Oh and btw, that s100 sensor was just AN EXAMPLE of a sensor with tiny pixel. I just said Canon has the technology to make high MP sensor. And unless that 120MP sensor was just a fake prop, which I doubt, they do have the technology.
I can guarantee you those sensors were not made with the same process as the S100 sensor.
As someone else said, the 120 MP protopye actually performed not so well. That is what we mean when we say they can't go any higher with the current large sensor fabs. The performance penalty would be too high.
theSuede has earlier said that Canon uses a 350 nm process for their large sensors, and there are no signs in the industry that they are rolling in newer equipment.
There are generally needed 4-6 line widths to support each photosite, in addition to the size of the photosite. That makes 1.4 - 2.1 um dead space between each photosite. This may work with pixel pitches in the 4 um range, but not much smaller.
alundeb wrote:
I can guarantee you those sensors were not made with the same process as the S100 sensor.
As someone else said, the 120 MP protopye actually performed not so well. That is what we mean when we say they can't go any higher with the current large sensor fabs. The performance penalty would be too high.
And the 300 mm sensor has huge pixels.
+1 very true
The performance penalty would be too high like you wrote.
alundeb wrote:
I can guarantee you those sensors were not made with the same process as the S100 sensor.
As someone else said, the 120 MP protopye actually performed not so well. That is what we mean when we say they can't go any higher with the current large sensor fabs. The performance penalty would be too high.
And the 300 mm sensor has huge pixels.
1st of all, get that S100 out of your head, that was just an example of Canon being able to cram tiny pixels in their sensor. I could use those phone camera sensor as an example as well. I don't know the size of the pixels in those tiny 8MP phone cameras, probably even smaller.
Now the context in which I wrote my post is that Canon has the ability to physically build a sensor, whether large or small, with tiny pixel. Of course larger sensor with the same tiny pixels would mean a much higher resolution.
With the 120MP prototype sensor, we know Canon has the ability to build a large sensor with pixels AT LEAST 2 micron in size.
Whether they perform well or not, that's besides my point.
As far as I remember this thread was based on "retros" link to a CR1 rumor about an actual fictive EOS 3D body. I know you all are genius (like me). But instead of personal believes of what Canon can do or what they can not, I would like to hear some more information about the OPs topic. Did anybody hear something new concerning this?
Camera manufacturers (and manufacturers of many other technology products) decide when to turn various technologies into shipping products for a whole range of reasons including pricing, costs of production versus potential increased sales, marketing, how they view the current competition, what they might hold in order to compete better later on, how the technology will affect their current product mix (to avoid "cannibalization" of their own products), and much more.
Ralph Conway wrote:
As far as I remember this thread was based on "retros" link to a CR1 rumor about an actual fictive EOS 3D body. I know you all are genius (like me). But instead of personal believes of what Canon can do or what they can not, I would like to hear some more information about the OPs topic. Did anybody hear something new concerning this?
Ralph
I am not aware of any other news since in regard to my OP. This thread has become somehow a main thread about high MP sensor technology and feasibility which is somehow related.....at least this allows us later in a couple of months to look back and see if any of those opinions stated here became true or not . No matter if one is right or not, we all share the same interest in photography and also technology! And this is a good thing ! Great discussions so far!
alundeb wrote:
The S100 sensor cannot be scaled to FF sizes for technical reasons.
Nobody said that the S100 sensor itself will be scaled to FF sizes.
This is a naive interpretation of what I said earlier.
To iterate my point:
The S100 sensor is quite obviously *not* made on the 300nm (350nm?) process, on which all other Canon sensors are still being made.
The significance of this fact is that Canon already has a non-300nm process in production.
Don't know if this is a 180nm process or something else but it's a non-300nm process for sure.
Historically, whenever Canon has implemented certain process improvements, they'd lock their process and make multiple sensor/pixel sizes based on this process.
For example, the 1DII and the original 5D, 1DIII and 40D, 1DIV and 7D, etc..
So, based on historical data, it is safe to assume that S100 process will be locked at some point (if not locked already) and Canon will start using this process for making sensors of larger sizes.
Let me clarify this again:
Canon will certainly not scale the S100 sensor to a FF sensor with 100+ megapixels (or whatever). It is indeed naive to think that.
But based on what Canon has been doing consistently in the past, it is quite certain that they will use the S100 process to make next generation 22-50mp FF sensors and 18-24mp 1.6x sensors.
You can be quite certain of that. Just think about it:
To make the S100 sensor, Canon must have spent 400+ USD millions to build the required facilities.
Do you think that they spent all this money to build sensors for low-margin P&S camera
When they can get these sensors - and are in fact gertting some of them - from Sony? For cheap.
The S100 sensor is the testbed for their next generation process, which will be scaled to FF and 1.6x.
(Again, they are not scaling the S100 sensor itself but the CMOS process, on which the S100 sensor is made).
If the line widht gets too small, and the process is intended for small chips, the fail rate may be too high for making large chips. Large sensors need the right line width, not too small and not too large.
Are you positive that the process used for the S100 sensor gives high enough yield to get even one working FF sensor out of a wafer?
alundeb wrote:
Are you positive that the process used for the S100 sensor gives high enough yield to get even one working FF sensor out of a wafer?
Yields are confidential information, which I certainly don't have.
Consider this, though:
The S100 sensor is not better than the Sony BSI sensors of the same size. In fact, it's worse.
So, manufacturing the S100 sensor in-house does not give Canon any performance advantage.
It's hard to believe that it gives them a cost advantage either, as Sony's BSI sensors of the same size sell for $5-15 USD.
So, if manufacturing the S100 sensor has neither cost, nor performance advantages over the Sony sensors, why did Canon decide to make the S100 sensors in-house?
We can speculate that Canon's board of directors has secretly hired Kodak's ex-management as advisors .
Or, that Canon has other plans on their minds for using the S100 sensor tech.
Did you know that Canon is big in DSLRs?
It is still not as easy as to just scale to larger sensors. The optics for the lithography with fine resolution are not able to illuminate a large enough area for FF sensors. They at least need to change steppers and a whole lot of other things. But maybe not build a new plant.
Again, don't think so much in terms of the fab's resolution capacity...I really, really, don't think this is the issue for raising the megapixels. Going from 22mp, to say...45mp...on a full frame sensor...is such a small increase in resolution required for the litho...however this doesn't even matter, because:
Canon's existing equipment already cannot expose something the size of a full frame sensor with only one exposure. Their existing equipment can already handle scaling up the 18mp crop sensor sized pixels to full frame...
I think the issue people should be asking is: Can Canon's process tech (NOT the machines themselves that Canon uses, Canon makes steppers capable of pretty modern digital logic) handle combined digital and analog logic on a chip? If not, what is Canon going to do to get their analog output chips up to snuff, DR wise (I think the can choose to do this, it just will be more difficult and pricey).
Does Canon even produce any digital logic in house? Does anyone know, or have evidence of this? I don't believe their digic chips are made in house. I also think their ADC chips are off the shelf components and not custom in house stuff. Anyone have any evidence of the contrary?
Lets face reality here, a 40mp DSLR from Canon with the same 11 stop DR is not going to cut it...they need to get the DR up, Canon didn't get where they are today by being stupid--they know this, too.
All this wild speculation about Canon's ability to fabricate a particular sort of sensor may be fun, but it bears little relationship to reality nor to what will actually happen, and it is largely uninformed.
I really looking forward to see what Canon may bring out... especially now that we are a few that beleives that the ff lines at Canon are NOT capable of producing the fine geometry that's needed. This time I really want to be wrong... Bring 'er on Canon
.