larger sensors

[rainer_v]

althoug it seems to be a question of the time ( over-next generation of sensors ? ) that the sensor sizes could increase to 6x6cm , wouldnt it be more logic and maybe even more usefull to enlarge future sensors to real 4,5x6 format of 41x56mm instead 36x48mm?
the size would be 1,35 bigger, all systems and cameras would be remain still usuable.

[rethmeier]

That's were the Hy6 will be unique!
I will be able to accept a large square sensor!
Cheers,
Willem.

[BJL]

That's were the Hy6 will be unique!
I will be able to accept a large square sensor!
Cheers,
Willem.
[a href=\"index.php?act=findpost&pid=91522\"][{POST_SNAPBACK}][/a]

Compatibility with an imagined future product that will almost certainly never exist (i.e. 56x56mm sensors, or even anything bigger than 37x49mm) seems a poor reason to put up with imposing a heavy 42% or more (area) crop on all lenses, not to mention using a mirror and viewfinder that are far bigger and heavier than necessary in order to accommodate the obsolescent 56x56mm film frame. Ironically, even the first film back for the Hy6 will be 645, not 6x6.

Will some people never recognize that new digital camera bodies, lenses, viewfinder systems and such are adapting to the new format sizes that make most sense with electronic sensors, rather than sensor makers striving to improve compatibility with old lenses and bodies? Kodak and Dalsa have apparently make it fairly clear that their sensors sizes for "medium format" cameras have topped out, at 37x49mm and 36x48mm respectively.

There is no reason to think that future chip fabrication equipment will be "upsized" to more easily or economically handle super-large chips, meaning anything bigger than about APS sensors, already larger than most micro-processors and such. On the contrary, the market for chip fabrication equipment is dominated by products other than DSLR sensors, in which the dominant trend is to smaller chip sizes, for the sake of faster operation, lower power consumption and lower cost.
DSLR sensors tend to use "trailing edge" fabrication technology, with larger feature sizes and larger chip sizes than most other applications, but the fabrication equipment does not last for ever, and so eventually fabrication will have to move to equipment designed for the generally smaller sizes of newer chips, potentially making super-large sensors even more difficult to produce.

[eronald]

As someone trained in semiconductor design, albeit a long time ago, I strongly disagree with the contents of the quoted post. Historically, wafer sizes and chip sizes have steadily got bigger. I expect sensors to get bigger, the only question is when.

Or maybe sensors will only get bigger in China, Europe and Japan, with those in the US getting smaller

By the way, I sign with my real name ...

Edmund Ronald, Ph.D

Compatibility with an imagined future product that will almost certainly never exist (i.e. 56x56mm sensors, or even anything bigger than 37x49mm) seems a poor reason to put up with imposing a heavy 42% or more (area) crop on all lenses, not to mention using a mirror and viewfinder that are far bigger and heavier than necessary in order to accommodate the obsolescent 56x56mm film frame. Ironically, even the first film back for the Hy6 will be 645, not 6x6.

Will some people never recognize that new digital camera bodies, lenses, viewfinder systems and such are adapting to the new format sizes that make most sense with electronic sensors, rather than sensor makers striving to improve compatibility with old lenses and bodies? Kodak and Dalsa have apparently make it fairly clear that their sensors sizes for "medium format" cameras have topped out, at 37x49mm and 36x48mm respectively.

There is no reason to think that future chip fabrication equipment will be "upsized" to more easily or economically handle super-large chips, meaning anything bigger than about APS sensors, already larger than most micro-processors and such. On the contrary, the market for chip fabrication equipment is dominated by products other than DSLR sensors, in which the dominant trend is to smaller chip sizes, for the sake of faster operation, lower power consumption and lower cost.
DSLR sensors tend to use "trailing edge" fabrication technology, with larger feature sizes and larger chip sizes than most other applications, but the fabrication equipment does not last for ever, and so eventually fabrication will have to move to equipment designed for the generally smaller sizes of newer chips, potentially making super-large sensors even more difficult to produce.
[a href=\"index.php?act=findpost&pid=91669\"][{POST_SNAPBACK}][/a]

[howiesmith]

Compatibility with an imagined future product that will almost certainly never exist (i.e. 56x56mm sensors, or even anything bigger than 37x49mm) seems a poor reason to put up with imposing a heavy 42% or more (area) crop on all lenses, not to mention using a mirror and viewfinder that are far bigger and heavier than necessary in order to accommodate the obsolescent 56x56mm film frame. Ironically, even the first film back for the Hy6 will be 645, not 6x6.

Will some people never recognize that new digital camera bodies, lenses, viewfinder systems and such are adapting to the new format sizes that make most sense with electronic sensors, rather than sensor makers striving to improve compatibility with old lenses and bodies? Kodak and Dalsa have apparently make it fairly clear that their sensors sizes for "medium format" cameras have topped out, at 37x49mm and 36x48mm respectively.

There is no reason to think that future chip fabrication equipment will be "upsized" to more easily or economically handle super-large chips, meaning anything bigger than about APS sensors, already larger than most micro-processors and such. On the contrary, the market for chip fabrication equipment is dominated by products other than DSLR sensors, in which the dominant trend is to smaller chip sizes, for the sake of faster operation, lower power consumption and lower cost.
DSLR sensors tend to use "trailing edge" fabrication technology, with larger feature sizes and larger chip sizes than most other applications, but the fabrication equipment does not last for ever, and so eventually fabrication will have to move to equipment designed for the generally smaller sizes of newer chips, potentially making super-large sensors even more difficult to produce.
[a href=\"index.php?act=findpost&pid=91669\"][{POST_SNAPBACK}][/a]

Speculating on what will be is risky.  A mere 20 years ago, no one had ever heard of a digital camera.  Who knows where the future will be.  Bigger sensors?  Maybe even something newer than digital.

A couple hundred years ago, the head of the US Patent Office thought the department should be abolished because everything had been invented.  In the '50s, the head of IBM thought the world could support 1 0r 2 big computers.

A real probelm with adapting to existing formats is "How can APS be called 35mm?"  Camera crop factors.  36x48mm is full frame what?  Certainly not "medium format" film, which seems to be the allusion they expect consumers to buy.

Currently, the size of sensors is limited by the number that sensors thta can be fit economically onto a single wafer.  Odd and/or large sizes may create a lot of waste.

[pss]

the more sensors are sold, the more money will be spent on developing newer, better and probably also larger sensors....
a look at canon says it all...if canon stays with "35mm" the will have to lower their prices....they pretty much cannot cram more pixels onto the existing chip, maybe 22mpix, but that's it and even that will have to come with a pricetag.....would i pay 20000 for a 22mpix 12bit (or even 16bit) camera system with a small finder and lenses already pushed to the limit with 16mpix sensors?
with prices for DMF coming down and a P21 and P30 available at pretty much competing prices (for the whole system) and much better file quality?
i think that the DMF market is working out some kinks right now, but someone will survive, there will always be a market there....and there are already much larger chips available, for military and such...so it is only a matter of time when it will come into the photo pro marketplace...
when is the logical update to the P45+? the P55++?kodak will reach a limit just like canon with how many pixels fit...and if you reach that limit...you make the sensor bigger...there is rom to grow for the next 5-10 years...and there are a lot of RZs out there...and don't forget that a lot of superhigh end editorial, advertising and art photographers still shoot film BECAUSE they want the size and they feel that there is room for improvement in quality as well.....there always is....
bytheway: the aptus and emotion lines are numbered not for their mpix, but for the diagonal of the sensor....A65, A75,??? and they both use dalsa chips and they both started this at the same time....maybe dalsa is closer to the 85 then we think?
one thing is for sure: if anyone comes out with a larger sensor within the next 2 years, hasselblad really screwed themselves by making the finder and lenses smaller to accommodate the smaller chips....

[Steve Kerman]

There is no reason to think that future chip fabrication equipment will be "upsized" to more easily or economically handle super-large chips, meaning anything bigger than about APS sensors, already larger than most micro-processors and such. On the contrary, the market for chip fabrication equipment is dominated by products other than DSLR sensors, in which the dominant trend is to smaller chip sizes, for the sake of faster operation, lower power consumption and lower cost.
DSLR sensors tend to use "trailing edge" fabrication technology, with larger feature sizes and larger chip sizes than most other applications, but the fabrication equipment does not last for ever, and so eventually fabrication will have to move to equipment designed for the generally smaller sizes of newer chips, potentially making super-large sensors even more difficult to produce.
[a href=\"index.php?act=findpost&pid=91669\"][{POST_SNAPBACK}][/a]

Huh?  Current medium format sensors are being fabricated on 150mm wafers.  That's really old technology.  Even 200mm wafers are rather dated.  Current technology is using 300mm wafers.

The trend has not been to smaller chip sizes.  It has been to denser chips.  But the biggest devices have been getting bigger for many years.  An 8086, as I recall, had around 30,000 transistors on it; the latest generation of parts have 100s of millions of transistors.  They do this by making them both denser and bigger.

It is certainly plausible that Dalsa and/or Kodak could produce larger sensors as they get hand-me-down 200mm and 300mm processing equipment.

[joern_kiel]

In 1994 i was presenting the new KODAK DCS 460 camera for Kodak at Photokina. Long time ago. At this time i started digital photography with a Nikon based KODAK DCS 200 with a filter wheel for three shot.

So i was in contact with some technicians of Kodak and i asked them for larger sensors, i.e. 4x5 inch. They told me that they have already produced 3 very large sensors for the U.S. Army for the use inside of spy sattelites. But they would be so expensive that no photographer could afford them in the next 20 years. I don´t know if that really was true but i believe that they exist today.

In 2014 i will ask Kodak again for the price of the chip ;-)

jørn

[eronald]

Huh?  Current medium format sensors are being fabricated on 150mm wafers.  That's really old technology.  Even 200mm wafers are rather dated.  Current technology is using 300mm wafers.

The trend has not been to smaller chip sizes.  It has been to denser chips.  But the biggest devices have been getting bigger for many years.  An 8086, as I recall, had around 30,000 transistors on it; the latest generation of parts have 100s of millions of transistors.  They do this by making them both denser and bigger.

It is certainly plausible that Dalsa and/or Kodak could produce larger sensors as they get hand-me-down 200mm and 300mm processing equipment.
[a href=\"index.php?act=findpost&pid=91706\"][{POST_SNAPBACK}][/a]

I think the CCD technology is basically getting older, but I would expect a change over to CMOS in medium format around the next generation. What this will do to image quality is anybody's guess.

Edmund

[BJL]

Current medium format sensors are being fabricated on 150mm wafers.  That's really old technology.  Even 200mm wafers are rather dated.  Current technology is using 300mm wafers.

The trend has not been to smaller chip sizes.  It has been to denser chips.
[a href=\"index.php?act=findpost&pid=91706\"][{POST_SNAPBACK}][/a]

As someone trained in semiconductor design, albeit a long time ago, I strongly disagree with the contents of the quoted post. Historically, wafer sizes and chip sizes have steadily got bigger.[a href=\"index.php?act=findpost&pid=91674\"][{POST_SNAPBACK}][/a]

Firstly, wafer size has nothing to do with the chip size trend that I was talking about. Clearly it is easy it fit a far larger sensor than current ones onto a wafer, but a dominant limitation on size and cost is the reticle size of the steppers (or the newer fab. options, scanners or stepper-scanners): the maximum size of a chip that can be made in a normal single exposure process. A good hint is the sizes of the chips that are being made and envisioned, which surely influences the capabilities of future stepper/scanner designs.

What is the recent trend in chip sizes? Look for example at the size of the recent dominant microprocessors from Intel and AMD: the trend is to both denser and smaller with the shift to smaller feature sizes such as with 65nm process. Even with the move to two cores on a die, die sizes are smaller now than with single core processors of a few years ago. The new Intel Core Duo processors have a die area of 111-143 sqmm, about half the size of the previous year's dual core Pentium D 900 at 280 sqmm, and distinctly smaller than 183-230 sqmm for recent AMD Opteron dual core processors using the older 90 nm process. (By a sense of scale, 4/3" sensors have die areas of over 243 sqmm, and DX format sensors are over 430 sqmm.)

The biggest IC's that I know of are in the Intel Itanium series, and this year's dual core Itanium 2 processors using the slightly older 90 nm process are 27.72 mm x 21.5 mm = 596 sqmm (about 1D sensor sized). This is down from about 750 sqmm for the first dual core Itanium processors from 2003-2004, which used 130 nm process. Itanium processors are likely to move to the smaller 65 nm process, so even combined with a predicted move to quad core, will likely stay smaller than the dual core Itanium's of a few years ago, staying at about "1D" size. Canon also predicts a move to 45 nm process in three year's a time, further reducing sizes.

The trend seems to be increasing component count without increasing die size, and often decreasing die size.


To Edmund:  as a professional scientist, I judge claims more by evidence than by the speaker's educational credentials (so I have not tried to support my claims by reference to my Ph. D. and record of publications in physics journals), so could you provide evidence of a recent trend to larger reticle sizes in fab. equipment suitable for making sensors, which is what would be required to make larger sensors more likely in the future?

[eronald]

BJL

Let's reason together.

A chip has to fit on a wafer so at a given technology there is always an absolute limit to chip size. However, wafer size is increasing with time, so wafer size is not an absolute intemporal limitation on chip size.

Then a chip neets to be printed - lithographed if you prefer. If this is done by mask exposure rather than direct write on wafer, there is an issue of reticle size, however, I believe chips like the Dalsa and Kodak designs are currently stitched by multiple exposures anyway, which is evidence that reticle size is not a permanent limitation on chip size.

This leaves the issue of local defects on the chip. Chip defects per infintesimal surface area at a given technology are proportional to this infinitesimal surface area . (p(ds is defect)=mu*ds) where mu is constant. Integrated, to reflect that *one* defect means a bad chip, this gives something like a Poisson yield statistic where you have a negative yield exponential going from 1 to 0 (asymptotically) as the chip surface increases.  Tiny chips are -almost- always good, huge chips -almost- always defect.  Luckily,there is empirical evidence that the constant mu has historically gone down with production epochs, ensuring that larger and larger chips can be made.

I suggest you now go and read up the above points via google. You will eventually hit upon the  Dalsa white papers, and various tutorials which document yield statistics with all the relevant maths, modelling and empirical data.

If  after doing the above research, you still feel a need to pour scorn on me you know where you can find me. Luckily, PhDs are awarded for life, and so I expect to be allowed to keep mine even after the abuse of wine, women and cameras has burnt out my last neurons.

Edmund

To Edmund:  as a professional scientist, I judge claims more by evidence than by the speaker's educational credentials (so I have not tried to support my claims by reference to my Ph. D. and record of publications in physics journals), so could you provide evidence of a recent trend to larger reticle sizes in fab. equipment suitable for making sensors, which is what would be required to make larger sensors more likely in the future?
[a href=\"index.php?act=findpost&pid=91806\"][{POST_SNAPBACK}][/a]

[josayeruk]

BJL

Let's reason together.

If  after doing the above research, you still feel a need to pour scorn on me you know where you can find me. Luckily, PhDs are awarded for life, and so I expect to be allowed to keep mine even after the abuse of wine, women and cameras has burnt out my last neurons.

Edmund
[a href=\"index.php?act=findpost&pid=91837\"][{POST_SNAPBACK}][/a]

Phd's and handbags at fifty paces!!!    

I've got a City and Guilds in E6 processing.    

Plus I would imagine that a megapixel count of a 56mm square sensor would be so bonkers that it would make it unusable with todays storage and computers???  Yes?

39 is enough for me thanks very much!

Jo S.x  

(Correct my maths if I am wrong but 70+ Mega Pixels?)

[Kumar]

Plus I would imagine that a megapixel count of a 56mm square sensor would be so bonkers that it would make it unusable with todays storage and computers???  Yes?
[a href=\"index.php?act=findpost&pid=91898\"][{POST_SNAPBACK}][/a]

Is it technically possible to have a larger sensor without increasing the pixel count? Would it impact picture quality? And of course, does it make marketing and financial sense?

Cheers,
Kumar

[josayeruk]

Is it technically possible to have a larger sensor without increasing the pixel count? Would it impact picture quality? And of course, does it make marketing and financial sense?

Cheers,
Kumar
[a href=\"index.php?act=findpost&pid=91917\"][{POST_SNAPBACK}][/a]

Yes, but then single shot quality would suffer if the pixel sites where enlarged again.

No easy answer!    

Jo S. x

[josayeruk]

PhDs are awarded for life, and so I expect to be allowed to keep mine even after the abuse of wine, women and cameras has burnt out my last neurons.

[a href=\"index.php?act=findpost&pid=91837\"][{POST_SNAPBACK}][/a]

A Phd is for life, not just for Christmas?  

[nik]

Without slagging anyone with a PhD, it's all very good talking about Intel and AMD fabrication and the trend they seem to be taking toward smaller size chips, but they are being designed with very different goals in mind, speed and energy efficiency. Very different to what the design goal of an imaging sensor is. If a larger sensor size will help solve the current limitations that Dalsa and Kodak face in future sensor design, then that's what they will do. No?

Hopefully they will at least reach the real 645 full frame size.

Go easy on flaming me if you must.

-Nik

BJL

Let's reason together.

A chip has to fit on a wafer so at a given technology there is always an absolute limit to chip size. However, wafer size is increasing with time, so wafer size is not an absolute intemporal limitation on chip size.

Then a chip neets to be printed - lithographed if you prefer. If this is done by mask exposure rather than direct write on wafer, there is an issue of reticle size, however, I believe chips like the Dalsa and Kodak designs are currently stitched by multiple exposures anyway, which is evidence that reticle size is not a permanent limitation on chip size.

This leaves the issue of local defects on the chip. Chip defects per infintesimal surface area at a given technology are proportional to this infinitesimal surface area . (p(ds is defect)=mu*ds) where mu is constant. Integrated, to reflect that *one* defect means a bad chip, this gives something like a Poisson yield statistic where you have a negative yield exponential going from 1 to 0 (asymptotically) as the chip surface increases.  Tiny chips are -almost- always good, huge chips -almost- always defect.  Luckily,there is empirical evidence that the constant mu has historically gone down with production epochs, ensuring that larger and larger chips can be made.

I suggest you now go and read up the above points via google. You will eventually hit upon the  Dalsa white papers, and various tutorials which document yield statistics with all the relevant maths, modelling and empirical data.

If  after doing the above research, you still feel a need to pour scorn on me you know where you can find me. Luckily, PhDs are awarded for life, and so I expect to be allowed to keep mine even after the abuse of wine, women and cameras has burnt out my last neurons.

Edmund
[a href=\"index.php?act=findpost&pid=91837\"][{POST_SNAPBACK}][/a]

[josayeruk]

Without slagging anyone with a PhD, it's all very good talking about Intel and AMD fabrication and the trend they seem to be taking toward smaller size chips, but they are being designed with very different goals in mind, speed and energy efficiency. Very different to what the design goal of an imaging sensor is. If a larger sensor size will help solve the current limitations that Dalsa and Kodak face in future sensor design, then that's what they will do. No?

Hopefully they will at least reach the real 645 full frame size.

Go easy on flaming me if you must.

-Nik
[a href=\"index.php?act=findpost&pid=91946\"][{POST_SNAPBACK}][/a]

No offence intended and all Good points Nik.    

I think what you say about the 645 frame full size is more likely in he future.

Jo S. x

[free1000]

I believe chips like the Dalsa and Kodak designs are currently stitched by multiple exposures anyway, which is evidence that reticle size is not a permanent limitation on chip size.

Yes... which is where I expect the dreaded centre fold comes from as the stitching process seems to result in non uniform halves to the chip...

maybe they will figure out how to do this properly in a year or two.

[Marsupilami]

I am very sceptical that a full frame chip for a mamiya Rz would make sense. I doubt that these old lens designs are able to cope with digital sensors well, specially when no crop is used.

Christian

[BJL]

BJL

Let's reason together.

A chip has to fit on a wafer so at a given technology there is always an absolute limit to chip size. However, wafer size is increasing with time, so wafer size is not an absolute intemporal limitation on chip size.

Of course, but wafers are already so much bigger than any sensors that they are in no significant way a limit to sensor size, so that increasing wafer size will have little effect on the economics of making, say, a 56x42.5 (645) sensor.

... I believe chips like the Dalsa and Kodak designs are currently stitched by multiple exposures anyway, which is evidence that reticle size is not a permanent limitation on chip size.

I know about multiple exposures needed for sensors larger than reticle sizes: Canon also says that its 24x36mm chips are also made with multiple exposures. The point is that multiple exposures substantially increase sensor costs (or at least Canon says so) through lower yields, and I would expect that each extra exposure increases costs more, so reticle size also creates cost notches at "maximum double exposure size",  "maximum triple exposure size" and so on. (My guess is that the smaller 33x44mm "medium format" sensor size is the largest possible with just two exposures of some fab. equipment.)

This leaves the issue of local defects on the chip ...
I suggest you now go and read up the above points via google.[a href=\"index.php?act=findpost&pid=91837\"][{POST_SNAPBACK}][/a]

No need to be condescending: there is nothing in your post that I did not already know. Of course there will be some downward trend in the cost of sensors of any given size due to improving "mu" value, but that is a relatively modest factor in chip cost reduction, compared to the method that is vastly more popular in the digital imaging industry: reducing the size of the photo-sites (and thus of the sensors) needed to achieve a given level of image quality.

P. S. It is perhaps noteworthy that the team of Sony and Nikon has not even bothered to go to sensors at the maximum size possible with a single exposure, which is reportedly something around Canon 1D size (reticle size about 26x33mm according to Canon, so sensor active area a bit smaller.) Nikon is a major maker of IC fab. equipment, so there should not be a problem of access to suitable tools.