The Amazing D200.

Started Jun 13, 2013 | Discussions thread
Cytokine
Contributing MemberPosts: 626
Like?
Re: Scientists not convinced by CMOS develop Scientific-cmos
In reply to olliess, Jul 24, 2013

olliess wrote:

Cytokine wrote:

olliess wrote:

I think it's hard to compare specialized scientific imaging sensors with those produced for the professional (and consumer) DSLR market. The cost and usage constraints are just way different.

For example, I think that most DSLR users would NOT be interested in a $10,000, 16-MP astronomical CCD that requires water cooling for best performance.

Oliess, I posted these quotations and link, because the two sensors are almost diametrically opposed in the way they work, therefore it is not surprising that some people can see a difference.

John,

I appreciate the link. It's an interesting read so far (although I admit to not being able to finish it in one go. )

It seems to me that the sensors could only be described as working in "diametrically opposed" ways from a certain perspective. After all, the main physical principle of both sensors types is the same: photons are received in a photosensitive area, which converts the photons to charges, and then the charges need to be counted. The major working difference is between moving the charges off-sensor first to count them, or doing it on the spot.

IMHO, it would be somewhat surprising for two modern commercial sensors, under normal operating conditions, to show such obvious differences due to sensor type that people could reliability tell which was which.

People said the same thing about tube amplifiers and transistors too, if you want to open up a whole can of worms.

Actually the two sensor types compliment each other, and allow photographers to choose the right tool for the job in hand.

Only if there's a job that one sensor can do and the other demonstrably cannot. Otherwise, photographers might want to just pick the best sensor available to them and not worry about CMOS vs CCD.

This is also the case in some scientific applications where CMOS is used to differentiate between high level light signal analysis, and CCD for low level light signal analysis, both complimenting each other.

CCD has better pixel dynamic range and CMOS better system dynamic range (given enough light).

I think these are overgeneralizations. Is this true for every CCD and every CMOS sensor?

To go back to my original point, the constraints are totally different between scientific applications and a mainstream DSLR. Even if there is a CCD which is measurably better in every way than the CMOS used in my camera, it doesn't do me any good if I can't afford it (or the manufacturer can't afford to put it in their camera, without going out of business anyway).

Any, that's just my $0.02. I've owned for many years (and continue to use) a D200, so I'm definitely not saying this just because I'm "too good for CCDs" or something.

Olliess,

" Dynamic range, the ratio of a pixel’s saturation level to its signal threshold. It gives CCDs an advantage by about a factor of two in comparable circumstances. CCDs still enjoy significant noise advantages over CMOS imagers because of quieter sensor substrates (less on-chip circuitry), inherent tolerance to bus capacitance variations and common output amplifiers with transistor geometries that can be easily adapted for minimal noise. Externally coddling the image sensor through cooling, better optics, more resolution or adapted off-chip electronics cannot make CMOS sensors equivalent to CCDs in this regard".

https://www.teledynedalsa.com/public/corp/Photonics_Spectra_CCDvsCMOS_Litwiller.pdf

This was written in 2001 and to some extent it is just as relevant today, as Cmos sensor real-estate now looks like New-York New-York. they are smaller with even more on-chip circuitry, relying on ever smaller micro lenses to bring some light to the light sensitive area.

With micro lenses came other side effects: limited light acceptance angles=Stealth ISO increases with wide aperture lenses.

With on chip A/D conversion came difficult to get colours. (No analogue colour preconditioning) (blowing the red channel etc.,).

However CMOS is a reasonable compromise if you want all the bells and whistles.

John

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
YesNew
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow