Given a $3,500 or so SRP, there is a lot of working room from
today's $1,800 effective US street price (B&H say) after rebates.
Another factor is that they might shrink the pixel slightly and
keep the multiplier down to 1.4 to 1.3X rather than 1.2X. With a
little better design/process they could keep the noise down. Even
if the noise PER PIXEL is slightly higher on a 6MP, assuming the
noise is random, for the same size output the effective noise would
be less.
Another BIG cost factor is that they may allow "dead" pixels. A
big killer in yielding large die is the probably of a single or few
defects. At 6 Megapixels, one could tolerate the occasional dead
pixel and map it out with processing in the camera. The whole
Bayer filter process means that to get a output pixel you are
already processing the fool out of what comes out of a number of
imager "Pixels."
At 3MP you don't have as much to work with for a reasonable size
output (Say 8x10). As the number of pixels goes up, you can be
more tolerant of an occasional bad pixel (somewhat analogous to
grain in film). Based on my 25 years of Semiconductor
experience, this the probably the way they will go. According to
Phil's Review of the Nikon 5000, the other camera companies (but
not the Nikon 5000 in his review) are mapping out bad pixels
already on consumer cameras.
Every DRAM today has "redundancy" or else the yields would be next
to zero. With redundancy the yield go from less than 1% to greater
than 80%. You could not afford a PC today if you did not accept
memories that used redundancy to map out bad pixels.
I'm guessing that an 8" wafer costs about $2500 to $3000.
Finishing the wafers and packaging would about double that (depends
on whether they can yield sort before or after packaging). A quick
guess is that that will give about 35 to 40 displays at an
UNYIELDED cost of about $150 dollars a piece. The cost per good
device is given by dividing $150 by the Yield (if say 50% then the
cost is $300).
Thus paradoxically it COULD be cheaper to build a 6MP device IF you
accept a few scattered defects at 6MP but could not tolerate them a
3MP.
Karl
To do 6MP right, the chip would have to be much bigger. A 1.2
Multiplier would be expected. I don't know how they could do this
given current technology and hit D30 pricing. If this is really
6MP, nearly full framed, and D30 priced, it should destroy the
competition.
Red Flags:
Now if the chip is the same size as the current chip and 6MP, then
it will be higher noise and most current lenses would not
adequately resolve 6mp in that small space. This would be the case
if they chose marketing over good design.
Chip size is what I see as the main stumbling block.
Peter
and what are the "red flags" to look for in the specs. (not that
they wouldn't be pointed out here)
--
Karl
