My Prediction for the new EOS Lineup

[Stu_bert]

Agreed, I am no engineer either so it is pure conjecture on my part, and I meant no offence in the discussion. I guess I am assuming that in the same way we keep cramming more transistors onto CPUs, and make bigger and higher res Plasma/LCD/LED devices, then the same cost-challenges with sensors can be overcome. Of course making a 12 MP FF sensor for $500 this year, which may have cost you $700 2 years ago to make, also means you can make an APS sensor for $300. So the economies of scale would apply to either.

I guess my point was more that overall cost, simplifying your whole range has an affect on a whole manner of elements - from develop/manufacture/market to support and thus help reduce your cost base and allow you to increase your margin or lower your prices. Whether that was standardise on FF or APS would have the same affect to your cost base. I was just biased as would hate them to standardise on APS sensors ;D

 

[Rocky]

Agreed, I am no engineer either so it is pure conjecture on my part, and I meant no offence in the discussion. I guess I am assuming that in the same way we keep cramming more transistors onto CPUs, and make bigger and higher res Plasma/LCD/LED devices, then the same cost-challenges with sensors can be overcome. Of course making a 12 MP FF sensor for $500 this year, which may have cost you $700 2 years ago to make, also means you can make an APS sensor for $300. So the economies of scale would apply to either.

I guess my point was more that overall cost, simplifying your whole range has an affect on a whole manner of elements - from develop/manufacture/market to support and thus help reduce your cost base and allow you to increase your margin or lower your prices. Whether that was standardise on FF or APS would have the same affect to your cost base. I was just biased as would hate them to standardise on APS sensors ;D

CMOS process is a relative stable process and has been around for a long time.. We cannot expect any substantial cost reduction in sensor from the process alone, except moving from 8 inch wafer to 12 in wafer. With the same amount of physical work (mostly done with automation), a little more chemical, and a slight modification in process and you end up twice as much sensor.
As for improving the sensor performance, we should have high hope. just look at the latest sensor from Nikon( Sony). However, we have already pushed beyond the capability of the lens resolution with 18MP APS_C sensor.

As a foot note: Most other electronic device are enjoying huge cost reduction in the last 40 years due to the advance in processing technology and cost reduction in the processing itself and hence the internal device on the silicon is getting smaller and smaller and resulted in the actual silicon in each IC is getting smaller and smaller and hence more and more IC per wafer. They are getting cheaper and cheaper. Unfortunately. This does not appy to the DSLR snesor. The sensor size is fixed. So there is no cost reduction in shrinking internal device. Instead we end up with higher pixel count. We have moved from 2MP (APS-C) to 18MP (APS-C) in the last 15 years or so.
As for the prine difference of FF and APS-C, The size ratio between them is about 2.63. So the cost ratio between FF and APS-C will be 5 to 7 times depends on the cleningness of the fabrication facility.

 

[Stu_bert]

Rocky, as you clearly have more understanding in this area, would appreciate a little more info/understanding....

If moving from 8" to 12" is not much additional cost but you end up with twice as much sensor, does it follow that you can similarly continue to scale the wafer size to benefit?

Secondly, did not quite get the maths on 2.63x larger, therefore 5-7x more expensive?

Thirdly, electronic shrinkage, is it not the case that all that happens is we get more for the same price - more transistors etc, of course coupled with better use of those transitors? Did not think in the case of CPUs they were getting smaller, just more on them - which is akin to CMOS sensors advancement?

Finally (almost there ) but is cost reduction of TV panels again just a maturity of the technology / manufacturing process?

Sorry so many questions - just like to learn

 

[Rocky]

Stu_bert, increase the wafer size is not as easy as it looks. It is very envolved. Also the equipment will be extremely expensive for large wafer. As far as I know, nobody can break the 12" barrier yet. As for yield, the ratio is not the ratio of the size but roughly the 1/ square of the ratio of size. The size of idividual trnssitors are getting smaller and smaller, As a result, Some IC are getting smaller and smaller and hence getting cheaper and cheaper. Some IC have multifunction. that is why most of the electronic device are getting cheaper including Digital camera. As for CPU, that is a slightly different story. The transistors are getting smaller and becomes faster. The CPU also evolved from single core to double, quad, even hex. I can predict tha we will see Octo soon.
As for flat panel TV, IT is not yet a mature process, so the cost is dropping due to increase in yield.

Hope this will help

 

[Mt Spokane Photography]

The quality of the wafer is yet another factor. Canon say they use a higher grade of wafers with as few defects as possible, because the yield can be very low with FF or APS-H sensors.

Still, I wonder if they are able to find all the defects before they assemble the sensor with microlenses, anti-aliasing, etc. If they have to do all that before testing, the cost of a good sensor goes up a whole lot.

 

[Rocky]

The quality of the wafer is yet another factor. Canon say they use a higher grade of wafers with as few defects as possible, because the yield can be very low with FF or APS-H sensors.

Still, I wonder if they are able to find all the defects before they assemble the sensor with microlenses, anti-aliasing, etc. If they have to do all that before testing, the cost of a good sensor goes up a whole lot.

There are so much at stake for the 12" or even 8" wafer. So everybody are using high quality wafer. Raw(starting wafer) are dirt cheap, compared to the finished wafer. Infact most of the defects are from each processing steps. there are at leat 30 steps, some can go as high as 60 plus steps. Testing sensor can be quick, it is all automatic problem is that you cannot notto add microlens in individual sensor. You have to do it on the whole wafer, for bothj good and bad sensors all at the sametime.

 

[Mt Spokane Photography]

The quality of the wafer is yet another factor. Canon say they use a higher grade of wafers with as few defects as possible, because the yield can be very low with FF or APS-H sensors.

Still, I wonder if they are able to find all the defects before they assemble the sensor with microlenses, anti-aliasing, etc. If they have to do all that before testing, the cost of a good sensor goes up a whole lot.

There are so much at stake for the 12" or even 8" wafer. So everybody are using high quality wafer. Raw(starting wafer) are dirt cheap, compared to the finished wafer. Infact most of the defects are from each processing steps. there are at leat 30 steps, some can go as high as 60 plus steps. Testing sensor can be quick, it is all automatic problem is that you cannot notto add microlens in individual sensor. You have to do it on the whole wafer, for bothj good and bad sensors all at the sametime.

See Canons white paper, a bit dated, but not a lot has changed, lots of small things, but the process and number of steps has not been reduced from 600 ti 60 or 30.

http://www.scribd.com/doc/49842014/Canon-CMOS-WP

And, as Canon says, they do not use cheap $30 wafers, but pay up to $5,000 for ultra high quality ones.

 

[macfly]

That is a great read, than you for posting the link.

 

[AG]

I cant see the 5D being split but i can see the 1D series being merged.

I can actually see the 7D being dropped and as for the 3D idea, unless it was released as a gimmick actual "3D" camera. I cant see it selling because unfortunately that is what the 3D label has come to represent.

I think that eventually Canon will realise that as nice as it is to have multiple choices for customers they will eventually do a similar strategy to someone like Apple and reduce the options.

The way i see it, it will be:

Fully Pro Camera - 1D series with ALL the bells and whistles but no video, purely a FF Photographers camera.

Prosumer Camera - 5D series, lots of MP & FF, geared around both Photo and Video, Best of both worlds.

Enthusiast - 60D series, pretty much as is but eventually gaining the 7D's AF and other benefits.

Amateur - 600D series, no changes from current models

Entry - 1100D series, no changes from current models or if anything add more funkier colours to take on the Pentax KR

I know that this will upset a few people but it would also mean that people would know what to gear towards. There would be a uniformity in the line ups and this would mean that Canon would only need to really focus on the top 2 cameras.
The rest would be a trickle down effect.

 

[Rocky]

The quality of the wafer is yet another factor. Canon say they use a higher grade of wafers with as few defects as possible, because the yield can be very low with FF or APS-H sensors.

Still, I wonder if they are able to find all the defects before they assemble the sensor with microlenses, anti-aliasing, etc. If they have to do all that before testing, the cost of a good sensor goes up a whole lot.

There are so much at stake for the 12" or even 8" wafer. So everybody are using high quality wafer. Raw(starting wafer) are dirt cheap, compared to the finished wafer. Infact most of the defects are from each processing steps. there are at leat 30 steps, some can go as high as 60 plus steps. Testing sensor can be quick, it is all automatic problem is that you cannot notto add microlens in individual sensor. You have to do it on the whole wafer, for bothj good and bad sensors all at the sametime.

See Canons white paper, a bit dated, but not a lot has changed, lots of small things, but the process and number of steps has not been reduced from 600 ti 60 or 30.

http://www.scribd.com/doc/49842014/Canon-CMOS-WP

And, as Canon says, they do not use cheap $30 wafers, but pay up to $5,000 for ultra high quality ones.

There are different ways to define step. In the industry, each step is one mask. Between each mask, there are a few process, So If you count everything, 600 "steps" are highly possible. As for the number of sensor an 8" wafer can produce, there are some mistakes too. There is no way an 8" wafer can produce 200 APS-C sensor. I worked out to number to be roughly 70 sensor sites for APS-C for a 8" wafer. There are a few holes in the paper. we can read it anyway we want. I read it as a paper to justify the high cost of FF sensor. I am speaking from my own experience about the CMOS process.. You can also read it anyway you want also.

 

[Stu_bert]

@Rocky - many thanks