Sign in with
Sign up | Sign in
Your question

dynamic range of digital image sensors

Last response: in Digital Camera
Share
Anonymous
April 5, 2005 7:23:03 AM

Archived from groups: rec.photo.digital (More info?)

Hello Ng,

My questions regards the dynamic range of digital image sensors (CCDs
and CMOS). I understand that this is mostly limited by the size of the
capacitor / well that holds the electrons and the amount of electronic
noise produced by the sensor. But if the capacitor size is a limiting
factor, why don´t the manufacturers use larger ones ? Is there a
special relationship between the size of the capacitor and the size of
of each pixel that keeps them from doing so ?

Thanks for your input!
Mr.Adams
Anonymous
April 5, 2005 11:29:40 AM

Archived from groups: rec.photo.digital (More info?)

"Mr.Adams" <mr.adams@gmx.net> wrote in message
news:ffeb7fc.0504050223.444ef308@posting.google.com...
> Hello Ng,
>
> My questions regards the dynamic range of digital image sensors (CCDs
> and CMOS). I understand that this is mostly limited by the size of the
> capacitor / well that holds the electrons and the amount of electronic
> noise produced by the sensor. But if the capacitor size is a limiting
> factor, why don´t the manufacturers use larger ones ? Is there a
> special relationship between the size of the capacitor and the size of
> of each pixel that keeps them from doing so ?
>
> Thanks for your input!
> Mr.Adams

There is only so much room available at the photosites on the sensor. When
they pack, say, 4 mp on the same sized sensor as a 3 mp sized sensor,
everything must be made smaller, thus noise and dynamic range become a
problem. Many new cameras do okay in the noise department due to noise
removal algorithms, but that is not a cure all - especially at higher ISOs.

DSLRs use a large sensor and the sensor is designed with less electronics at
each photosite so you get amazingly low noise at very high ISOs. Due to this
design (not to mention the swinging mirror being in the way), the sensor
can't do live video. The images look smooth and unprocessed and have better
dynamic range.
bg
Anonymous
April 5, 2005 12:40:14 PM

Archived from groups: rec.photo.digital (More info?)

Mr.Adams wrote:

> Hello Ng,
>
> My questions regards the dynamic range of digital image sensors (CCDs
> and CMOS). I understand that this is mostly limited by the size of the
> capacitor / well that holds the electrons and the amount of electronic
> noise produced by the sensor. But if the capacitor size is a limiting
> factor, why don´t the manufacturers use larger ones ? Is there a
> special relationship between the size of the capacitor and the size of
> of each pixel that keeps them from doing so ?
>
> Thanks for your input!
> Mr.Adams


Because of a limitation on the number of layers in IC processing, the
area of the capacitor is directly proportional to the area of the pixel.
If we keep the overall size of sensor chips the same (silicon real
estate is expensive) and shrink pixel size to get more pixels on the
chip, then the size of the capacitor shrinks also, making a smaller
capacitance.

Note that the size of the capacitor is only one of many potential noise
sources. Whether the capacitance is the limiting noise factor on not is
a tough question for a particular sensor. However, if we reduce other
sources of noise, the capacitance issue will indeed eventually bite us.
Related resources
Anonymous
April 5, 2005 1:35:36 PM

Archived from groups: rec.photo.digital (More info?)

The size of the capacitor is a volume problem. The area of the pixel
fixes two of the dimensions, the ability to create deep diffusions
fixes
the other dimension. If one tries to make a really deep capacitor, one
runs into the problem that one cell might short out to its neighbor.

So, as pixels shrink in area, the capacitance shrinks in volume.

Mitch
Anonymous
April 5, 2005 3:03:44 PM

Archived from groups: rec.photo.digital (More info?)

Mr.Adams wrote:
> Hello Ng,
>
> My questions regards the dynamic range of digital image sensors (CCDs
> and CMOS). I understand that this is mostly limited by the size of the
> capacitor / well that holds the electrons and the amount of electronic
> noise produced by the sensor. But if the capacitor size is a limiting
> factor, why don´t the manufacturers use larger ones ? Is there a
> special relationship between the size of the capacitor and the size of
> of each pixel that keeps them from doing so ?
>
> Thanks for your input!
> Mr.Adams

There are limits to the capacitor area directly related to the pixel size.
If you could capture more photo-electrons in the well, you would get an
improved signal-to-noise ratio, but needing more photon would result in a
decreased sensitivity. We already see sensitivities down to ISO 50 in
some 8MP cameras - do we want lower?

Cheers,
David
Anonymous
April 5, 2005 6:20:57 PM

Archived from groups: rec.photo.digital (More info?)

"Not sure if I understand this. With DRAM we are at the point where
they
make vertical capacitors, because if they made a simple horizontal
capacitor, this would not be able to store enough charge. Why can't
they
make vertical capacitors also for CCDs ? "

Verticle electron storage blocks the incomming light! This would not be
an
issue with back illuminated sensors.
Anonymous
April 6, 2005 12:13:26 AM

Archived from groups: rec.photo.digital (More info?)

In article <kgu4e.4269$G8.1906@text.news.blueyonder.co.uk>, David J
Taylor says...

> There are limits to the capacitor area directly related to the pixel size.
> If you could capture more photo-electrons in the well, you would get an
> improved signal-to-noise ratio, but needing more photon would result in a
> decreased sensitivity. We already see sensitivities down to ISO 50 in
> some 8MP cameras - do we want lower?

Well no, sensitivity is not decreased, because you don't need more
electrons. With a larger well you essentially have more dynamic range
(it takes longer before each pixel saturates).
--

Alfred Molon
------------------------------
Olympus 4040, 5050, 5060, 7070, 8080, E300 forum at
http://groups.yahoo.com/group/MyOlympus/
Olympus 8080 resource - http://myolympus.org/8080/
Anonymous
April 6, 2005 12:13:27 AM

Archived from groups: rec.photo.digital (More info?)

Alfred Molon wrote:
> In article <kgu4e.4269$G8.1906@text.news.blueyonder.co.uk>, David J
> Taylor says...
>
>> There are limits to the capacitor area directly related to the pixel
>> size. If you could capture more photo-electrons in the well, you
>> would get an improved signal-to-noise ratio, but needing more photon
>> would result in a decreased sensitivity. We already see
>> sensitivities down to ISO 50 in some 8MP cameras - do we want lower?
>
> Well no, sensitivity is not decreased, because you don't need more
> electrons. With a larger well you essentially have more dynamic range
> (it takes longer before each pixel saturates).

With a constant pixel area and lens f/number, the photon flux is constant.
To get the improved SNR, you need to collect more photons, therefore you
need to expose for a longer time. I.e. to take advantage of a bigger
well, and get a better SNR, you need to reduce sensitivity. Simply
getting a larger dynamic range may only help you capture specular
highlights better.

Cheers,
David
Anonymous
April 6, 2005 12:15:37 AM

Archived from groups: rec.photo.digital (More info?)

In article <1112718936.188191.47470@l41g2000cwc.googlegroups.com>,
MitchAlsup@aol.com says...
> The size of the capacitor is a volume problem. The area of the pixel
> fixes two of the dimensions, the ability to create deep diffusions
> fixes
> the other dimension. If one tries to make a really deep capacitor, one
> runs into the problem that one cell might short out to its neighbor.

Not sure if I understand this. With DRAM we are at the point where they
make vertical capacitors, because if they made a simple horizontal
capacitor, this would not be able to store enough charge. Why can't they
make vertical capacitors also for CCDs ?
--

Alfred Molon
------------------------------
Olympus 4040, 5050, 5060, 7070, 8080, E300 forum at
http://groups.yahoo.com/group/MyOlympus/
Olympus 8080 resource - http://myolympus.org/8080/
Anonymous
April 6, 2005 4:15:56 AM

Archived from groups: rec.photo.digital (More info?)

In article <1112736057.898236.305320@z14g2000cwz.googlegroups.com>,
MitchAlsup@aol.com says...

> Verticle electron storage blocks the incomming light! This would not be
> an
> issue with back illuminated sensors.

Then let's make a vertical storage back-illuminated CCD - more
sensitivity and dynamic range!
--

Alfred Molon
------------------------------
Olympus 4040, 5050, 5060, 7070, 8080, E300 forum at
http://groups.yahoo.com/group/MyOlympus/
Olympus 8080 resource - http://myolympus.org/8080/
Anonymous
April 6, 2005 11:08:34 AM

Archived from groups: rec.photo.digital (More info?)

"Mr.Adams" <mr.adams@gmx.net> wrote in message
news:ffeb7fc.0504050223.444ef308@posting.google.com...
> Hello Ng,
>
> My questions regards the dynamic range of digital image sensors (CCDs
> and CMOS). I understand that this is mostly limited by the size of the
> capacitor / well that holds the electrons and the amount of electronic
> noise produced by the sensor. But if the capacitor size is a limiting
> factor, why don´t the manufacturers use larger ones ? Is there a
> special relationship between the size of the capacitor and the size of
> of each pixel that keeps them from doing so ?
They do in fact make sensors with larger sensels (each individual sensor
element), and they do deliver better signal/noise ratios. But there are some
drawbacks:
The overall sensor size equals sensel size*megapixels, so in the push to get
more megapixels, either the overall sensor size must be larger, or the
individual sensel size must be smaller.
Larger sensor size = larger lens=bigger heavier camera=(generally) less
market appeal.
Just like all integrated circuits, the larger the sensor, the lower the
yield of acceptibly good chips, hence manufacturing costs climb
significantly.
Larger sensels require more light to saturate them, hence they have a lower
native ISO equivalence. So while a sensor made with larger sensels may
deliver exceptional SN ratio, it might be at it's best at ISO 25
equivalence. Once amplification has been applied to deliver ISO 100 or
higher equivalence, the SN ratio will have dropped, and may not be
significantly better than a smaller sensel that has a native ISO 100
equivalence.
Thus, the actual sensor chips used are at a size where the manufacturer
feels they have the best trade-off between SN ratio on the one hand vs cost
and overall size on the other hand.

>
> Thanks for your input!
> Mr.Adams
!