The penalty of Canon not having the volume to put sufficient R&D funding into its sensors is just playing out. They are 10 years behind Sony with column ADC.
Hi Bobn2: I realize you have far more knowledge on sensor fab than myself. That being said, it may true Canon started 10 years behind Sony it does not mean they are actually 10 years behind presently. Canon does not need to reinvent the wheel so it is possible Canon column ADC will catch up to the current Sony chips within a couple of years max (and Sony will make some incremental updates in the meantime). Within a few years the column ADC will be very mature tech and there may be little difference between manufacturers at that time. Then again, I could be completely out in left field on this
Of course somebody else will eventually come up with another breakthrough in another area of sensor fab and reset the bar even higher

Cheers.
The point is that Canon people tend to concentrate on ownership sensor fab. It's not a big issue. Most image sensing producers use foundry fab, the reason being that until you get as big as Sony, there is not enough business to keep a fab line busy enough to justify keeping it up to date.
depends on the size of the fab - canon needed two plants to manufacturer 6 million sensors per year. since they still manufacturer around 5.8 milllion sensors per year, going by initial volumes, they are keeping both plants busy.
That is not clear. The gist of the press release for the newer fab was that it was to make sub-APS-C sensors, which at a time it was getting from Sony.
Which of course leads to another question - how did canon produce at one time, 10 million a year.
Quite simply, it had two fabs running, the 500nm one for APS-C and bigger and the 180nm one for the in-house 1/7" sensors it was making (probably was selling several million of those, they put it in several high volume cameras). Now, its buying the1/7" from Sony again (and some of those products have gone to 1") and moving APS-C and above to the 180nm line, so the 500nm line is phasing out.
they of course may have improved / updated their equipment without notifying dpreview.
They never have informed DPReview, when they open a new line, thy put aout a press release and write about it in their annual report. You'd wonder why they'd want to keep quiet about major investments, particularly in their annual report.
Using foundry, your sensors are mixed in with other business, and you can choose an up to date fab line when you need it. Canon's problem has been that they are locked into two fab lines, and only one has small enough geometry to do complex circuitry like column ADCs, they couldn't just abandon the old one, because that would have meant writing off the capital.
the capital would have been written off years ago.
That depends, doesn't it? A fab line might cost $1bn for a cheap one. Lets' assume the Canon ones weren't new, so used price $500M. If they amortise $10 capital costs per señor, that's 50M sensors they have to make to pay for the fab line, that's about 10 years.
not to mention it would be been simply equipment replacement what would have been near EOL anyways. you don't have to "abandon" a fab because the equipment is no longer in use. you CAN upgrade the equipment. it's costly, but it depends on the wafers/ month to how much it does scale for cost. considering that most of the equipment would have been from canon themselves - what really would have been the cash cost?
the first plant was built I believe 10 years ago, the second, 7 years ago.
That would fit, with the figures above. The first line has paid for itself, it can go. The second, maybe not (because you can't really add $10 to the cost of a 1/1.7" sensor without making it completely uncompetitive)
This new generation of Canon sensors probably means that they have moved onto the newer 180nm fab line for their large sensors, and will finally retire the old 500nm line.
one small note though - the 250MP prototype was done on 130nm. there's an entire possibility that canon simply shut down and replaced the 180nm line altogether.
That would be very strange. More likely that the 250MP was made in a foundry, maybe an extension of the contracts Canon already have to do part of their wafer processing (the BEOL) at foundries.
that acquisition cost would have been pocket change in canon's R&D expenditures.
Not really. It is a big enough investment it would have had to have appeared in their annual report to investors.
They probably moved to it to smaller geometries with DPAF since the QE from that generation onwards - jumped, and the noise characteristics from the sensors changed. for starters, there's an extra switch for DPAF, wiring, not to mention the pixel itself is cut in half, and with that the QE still jumped over the older 18Mp sensor.
Speculation, we'll see with the next Chipworks report.
it also explains why they "sat" at 18MP for so long; as since then canon has come out with different sensors for the 70D, 7DII, 80D, T6 in basically 2.5 years. Canon's rapidly changing it's sensors now like they used to.
I think the step to 180nm would explain all that. Finally they decided to go to Sony for the new 1/1.7" sensors, retire the old line and use the newer one for big sensors. Fits what's known better than the idea that they have an unannounced new 130nm fab line. Moreover, even 180nm was right at the limit of i-line steppers, and Canon does not have an immersion product. So, if Canon really does have its own 130nm facility, it will have to be equipped with Nikon or ASML kit (possibly that would be why they didn't want to publicise it

)
as far as "catching up" it's usually easier to catch up then it is to break new ground, but it's a matter of patent portfolio. it's not a matter of canon needing years to catch up if they managed to secure the patents necessary to develop a suitable sensor. Some of those patents cleared very late last year.
Personally, In don't think it has much to do with patents. Much of the performance has to do with fine tuning over time, rather than patents. But sure, yes, Canon won't take 10 years from where they are now to get to where Sony are now. But when they get there, Sony will have moved on. They are already making BSI FF sensors, and I have a suspicion that what's in the D500 is a Sony APS-C stacked sensor.
Also the differences now in between the senors are all within the realms of diminishing returns. the difference between an 80D and a latest tech A6300 is negligible, at the level of pushing you have to do to see the difference, you have other problems cropping up - loss of microcontrast, resolution, clarity, color casting,etc.
It's moving to a different sphere with stacked sensors and BSI. That's more about fast lens performance and full frame video read-out rates, global shutters and other facilities.
--
Bob.
“The picture is good or not from the moment it was caught in the camera.”
Henri Cartier-Bresson.