Nikon's 'Hundred's' line to be discontinued?

[Everythingisnt]

With the release of the D700, Nikon's response to Canon's 5D, I think that we're seeing how Full frame is slowly becoming the industry standard for DSLR systems..

Seeing as how the imminent release of the D90 will see a d80 body sporting a D300 sensor, and that the D700, D3, and the future D3x form a very encompassing 'pro' lineup, do you think that Nikon will drop their 'hundreds' line in place of cheaper semi-pro FX format cameras and high end consumer DX models? I think that the price difference between the D300 and D700 is to small to discern them, and seeing as cheaper cameras like the D90 will give D300-quality with a smaller price tag, Nikon may decide to drop the hundred's line and widen the gap..



Any thoughts?

[yrsonicdeath]

Personally the D700 is too expensive for me right now and the D300 is not so that price difference isn't too little to me. I also think that all of its specs taken into account the D300 holds a place for some users in the Nikon line-up.

[Westside guy]

I can't follow the logic here - $1000 is too small a price difference to differentiate between the D300 and D700? And we're not talking about the difference between $8999 and $9999 - we're talking at the low end, $1800 versus $3000. That's a huge difference.

Do you feel the price difference between the D80 and D300 is too small to discern them? Because that gap is comparable to the gap between the D300 and D700.

I wouldn't be surprised if Nikon has several different plans written up, and sales from the next 1-2 years may determine which path they take. Some people (e.g. me) wanted a mid-tier full frame camera, but others really do prefer DX. Nikon may or may not have an idea of the respective popularity of full frame and DX will be - but they may want to see how D300 versus D700 sales shake out before committing to how they go from here on out.

In any case I can't see them dropping part of their price lineup; just possibly moving other items into existing slots.

[NEiMac]

I know that I for sure plan on getting a D300 or perhaps its replacement in the future. The D700 is about $1000 more then I will be willing to spend on a camera anytime soon and to be honest I really don't need the thing. I would much rather spend the extra dollars on better glass. This is all just a hobby to me though, I want the best gear possible but within reason. :D

[yrsonicdeath]

Some people (e.g. me) wanted a mid-tier full frame camera, but others really do prefer DX.

That's what I was hoping for too, but with the D700 announced I've decided to buy a D300 and just invest in some lenses for my next camera, which with the way cameras are going I can say with some certainity will be FF.

[Grimace]

Nope. Stratifying the camera lines is not helpful so as many distinct price points as Nikon/Canon can create will actually help them.

$500-900 Consumer
$900-1200 Consumer/Enthusiast
$1200-1500 Prosumer
$2500-3000 Semi/Pro
$3500-4000 Pro
$5000-8000 Pro hi res

[OreoCookie]

I don't agree with your conclusion: FX sensors will remain a lot more expensive than DX sensors, because the yield is smaller and you fit less sensors on a wafer (this is different from computer chips where the die area tends to shrink -- not an option if you want to keep the size constant).

I think DX sensors will be here to stay -- as are FX sensors. We might see that a successor of the D300 merges with a successor of the D700, but I don't think neither Nikon nor Canon will go further down than that. You have a full complement of lenses for crop sensor (on both sides of the aisle), sensors that deliver very good quality at a moderate price.

I'm sure that anything below a D80+/Canon 450D will stick to crop sensors.

Even if you're right and in 7 years all dslrs will have crop sensors, I don't think it's particularly useful to pretend as if you're going to get a full-frame sensor dslr next month. Unless you really will ;)

[ksz]

FX sensors will remain a lot more expensive than DX sensors, because the yield is smaller and you fit less sensors on a wafer (this is different from computer chips where the die area tends to shrink -- not an option if you want to keep the size constant).
I am not so pessimistic about relative pricing of DX and FX sensors. I think FX sensors are already coming down in price. Canon's 5D started at $3500 and now sells for about $2000. While I'm not sure how much of that decline is due to smaller margins being charged by retailers and how much is due to component cost reduction by Canon, it does seem that a full frame dSLR body can be sold profitably for $2000 today versus $3500 about 18-24 months ago.

At $2000, the Canon 5D is only about $300-$400 more expensive than the Nikon D300.

Canon's successor to the 5D may get priced at $3000 or less, and the pixel count may jump from 12.x to about 16 MP.

As wafer yield matures and unit volume increases, cost goes down.

[OreoCookie]

At $2000, the Canon 5D is only about $300-$400 more expensive than the Nikon D300.
Canon sells it at a discount, because the current model is `old' (≠ necessarily bad).
As wafer yield matures and unit volume increases, cost goes down.
But not to the degree that it makes financial sense to insist on a full-frame sensor line-up from the smallest to the largest camera. You cannot improve wafer yields beyond certain points and DX sensors will always have a higher yield and will cost less.

[ksz]

Canon sells it at a discount, because the current model is `old' (≠ necessarily bad).
But if it is still profitable for Canon, it means that a full-frame dSLR with 12.x MP can be sold for $2000 (actually, under $2000 at several retailers).

But not to the degree that it makes financial sense to insist on a full-frame sensor line-up from the smallest to the largest camera. You cannot improve wafer yields beyond certain points and DX sensors will always have a higher yield and will cost less.
Wafer yield, which in turn is related to individual die yield, is affected not so much by the size of the die, but by the density of discrete components such as pixels or transistors. If you have a full-frame sensor with 5 million pixels, you will have more interstitial space (pixel-to-pixel gap) than a DX sensor with 5 million of the same-sized pixels. This pixel density, which relates to the amount of space between pixels, also affects fabrication yield because a sparse distribution is less impacted by defects than a dense distribution in which critical dimensions are smaller and hence manufacturing tolerances are tighter.

But I agree that if we keep pixel density the same or increase the size of pixels on a full-frame sensor, then a FF sensor is more expensive due to both defectivity and die-per-wafer.

Wafer yields are actually pretty high on long-life parts in 300mm fabs around the world, with greater than 90% sustained yield on even relatively large die sizes. Short-life parts where only a handful of lots are manufactured will most likely not see mature yields, but a part that exists for a long time ramps up to mature yield levels.

[compuwar]

With the release of the D700, Nikon's response to Canon's 5D, I think that we're seeing how Full frame is slowly becoming the industry standard for DSLR systems..


1. Nikon has repeatedly stated that they do not match Canon model-for-model. The D700 is Nikon's response to Nikon's marketing plan- not their "response" to a 5D. A "response" usually doesn't price out at 30% more than the thing being responded to.

2. Full-frame 35mm will not become the "industry standard," the sensor is still the single-most expensive component, and hence the least-profitable part of the whole deal. Those economics are not destined to change- 2x the sensor, 4x the cost is likely to hold true for as long as we're dealing with CMOS and CCD devices, perhaps longer unless someone creates a magic "make a perfect wafer every time" device- in which case it'll still be 2x the cost.

3. Doubling from 2 to 4 models is hardly creating a "new standard-" even if you go to six, you're still nowhere near "standard."


Seeing as how the imminent release of the D90 will see a d80 body sporting a D300 sensor, and that the D700, D3, and the future D3x form a very encompassing 'pro' lineup, do you think that Nikon will drop their 'hundreds' line in place of cheaper semi-pro FX format cameras and high end consumer DX models? I think that the price difference between the D300 and D700 is to small to discern them, and seeing as cheaper cameras like the D90 will give D300-quality with a smaller price tag, Nikon may decide to drop the hundred's line and widen the gap..



Any thoughts?

Highly unlikely. There are advantages to having denser sensors in the professional line, and "cheaper" is much, much more difficult to do in FX than it is in DX because you get significantly smaller yields per wafer- I would suggest you read Canon's 2006 whitepaper "Canon's Full-Frame CMOS Sensors." It's the best look into the economics we'll get, and other than narrowing the number of steps for a full-frame sensor, it's still up to date. I suggest you pay particular attention to Section IV, "The Economics of Image Sensors" which includes the following data:

1 8" wafer ($450-500, $1,000 or $5,000) through 400-600 process steps gives you a wafer of sensor material. If you take that wafer and make APS-C sensors you get about 200 sensors. If you make APS-H sensors, you get about 46 sensors per wafer. For full-frame, you get 20.

Let's do that again- APS-C 200 sensors, FX 20 sensors.

That's a base 10x price difference that Canon says goes from 10x-20x because of handling and other defect issues. Turning that wafer into a sensor costs a lot of money- but that doesn't even make much difference, if I make 1M cameras and I can spend $22.50 less per unit on raw materials, that makes me $22.5M more profitable.

Let's say that in today's market an APS-C sensor is $75 in finished form, an FX sensor is going to be at least 3x and more likely between 6x and 9x that amount. So that'd put it in the $225 and up range. That's a $150/unit price increase before you look at bigger mirrors, hot mirrors, AA filters and prisms. Suddenly, my 1M cameras become almost a quarter of a billion more profitable.

I don't know why everyone keeps baying for "cheaper" FX cameras like the manufacturers are putting a thousand dollars into their pockets. The economics are always going to be in the APS-C camp- and the margins aren't staggering (they're good, but not phenomenal.) FX sensors are not cheap, they may get cheaper, but not on a ballistic curve and never relative to APS-C sensors.

Wishful thinking. Nikon will continue to produce what bodies make economic sense in the market, and pro APS-C bodies are "good enough" for lots of folks- I see no reason they'd ditch them any more or less than Canon would- that's where the best margins are, or you can give up some margin for volume sales- I don't see a rational reason for either company giving up that flexibility.

[compuwar]

with greater than 90% sustained yield on even relatively large die sizes.

I don't think I've seen anyone claiming better than ~20% yields for FX image sensors and ~60% for DX- do you have any citations?

[telf22]

i dont think the 'd90' will have the d300's sensor.

[ksz]

I don't think I've seen anyone claiming better than ~20% yields for FX image sensors and ~60% for DX- do you have any citations?
I am talking about wafer yields in general for relatively large devices with the point being that devices with long lifetimes get ramped up to high yield because of a continuous focus on yield improvement. Costs are reduced as process technologies improve to support fewer mask layers (fewer process steps), provide more resistance to fabrication problems, and reduce cycle time.

We should not assume that a low yielding part today will remain low yielding forever. That would be crazy.

A recent example is the advent of immersion lithography for the fabrication of sub-65nm feature sizes, in which a wafer is immersed in a thin layer of water, which has the effect of increasing the numerical aperture of a lithography lens while still using 193nm wavelength of light. Even more recent examples are experiments with double patterning - or pitch splitting - in which 32nm patterns can be printed while again using 193nm lithography. Etch chambers are also under tighter process control, and 300mm wafers live in FOUPs that rarely expose them to the ambient environment.

The semiconductor industry is never satisfied with the status quo.

[ChrisA]

I think it's clear that Nikon does not plan to drop the 100 line. Why else did they leave a hole in the numbers between D300 and D700. They plan to drop in some models There is room to continue the 100, 200, 300 line or make a cheaper FX body.

When people buy cameras they pretend to think about what they need but we all know they just look at the price and buy whatever matches their budget. Every consumer electronics makers knows this too and is very careful to have a model that fits each price point. My bet is that the mid range price point (where the 100 series lives) at just over $1K is the most profitable. I'm sure a D300 can't cost that much more than the D60 to make. Nikon would not give that up.

[OreoCookie]

A recent example is the advent of immersion lithography for the fabrication of sub-65nm feature sizes, in which a wafer is immersed in a thin layer of water, which has the effect of increasing the numerical aperture of a lithography lens while still using 193nm wavelength of light. Even more recent examples are experiments with double patterning - or pitch splitting - in which 32nm patterns can be printed while again using 193nm lithography. Etch chambers are also under tighter process control, and 300mm wafers live in FOUPs that rarely expose them to the ambient environment.

The semiconductor industry is never satisfied with the status quo.
You don't really understand: smaller structures help you keep the die size small while keeping the number of transistors constant/increasing them. The number of defects is strongly correlated to the size of your chip. Larger chips will always be more expensive than smaller chips.

A simple model: an APS-C-sized sensor has approximately half the area of a full-frame sensor. Since wafers are circular (and do not have a rectangular cross section), this means, you can get less than half of the number of full-frame sensors on one wafer. Then assume you use the same process and that defects are only related to dust on the wafer (there are others, larger masks require more precise alignment to the wafer's surface, etc.). The number of dust particles per wafer area is constant. Then full-frame sensors are twice as likely to be defective than crop sensors while there are less than half of the number of sensors on the die.

Obviously, larger masks for sensors are harder to handle and other factors worsen the yield of full-frame sensors significantly. Canon/ have already optimized the processes, but you cannot escape that truth that full-frame sensors are significantly more expensive than crop sensors. In the budget segment, it would also make other components (mirror, pentaprism/pentamirror) more expensive, cameras would be larger.

I think it's clear that Nikon does [I]not plan to drop the 100 line. Why else did they leave a hole in the numbers between D300 and D700. They plan to drop in some models There is room to continue the 100, 200, 300 line or make a cheaper FX body.
Agreed.
When people buy cameras they pretend to think about what they need but we all know they just look at the price and buy whatever matches their budget. Every consumer electronics makers knows this too and is very careful to have a model that fits each price point. My bet is that the mid range price point (where the 100 series lives) at just over $1K is the most profitable. I'm sure a D300 can't cost that much more than the D60 to make. Nikon would not give that up.
Well, I'd be a little more careful about that. The D300 uses a pentaprism viewfinder, the D60 has a smaller pentamirror viewfinder. The D300 has a metal chassis, the D60 does not. There is definitely a difference in cost (although I would bet it's nowhere near as big as the price difference between the two cameras).

[Hmac]

1. Nikon has repeatedly stated that they do not match Canon model-for-model. The D700 is Nikon's response to Nikon's marketing plan- not their "response" to a 5D. A "response" usually doesn't price out at 30% more than the thing being responded to.

Absolutely correct IMHO. I don't think either of these two companies are competing for a given market niche...I think they are working to create market niches. Nikon was pretty clever with the D3 - it covered a lot of ground and created a lot of anxiety in owners of the 5D, the 1DmkIII, and even got some 1Ds MkIII users thinking hard. Now, they've created a whole new niche with the D700. The concept makes me suspect that the 5D MkII, if there ever is such a thing, won't be a direct response to the D700. I think Canon will work hard to define a niche of their own rather than compete in the one Nikon created.



2. Full-frame 35mm will not become the "industry standard," the sensor is still the single-most expensive component, and hence the least-profitable part of the whole deal. Those economics are not destined to change- 2x the sensor, 4x the cost is likely to hold true for as long as we're dealing with CMOS and CCD devices, perhaps longer unless someone creates a magic "make a perfect wafer every time" device- in which case it'll still be 2x the cost.

Yeah, that may be right, but I doubt it. I think market forces will decide whether or not the move is to full frame, not economics of sensor manufacturing, which have been, are, and will be subject to change over time (with traditionally decreasing cost). Obviously, those economics will affect the market, but I think it's clear that the market interest these days is on full frame and I suspect the camera-buying public won't be denied. I would be surprised if, in 3-5 years, it would be possible to even buy a professional-level cropped sensor camera from either Canon or Nikon.

[ksz]

You don't really understand: smaller structures help you keep the die size small while keeping the number of transistors constant/increasing them. The number of defects is strongly correlated to the size of your chip. Larger chips will always be more expensive than smaller chips.
You keep missing the point... :(

Process technology continues to improve. Yield on long-life parts continues to improve. Cost of production for these parts continues to decline over time. Sometimes process improvements and cost reductions are dramatic, sometimes they are more subtle.

You seem fixated on the idea that full-frame sensors will never reach a price point where they can be sold to the masses at price points of under $1000. I just don't know what fuels your pessimism about the yield and cost of full frame sensors.

Large die sizes are more of a challenge, but the challenges are workable. The semiconductor industry has always innovated itself out of brick walls. The examples I gave above regarding immersion lithography and pitch splitting techniques are examples of such innovation which keep costs down by extending the life of 193nm lithography.

With the use of FOUPs, wafers are not subjected to the environment, and so dust or fall-on particles are not the yield-limiting factor. Systematic issues, particularly with etch and CMP (chemical mechanical planarization), are more of a concern although even these process steps have seen significant improvements. At sub-45nm, lithography remains a challenge because a 193nm lithographic wavelength cannot be squeezed too much further. Nevertheless, when it comes to manufacturing yield, there is significant progress made over time.

Again, to argue that 20% yield on full frame sensors will remain that way indefinitely is kinda foolish. :)

A simple model: an APS-C-sized sensor has approximately half the area of a full-frame sensor. Since wafers are circular (and do not have a rectangular cross section), this means, you can get less than half of the number of full-frame sensors on one wafer. Then assume you use the same process and that defects are only related to dust on the wafer (there are others, larger masks require more precise alignment to the wafer's surface, etc.). The number of dust particles per wafer area is constant. Then full-frame sensors are twice as likely to be defective than crop sensors while there are less than half of the number of sensors on the die.
I agree that larger masks are more troublesome, but not because of stepper alignment issues, but because of edge defocus issues. A larger reticle will be more in-focus at its interior, and focus will drift near the edges. But if the pattern to be printed is symmetric on the left and right sides, a smaller reticle can be used that exposes the left and right sides in two steps.

Obviously, larger masks for sensors are harder to handle and other factors worsen the yield of full-frame sensors significantly. Canon/[insert sensor manufacturer here] have already optimized the processes, but you cannot escape that truth that full-frame sensors are significantly more expensive than crop sensors. In the budget segment, it would also make other components (mirror, pentaprism/pentamirror) more expensive, cameras would be larger.
I don't argue that full frame sensors will be more expensive than crop sensors. My argument is that full frame sensors can achieve economies of scale (and also inherent improvements in yield) to allow full frame dSLR bodies to be sold at near-DX prices of today.

The Canon 5D is already close to the price of Nikon's D300. As more full frame bodies are introduced and more consumer interest is generated by continuing declines in prices, I fully expect full frame bodies to break through the $1500 barrier and subsequently break the $1000 barrier.

[OreoCookie]

You keep missing the point... :(

Process technology continues to improve. Yield on long-life parts continues to improve. Cost of production for these parts continues to decline over time. Sometimes process improvements and cost reductions are dramatic, sometimes they are more subtle.
I don't keep missing the point: any improvement in the manufacturing process will apply to smaller sensors as well. I don't expect full-frame sensors in anything below a D300-class camera any time soon. It might become a feature of most pro bodies (i. e. a common feature in $1300+ cameras) within the next, say, five, six years (two generations).
I just don't know what fuels your pessimism about the yield and cost of full frame sensors.
It's not pessimism is math and facts.
Large die sizes are more of a challenge, but the challenges are workable. The semiconductor industry has always innovated itself out of brick walls. The examples I gave above regarding immersion lithography and pitch splitting techniques are examples of such innovation which keep costs down by extending the life of 193nm lithography.
I don't argue that companies even could (if they insisted) put full-frame sensors in virtually all bodies -- but it wouldn't make too much sense.

The high quality of crop sensors coupled with most consumers using cheap lenses doesn't give a lot of advantages to full-frame sensors (if any) if you use low quality glass. It's like these compact cameras with 12 MP sensors and lenses smaller than most coins. So even if there was a way to offer full-frame sensors at a similar price point than crop sensors, even then I'm not convinced they'd be put into consumer-oriented dslrs.
With the use of FOUPs, wafers are not subjected to the environment, and so dust or fall-on particles are not the yield-limiting factor.
It was a simple example. In real-life applications, the size of the sensor correlates not just linearly with size.
Again, to argue that 20% yield on full frame sensors will remain that way indefinitely is kinda foolish. :)
I haven't made that claim.
But any improvement in processing larger sensors will also apply to smaller sensors and I doubt the offset (relative difference in yield) will decrease substantially.
The Canon 5D is already close to the price of Nikon's D300.
The Canon 5D is old and is sold `at a discount'. Most people expect its successor to be sold at a similar price point as the Nikon D700 (which was the same price point the 5D started selling at).

[Westside guy]

Not that it's going to stop you, but I don't believe either of you is going to convince the other of a) which one is "missing the point"; and b) your individual position on this. :D

[ksz]

^^^ I agree with Westside guy, but I want to address one comment before dropping the argument.

The high quality of crop sensors coupled with most consumers using cheap lenses doesn't give a lot of advantages to full-frame sensors (if any) if you use low quality glass. It's like these compact cameras with 12 MP sensors and lenses smaller than most coins. So even if there was a way to offer full-frame sensors at a similar price point than crop sensors, even then I'm not convinced they'd be put into consumer-oriented dslrs.
Crop sensors have improved and likely will continue to do so. But as I argued elsewhere in this forum, a full frame sensor has more headroom for future advancement in pixel quantity without sacrificing pixel quality. As full frame camera bodies become more plentiful and more affordable -- the cost of the sensor is irrelevant to customers -- it makes sense to invest in full frame lenses for people like me who would like to move into medium format territory.

Crop sensors for DSLRs were invented in the first place to address the cost issue. At a time when everyone with film SLRs had an investment in full frame lenses, and film SLRs were priced from a few hundred dollars to about two thousand dollars (for the Nikon F5), digital SLRs would not have been able to succeed in the market if they sold for $8000 to $10,000 with full frame sensors. Crop sensors were therefore invented, along with DX lenses.

This wasn't a bad idea. I have a D200 and I've enjoyed using it for the past 18 months. But I have also been concerned about the longevity of crop DSLRs and have limited my DX lens collection to merely 2.

Looking forward, I am much more excited about the potential of full frame sensors and I feel more comfortable investing in full frame lenses (specifically, I want to replace 3 of my 6 full-frame lenses with new models).

I also see full frame DSLRs becoming more mainstream as prices decline. If the D700 becomes a runaway success for Nikon, it will send a clear message that there is a good size segment of the market that is willing to spend 50- to 70-percent more (and eventually less) for full frame over cropped frame (but not 100% or 300% more).

[Hmac]

If the D700 becomes a runaway success for Nikon, it will send a clear message that there is a good size segment of the market that is willing to spend 50- to 70-percent more (and eventually less) for full frame over cropped frame (but not 100% or 300% more).

The D3 has been very successful, the D700 apparently so (according to Adorama anyway), and judging by the tone of the FX/DX remarks on DPR and the many, many enthusiastic comments on the D900 and/or D3X that I see there, there's a lot of interest in a 24mp full-frame Nikon as well.

I think the handwriting is on the wall, but time will tell.

[QuantumLo0p]

With the release of the D700, Nikon's response to Canon's 5D, I think that we're seeing how Full frame is slowly becoming the industry standard for DSLR systems..

Would have been nice for full frame to be standard from the start.

From the Art Bell voices in my head: Perhaps the big OEM's revenue stream was a bit flat so they simply decided to to manufacture DX lenses to supplement revenue.

:eek:

[compuwar]

I am talking about wafer yields in general for relatively large devices with the point being that devices with long lifetimes get ramped up to high yield because of a continuous focus on yield improvement. Costs are reduced as process technologies improve to support fewer mask layers (fewer process steps), provide more resistance to fabrication problems, and reduce cycle time.


But we are discussing imaging sensors, where the pixels have to be mapped to a particular geographic location on the sensor, not yields for general semiconductors. Claiming that there's a possibility of a 90% yield in the current CMOS processes for 35mm sized imaging sensors is disingenuous.

We're nowhere near 90% yields for 35mm sensors- heck we're nowhere near 50% yields for 35mm sensors. Most improvements are modest- 2-3% other than cutting where we're seeing maybe a 20% improvement. The fact that every fab focuses on yield and we're at about 20% for 35mm sensors seems to indicate that leaps and bounds are not likely, and the delta will remain rather significant for the time being. Any process improvement also applies to smaller sensors, so you really have to get APS-C to near 100% then make more improvements, then go to square wafers before you get parity.


We should not assume that a low yielding part today will remain low yielding forever. That would be crazy.


Nobody has advanced that position yet. The position advanced is that the _delta_ between APS-C and 35mm sized imaging sensor yields will remain relatively high for the foreseeable future. If we see a 30% improvement at the 35mm end, the APS-C end should gain enough that you're still talking a significant difference, perhaps even a larger delta- and that still only puts you at a 50% yield for 35mm sensors. Costs will certainly come down over time, but that won't be short-term without some real groundbreaking advances in sensor creation. Nobody in the fab industry seems to see anything groundbreaking coming. There's also a good bit of profit margin in the prosumer market that can be traded for sales volume with APS-C that isn't going to be there anywhere near as quickly or proportionate with 35mm.

Going to smaller process sizes for imaging sensors really doesn't affect yields like it does for other semiconductors because you can't remap a pixel somewhere else on the sensor and get the same results. The smaller your photosites the quicker diffraction rears its ugly head, and the more photosites are likely to be affected by a single defect on the wafer. That's why image sensor defects that aren't manufacturing process issues (like damage while cutting the wafer) have kept yields relatively static between different sized sensors. If it were otherwise, you'd see Hasselblad taking on the full frame folks at the high end and kicking their butts- but they're penalized even more than the 35mm sensor folks from yields.

[cube]

I am not saying that full frame sensors are going to become affordable any time soon, but where do you get your yield figures from?

[OreoCookie]

I am not saying that full frame sensors are going to become affordable any time soon, but where do you get your yield figures from?
I would suppose from the source he has quoted before, Canon's full frame white paper (http://www.robgalbraith.com/public_files/Canon_Full-Frame_CMOS_White_Paper.pdf). (Don't hesitate to correct me if I'm wrong, though, compuwar).

[ksz]

Claiming that there's a possibility of a 90% yield in the current CMOS processes for 35mm sized imaging sensors is disingenuous.
Please be careful. I have been saying that the semiconductor industry has always innovated itself out of brick walls, and I have provided examples of such innovation. I have not said that current CMOS processes for full frame imaging sensors will get to 90% yield, but that process improvements take place continuously and there is significant incentive to do just that as the market for full frame sensors grows.

Like OreoCookie, you are also fixated on the notion that improvements in sensor yields will not occur for the foreseeable future, but you are not citing objective data. A 2006 white paper from Canon is not representative of sensor yields today.

Any process improvement also applies to smaller sensors, so you really have to get APS-C to near 100% then make more improvements, then go to square wafers before you get parity.
Fabrication of large image sensors presents some unique challenges that are not applicable to small sensors. Large reticles with edge defocus issues may be one of these unique problems. Your absolute statement that "any process improvement" also applies to smaller sensors is naive.

I don't know why you keep bringing up "square" wafers. That's not something that's going to happen any time soon. It should not be part of any realistic argument.

The semiconductor industry is currently researching a move to 400mm wafers (Intel made some noise about 450mm wafers a couple of years ago). The transition will not happen for at least another 5 years. Obviously this will increase the number of full frame sensors per wafer, but I won't argue this line because it is too far out.

The position advanced is that the _delta_ between APS-C and 35mm sized imaging sensor yields will remain relatively high for the foreseeable future. If we see a 30% improvement at the 35mm end, the APS-C end should gain enough that you're still talking a significant difference, perhaps even a larger delta
This again depends on the premise that any process improvement for full frame sensors also applies to smaller sensors, which again is a naive assumption.

Nobody in the fab industry seems to see anything groundbreaking coming.
Citations please.

Going to smaller process sizes for imaging sensors really doesn't affect yields like it does for other semiconductors because you can't remap a pixel somewhere else on the sensor and get the same results.
And no one is claiming that.

The smaller your photosites the quicker diffraction rears its ugly head, and the more photosites are likely to be affected by a single defect on the wafer. That's why image sensor defects that aren't manufacturing process issues (like damage while cutting the wafer) have kept yields relatively static between different sized sensors. If it were otherwise, you'd see Hasselblad taking on the full frame folks at the high end and kicking their butts- but they're penalized even more than the 35mm sensor folks from yields.
This is not relevant because no one is claiming that image sensors are subject to die shrink.

[eddx]

Im gunna say no, the D300 / D200 / D100 line is here to stay. I would expect to see a D400 in the next 18 months with a 21mp DX sensor. It is a prosumer DX sensor DSLR.

D80/D90 is the flagship DX sensor and D40x/D60 is the basic DX DSLR.

DX for amteurs and FX for pros is the way I see Nikon going. Otherwise why would they be bringing out more DX lens?

By the way, I am a pro who is happy with the quality I can get from the DX sensor and my work doesn't require or justify FX at this moment in time.

[Westside guy]

Im gunna say no, the D300 / D200 / D100 line is here to stay. I would expect to see a D400 in the next 18 months with a 21mp DX sensor. It is a prosumer DX sensor DSLR.

I agree that the D400 will come, but I will be very surprised if the pixel density is that high. With photosites that small, diffraction would set in at (I'm guessing) about f/8. That'd put a severe restriction on the types of shots you could take with the camera.

I suspect Thom Hogan is right - the DX line will top out at 14-16 megapixels.

[Driver]

Yeah, that may be right, but I doubt it. I think market forces will decide whether or not the move is to full frame, not economics of sensor manufacturing, which have been, are, and will be subject to change over time (with traditionally decreasing cost). Obviously, those economics will affect the market, but I think it's clear that the market interest these days is on full frame and I suspect the camera-buying public won't be denied. I would be surprised if, in 3-5 years, it would be possible to even buy a professional-level cropped sensor camera from either Canon or Nikon.

It's inevitable. They have to increase mega pixels (and sensitivity) just to get people to upgrade. It's an industry that counts on people upgrading every 3-4 years. And they'll have to offer significant new cameras to force consumers to plunk down the cash. A full frame chip gives them room to to drive these changes (more pixels, better sensitivity, etc..), regardless of the price, which naturally will come down with time and process improvements.

Even though I've only had my D300 for a month I guaranty I don't go looking for DX lenses for it. I'll buy a lens for a full frame camera and make do until I can upgrade to the D5 in several years. The proliferation of DX lenses will have that segment hang on a little longer than I think some anticipate but Nikon will cut them loose from their ProSumer/Professional segment without much fanfare. They'll keep a low end slr in their lineup for 10 years or so just for users of those lenses but the middle to high end SLRs will be Full Frame soon enough. My guess is more like 7-8 years as the photography chip market moves a lot slower than the computer chip market. If they were equal we'd all be shooting with 80MP cameras by now.

By the time I go to a full frame camera I expect the iPhone will have a 4-6MP sensor in it. :cool:

[compuwar]

Please be careful. I have been saying that the semiconductor industry has always innovated itself out of brick walls, and I have provided examples of such innovation. I have not said that current CMOS processes for full frame imaging sensors will get to 90% yield, but that process improvements take place continuously and there is significant incentive to do just that as the market for full frame sensors grows.


You said "Wafer yields are actually pretty high on long-life parts in 300mm fabs around the world, with greater than 90% sustained yield on even relatively large die sizes. Short-life parts where only a handful of lots are manufactured will most likely not see mature yields, but a part that exists for a long time ramps up to mature yield levels."

In a discussion about sensor yields- intimating (and certainly not disclaiming) that 35mm sensors (parts that exit for a long time- at least in terms of Canon's fab plant) on relatively large die sizes (and I'm assuming here that you consider a 35mm sensor to be a relatively large size) get greater than 90% yields. So either you're intimating that 35mm sensors should be getting 90% yields, or you're drawing a conclusion about the semiconductor industry that hasn't applied to imaging sensors so far and that none of the current imaging sensor manufacturers I've seen are trumpeting, telegraphing or making any noise at all about.


Like OreoCookie, you are also fixated on the notion that improvements in sensor yields will not occur for the foreseeable future, but you are not citing objective data. A 2006 white paper from Canon is not representative of sensor yields today.


Prove that it is not *representative*. I've looked at annual reports, patents, and everything else, and while nobody's pushing numbers like they were ~24 months ago, that's not a huge amount of time- and nobody's annual or even quarterly reports are talking about killer yield numbers. The relative numbers are the same, even if the absolute costs have gone down- and even if yields have gone up some, it's still a surface area problem.

Show me a more recent representative paper.

Finally, are you asserting that Canon replaces its steppers every 2 years for the same sensor line? While the numbers will have changed, their relative representations really don't seem to have changed- but I'd be happy to see some citations, because all the good analysis that I have is either 2 years out or difficult to sustain attempts to reverse the sensor price from the camera price and approximate profit margins.


Fabrication of large image sensors presents some unique challenges that are not applicable to small sensors. Large reticles with edge defocus issues may be one of these unique problems. Your absolute statement that "any process improvement" also applies to smaller sensors is naive.


Present a process improvement that will only benefit large sensors- the gain to smaller sensors may not be as material, but it will still apply- edge defocus issues have been solved, which is why it (reportedly) no longer takes more than one patterning step to produce a 35mm sensor- but multiple exposures didn't do anything to yield, it just spoke to process improvements- and it meant that you could expose more smaller sensors in a step- so it's not unique to larger sensors. It still doesn't address yields, but it's not a large sensor only improvement.


I don't know why you keep bringing up "square" wafers. That's not something that's going to happen any time soon. It should not be part of any realistic argument.


What part of "to achieve parity" don't you understand? Your mythical, but unknown "leap" isn't realistic as far as I can see.


The semiconductor industry is currently researching a move to 400mm wafers (Intel made some noise about 450mm wafers a couple of years ago). The transition will not happen for at least another 5 years. Obviously this will increase the number of full frame sensors per wafer, but I won't argue this line because it is too far out.


But again, it does nothing to address yields, so it's a strawman- more 35mm sensors means even more APS-C sensors, larger wafers isn't a sensor yield issue, it's a sensors-per-process-step issue.


This again depends on the premise that any process improvement for full frame sensors also applies to smaller sensors, which again is a naive assumption.


Unlike your assertion that alignment issues aren't a big deal, Canon's last annual report trumpets their new solution for alignment. They're also focused on inspecting for defects up front- which doesn't exactly scream "way cool improvements in yields by solving a problem with the wafer" now does it?

Once again, even if 35mm yields doubled, they'd still be 30% behind APS-C yields from two years ago. Nobody seems to be trumpeting a 100% increaase in 35mm sensor yields in the last two years, and since only Canon and the MF manufacturers were doing 35mm processes in volume it's not like there was a huge universe of folks attacking the problem. Feel free to show some real data and anything other than blind hope that the "industry" is going to "make a leap" that has no current substance.

[compuwar]

If they were equal we'd all be shooting with 80MP cameras by now.


It's not just how fast the industry moves, it's the difference that progress gets you. An 80MP 35mm sensor starts to show loss of sharpness due to diffraction at f/5.3. Not very useful in general photography.

[ksz]

In a discussion about sensor yields- intimating (and certainly not disclaiming) that 35mm sensors (parts that exit for a long time- at least in terms of Canon's fab plant) on relatively large die sizes (and I'm assuming here that you consider a 35mm sensor to be a relatively large size) get greater than 90% yields. So either you're intimating that 35mm sensors should be getting 90% yields, or you're drawing a conclusion about the semiconductor industry that hasn't applied to imaging sensors so far and that none of the current imaging sensor manufacturers I've seen are trumpeting, telegraphing or making any noise at all about.
You keep missing the point. The argument that yield is related to die size is true, but die size is not the only factor, nor the most significant factor. The TI digital micro-mirror device (DMD) has a relatively large die size, but the most complex part of the process is fabrication of the tiny micro mirrors that actually twist. Most of the yield loss comes from this part of the process. Drawing a blind argument between die size and yield is naive. Like an image sensor, a DMD chip does not have redundancy which could be used to reroute electrical signals from faulty circuits to redundant or backup circuits. But TI has developed other techniques to recover yield. There are "correctable" faults even on a DMD.

My statement about 90% yield on long-life parts is exactly meant to assert that achieving high levels of yield on FX sensors should not be considered impossible. People who believe in the impossible are, in my view, useless people. They are the ones who thought a mission to the moon was impossible, and they would never have undertaken the challenges needed to make it a reality. The semiconductor industry has solved numerous very challenging problems, and is about to solve the challenges for 22nm. To think that they cannot bring FX sensor yields above 20% is lunacy.

Prove that it is not *representative*.
You don't understand the rules pertaining to burden of proof. It is incumbent upon you to prove your assertion.

I've looked at annual reports, patents, and everything else, and while nobody's pushing numbers like they were ~24 months ago, that's not a huge amount of time- and nobody's annual or even quarterly reports are talking about killer yield numbers. The relative numbers are the same, even if the absolute costs have gone down- and even if yields have gone up some, it's still a surface area problem.
Do you know why you cannot readily find yield numbers? Fab yield is one of the most closely guarded secrets.

The market for FX sensors is currently very small, which does not provide incentive for companies to pour resources into developing new process technologies. For general semiconductor devices, development of new process technologies is very expensive, and getting more so. Consortiums such as Sematech and IMEC are formed to spread the costs. Once the market for FX sensors grows, I see every reason for R&D expenses to rise to meet the demand for volume and price.

Finally, are you asserting that Canon replaces its steppers every 2 years for the same sensor line?
No.

Present a process improvement that will only benefit large sensors- the gain to smaller sensors may not be as material, but it will still apply
Explain clearly why the larger die size of a Full Frame sensor is so difficult to manufacture. What are the specific fabrication problems? Now explain why these fabrication problems don't affect smaller die sizes, but only affect large die sizes. Now if we find a solution for the large die size problem, it will have the greatest benefit for large die and maybe nothing for small die.

Now let's say we have 100% yield on small sensors because we've made them really small. And we have 20% yield on large sensors because, well, because they're large. Now if I increase my yield on large sensors from 20% to 50%, will I increase my yield on small sensors from 100% to 120%? I would really like to see that!

The point is clear: you have to understand the actual fabrication issues that limit yield. Citing only die size is naive and misleading.

- edge defocus issues have been solved, which is why it (reportedly) no longer takes more than one patterning step to produce a 35mm sensor- but multiple exposures didn't do anything to yield, it just spoke to process improvements- and it meant that you could expose more smaller sensors in a step- so it's not unique to larger sensors. It still doesn't address yields, but it's not a large sensor only improvement.
The Canon APS-H sensor used in the 1D (not 1Ds) body is, according to Canon's white paper, the largest die size that can be printed with one exposure. Their full frame sensor is printed with 3 exposures, but these are 3 smaller reticles than the single APS-H reticle because the FF sensor is not 3 times the size of the APS-H sensor. Canon solved the edge defocus issue by exposing the die 3 times, but the APS-H sensor is more sensitive to edge defocus.

Now are you saying that 3 separate exposures with smaller reticles did nothing to improve Canon's yield on FF sensors? Wow. I really need a reference for that.

What part of "to achieve parity" don't you understand? Your mythical, but unknown "leap" isn't realistic as far as I can see.
Because I don't think you're really looking. You really seem fixated on the notion that nothing can be done to improve yield on FF sensors, so the industry might as well throw up its arms and give up. That's a really wonderful attitude...

But again, it does nothing to address yields, so it's a strawman- more 35mm sensors means even more APS-C sensors, larger wafers isn't a sensor yield issue, it's a sensors-per-process-step issue.
But it does reduce cost because the number of good die per wafer increases. After all, we are talking about bringing down the price of FF sensors. That can be done by yield improvement and by more die per wafer. I'm surprised you didn't follow this argument.

Unlike your assertion that alignment issues aren't a big deal, Canon's last annual report trumpets their new solution for alignment. They're also focused on inspecting for defects up front- which doesn't exactly scream "way cool improvements in yields by solving a problem with the wafer" now does it?
There are two types of stepper alignment. One is die-to-die alignment. The other is die-on-die overlay alignment for subsequent mask layers. Both are well-understood problems and have been for a while. By using 3 exposures for a FF sensor, Canon has to accomplish both die-on-die and die-to-die alignment in each pass, but even this isn't new.

Again, we have to know what exactly causes yield loss on large sensors to have a meaningful debate. Talking about die size alone is devoid of reason.

Once again, even if 35mm yields doubled, they'd still be 30% behind APS-C yields from two years ago. Nobody seems to be trumpeting a 100% increaase in 35mm sensor yields in the last two years, and since only Canon and the MF manufacturers were doing 35mm processes in volume it's not like there was a huge universe of folks attacking the problem.
Now here we agree. As the market for full frame sensors grows, so will the R&D investment needed to reduce costs.

Feel free to show some real data and anything other than blind hope that the "industry" is going to "make a leap" that has no current substance.
Blind hope? Is that what John F. Kennedy had when he commissioned NASA to put a man on the moon? Is that what the semiconductor industry had when it solved the lithography problem for 45 and 32 nanometer? Yield of a full frame sensor is just another technological problem. You seem to suffer from blind pessimism.

[compuwar]

You keep missing the point. The argument that yield is related to die size is true, but die size is not the only factor, nor the most significant factor.

Die size is the most significant factor in price to yield issues.

http://www.naturescapes.net/092006/ej0906.htm

The bane of semiconductor manufacturing is die yield or in our context, sensor yield. Digital sensors, like all other semiconductor devices are made on wafers that have a number of sensors printed on them. But only a percentage of these are good and the rest are scrap. The percentage of good sensors as their size goes up is reduced dramatically.


My statement about 90% yield on long-life parts is exactly meant to assert that achieving high levels of yield on FX sensors should not be considered impossible. People who believe in the impossible are, in my view, useless people. They are the ones who thought a mission to the moon was impossible, and they would never have undertaken the challenges needed to make it a reality. The semiconductor industry has solved numerous very challenging problems, and is about to solve the challenges for 22nm. To think that they cannot bring FX sensor yields above 20% is lunacy.


Once again-Nobody but you has said that high yields on FX sensors are impossible. Nobody but you has said that sensor yields can't be brought above 20%- what has been said is that there is not a miracle over the horizon that will bring FX sensors to 90% yields soon. That's in terms of the current manufacturers who are actually manufacturing FX sensors. But even if they did, the economics still favor APS-C, and they will always favor APS-C.


Do you know why you cannot readily find yield numbers? Fab yield is one of the most closely guarded secrets.


Which is why sensor size must be extrapolated as a cost. However, yield increases are often trumpeted by fabs- in terms of imaging sensors there are basically five markets- medical imaging, space imaging, security devices, scanning and video/still cameras- in the first two, the cost of the sensor isn't as closely tied to the cost of the device as it is in the last three, and still cameras are the place where we can get the best data because of relatively low margins and high production rates.


The market for FX sensors is currently very small, which does not provide incentive for companies to pour resources into developing new process technologies. For general semiconductor devices, development of new process technologies is very expensive, and getting more so. Consortiums such as Sematech and IMEC are formed to spread the costs. Once the market for FX sensors grows, I see every reason for R&D expenses to rise to meet the demand for volume and price.


Again, any process improvements will move the entire market, not just the FX fab market.

Now explain why these fabrication problems don't affect smaller die sizes, but only affect large die sizes.


They don't- other than more process steps giving more opportunity to damage a wafer- but as I said in my last posting, I believe that Canon has reduced that in their newer processes. However *everything cited to date* has assumed _the exact same defect rates regardless of die size- that is, the process problems that cause defects are assumed to be the same for the same wafer size- that's why the economics are what they are.
[/quote]


Now let's say we have 100% yield on small sensors because we've made them really small. And we have 20% yield on large sensors because, well, because they're large. Now if I increase my yield on large sensors from 20% to 50%, will I increase my yield on small sensors from 100% to 120%? I would really like to see that!


Now you're just being asinine.


The point is clear: you have to understand the actual fabrication issues that limit yield. Citing only die size is naive and misleading.


No it isn't- everyone in the industry gives yield numbers based on defects-per-wafer and costs larger die sizes based mostly (but not completely) on the same number of defects-per-wafer. Obviously, if there are more process steps, or if there are process defects those make yields lower for larger dies, but there is very little inclusion of those issues in the overall math. If you normalize those process differences, so long as you've still got defects on a wafer, larger die sizes lose more yield.


The Canon APS-H sensor used in the 1D (not 1Ds) body is, according to Canon's white paper, the largest die size that can be printed with one exposure. Their full frame sensor is printed with 3 exposures, but these are 3 smaller reticles than the single APS-H reticle because the FF sensor is not 3 times the size of the APS-H sensor. Canon solved the edge defocus issue by exposing the die 3 times, but the APS-H sensor is more sensitive to edge defocus.

Now are you saying that 3 separate exposures with smaller reticles did nothing to improve Canon's yield on FF sensors? Wow. I really need a reference for that.


No, it didn't- because Canon wasn't then doing a single exposure- so it's yields have pretty-much always been from 3 exposures. My understanding is that they can now do a single exposure, which may have improvied yields due to fewer process steps and fewer alignment issues, but not by enough to make significant changes in the price of the sensor. APS-H sensors were the largest single-exposure sensor because they didn't lose yield because of defocus- or they would have done two exposures for APS-H.


Because I don't think you're really looking. You really seem fixated on the notion that nothing can be done to improve yield on FF sensors, so the industry might as well throw up its arms and give up. That's a really wonderful attitude...


You're trying to put words into my mouth again- you obviously have a reading comprehension issue, so I'm done. I never said "nothing can be done to improve yield," or that yield improvement shouldn't be advanced- your continuing attempts to twist my words makes any rational discourse impossible.

As we used to say on USENET, *plonk*.

[ksz]

^^^ Well, this has been pretty futile. I am tempted to rebut every statement of yours once again, but it will be a waste of my time.

[Hmac]

Name-calling and personality issues aside, I found the discussion interesting and educational. I don't think that sensor manufacturing plays as big a role in future camera development as consumer-driven market forces, but the conversation helps one understand the landscape.