equal of ccd for imaging applications. I note that many a machine vision application has moved on to commodity cmos sensors due to their cost and where their performance allows, but the high end stuff which incidentally includes the images that inspect cmos wafers iin-process as well as astronomical applications are almost exclusively ccd.
So you can take your physics be damed point of view that economics rules all which might be true when ccd just gets too expensive and the top 1% of those needing quality decide that it is just too expensive and are overwhelmed by the numbers who find mediocre excellent enough.
Now enough of this foolishness, lets get on with showing some images.
-bob