Upcoming Nikon Z9 Sensor by TowerJazz (now Semiconductor)??

Started 2 months ago | Discussions
BGD300V1
BGD300V1 Senior Member • Posts: 2,757
Re: Nikon Sensor Development
1

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

-- hide signature --

I'm a photographer, Jim, not a graphic artist!
My photo blog: http://birdsnbugs.com
RF Stock Portfolio - http://www.dreamstime.com/resp129611

 BGD300V1's gear list:BGD300V1's gear list
Nikon 1 V2 Nikon 1 J5 Nikon Z7 Nikon Z6 Nikon Z50 +12 more
bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
10

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

OK. Continually using the D850 sensor as an example of the work of Nikon's sensor design team, when that is known to be a Sony designed sensor - at least the silicon, and most of the article is about the process of designing the silicon. Statements like this:

"And from our reader’s side, given that the D850 won our Camera of the Year award for 2017 and its sensor recently receiving the first DxO Mark score of 100 for an SLR, I thought there’d be a lot of interest in hearing some of what went into its making.", which lead you to believe that the article describes the D850 sensor design. You can not that all the example illustrations of a sensor being designed are of the D5 sensor, which was designed by Nikon's sensor design team.

Then in the section 'what exactly do Nikon's designers do', it describes the process of designing the silicon, and says 'this is what Nikon does', but fails to mention that Nikon does this for only a small selection of the sensors they use (currently, only the D6 sensor). Then sections like this:

'In answering my question, Sanbongi-san talked in part about the need to match sensor designs to the capabilities of Nikon's EXPEED image processing engines (which are also designed in-house), and about being able to tailor performance to the level of camera each sensor will be used in.'. Sure, there is a need to match the image processor and sensor, but the majority of Nikon's cameras use commodity Sony sensors with their standard SLVS-EC interface - which the article goes on to talk about as though it was developed by Nikon. It wasn't, it is Sony IP.

So it says 'Finally of course, doing their own R&D lets them create sensors like the one in the D850, with its low-ISO ability (a true ISO 64), and the one in the D5, with its exceptional light sensitivity and quantum efficiency.'. Nikon's sensor design team did not 'create' the D850 sensor. It's a product of Sony's design team and the whole approach and architecture is different from the D5 sensor - because they were designed by different teams. Then it repeats uncritically Nikon marketing guff. 'This is one area where Nikon’s dedicated sensor design has paid off for them: The D850 has the first true ISO 64 capability in an SLR. (Other cameras have special “Lo” ISO settings that will get there, but those come at the expense of poor tonal qualities and blown highlights.)'. The D810 has a 64 base ISO, and Nikon has never said that is at the cost of poor tonal qualities and blown highlights, and it isn't. Later on it says 'While we’re ultimately limited at the low-light end of things by the fact that we’re basically counting individual photons, under bright conditions, there are still advancements to be found, as they did with the ISO 64 capability of the D850’s sensor.' Given that the whole article is about Nikon's sensor design team, the implication is that this is a Nikon 'advancement', when the D850 is a Sony designed sensor with Sony BSI pixel design. In any case, there is no real difficulty in making a sensor have sufficient saturation capacity to work down to a base ISO of 64, it's just a case of increasing the pixel capacitance. The penalty is a bit of a rise in read noise, but if you have Sony's dual gain architecture, where the capacitance is switchable, that isn't an issue.

Then he talks about the sensor teams, "(So there was a D850 sensor team, a D7500 one, etc.)'. The D7500 uses the same sensor as the D500, and was promoted by Nikon as doing so when launched, so why would there be a 'sensor team' to design it? The D850 is a Sony sensor, so why would there be a Nikon team to design it? I doubt that Nikon ever told Etchells that there were teams that designed these sensors, it's just text he made up, without checking. What there might have been is a group within the camera design team that made a selection as to what would be the best sensor for that model, whether an existing Sony product, a new design commissioned from Sony, or a Nikon design, but there won't be Nikon design teams when Nikon didn't design the sensor.

'Nikon’s sensor R&D goes on more or less continuously'...'the process takes multiple years for advanced sensors like those in the D5 and D850'. Very likely, but the impression given here is that the D5 and D850 sensor went through the same design process, they very clearly didn't. One is a fairly standard Exmor BSI sensor, the other is Nikon pixel array design linked to Toshiba's column ADC design. The look and behave completely differently. In fact, you can see this difference just in the little chip photo of the D5 sensor (there are no examples of the D850 sensor given, though the article talks about it repeatedly).

So that is the gist of it. Most of the sensors that Nikon uses are either solely designed by Sony SS, or designed by Sony SS to Nikon specifications, possibly with Nikon adding specific CFA and microlens layers. This article fails to make that differentiation and suggests that all Nikon's sensor design work is done in house, which is false. My suspicion is that Nikon didn't lie. Instead Etchells was fed weasel words, and let the weasel bite him.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

lattesweden
lattesweden Veteran Member • Posts: 4,190
Re: Nikon Sensor Development
1

bobn2 wrote:

the D850 is a Sony designed sensor with Sony BSI pixel design. In any case, there is no real difficulty in making a sensor have sufficient saturation capacity to work down to a base ISO of 64, it's just a case of increasing the pixel capacitance. The penalty is a bit of a rise in read noise, but if you have Sony's dual gain architecture, where the capacitance is switchable, that isn't an issue.

First thanks for pointing out the Wiki page for Sony sensors in another post as an answer to my question. It gives a little better view and understanding of what Sony has made sensor wise, but unfortunately not what the others manufacturers have done and the most interesting one - which camera body uses which sensor, though there where some examples. I think many would like that sort of list with camera model, sensor model and maybe a small comment if needed. I don't have the knowledge to do it, so I have to hope someone else puts such a list together.

I try to learn more about sensors in general, so I find it an interesting topic and I'll throw in two more questions I think you might know the answer to:

If Sony sensor division can make 64 ISO sensors, why don't Sony imaging use that in their cameras as well? Is there a drawback to use 64 ISO vs 100? Or why settle for 100?

Same question goes for an earlier discussion about Nikon using the 14-bit read out mode and Sony stuck with the 12-bit one in their camera using the same sensor. Same question here, is there a drawback with using the 14-bit mode vs the 12-bit one? If not, why settle for 12?

Thanks!

-- hide signature --

Best regards
/Anders
----------------------------------------------------
Mirrorless, mirrorless on the wall, say which is the best camera of them all?
When I put my camera in Manual mode, why don't I get any instructions?
Some images:
https://www.dpreview.com/forums/post/64169208
https://www.dpreview.com/forums/post/64221482

bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
4

lattesweden wrote:

bobn2 wrote:

the D850 is a Sony designed sensor with Sony BSI pixel design. In any case, there is no real difficulty in making a sensor have sufficient saturation capacity to work down to a base ISO of 64, it's just a case of increasing the pixel capacitance. The penalty is a bit of a rise in read noise, but if you have Sony's dual gain architecture, where the capacitance is switchable, that isn't an issue.

First thanks for pointing out the Wiki page for Sony sensors in another post as an answer to my question. It gives a little better view and understanding of what Sony has made sensor wise, but unfortunately not what the others manufacturers have done and the most interesting one - which camera body uses which sensor, though there where some examples. I think many would like that sort of list with camera model, sensor model and maybe a small comment if needed. I don't have the knowledge to do it, so I have to hope someone else puts such a list together.

I try to learn more about sensors in general, so I find it an interesting topic and I'll throw in two more questions I think you might know the answer to:

If Sony sensor division can make 64 ISO sensors, why don't Sony imaging use that in their cameras as well? Is there a drawback to use 64 ISO vs 100? Or why settle for 100?

Same question goes for an earlier discussion about Nikon using the 14-bit read out mode and Sony stuck with the 12-bit one in their camera using the same sensor. Same question here, is there a drawback with using the 14-bit mode vs the 12-bit one? If not, why settle for 12?

Thanks!

First thing, a sensor does not have an 'ISO', so there is no such thing as a 64 ISO or 100 ISO sensor. What a sensor has is a saturation exposure ultimately determined by the 'full well capacity', the sensor's quantum efficiency and the size of the pixel. In deciding the lowest ISO to have on your camera the decision is how much of that saturation exposure to use, and not all design teams come to the same conclusion. The closer you get to the saturation exposure the more non-linearity you get, so it's not a free lunch. Also, the raw saturation level is usually somewhat higher than the JPEG 100% level which is what determines the ISO rating. Different designers have different views on how much raw headroom to leave. Then there is marketing input. Some marketing teams think that a high ISO rating is more attractive than a lower one. Nikon used a 200 base ISO for years when competitors using the same sensors were using 100. Olympus went on with 200, and over a stop of raw headroom, even though the smaller sensor could have really benefitted for a bigger exposure. In Nikon's case, once they abandon a practice, they often swing over in the other direction. For years DX was the best solution, when they finally adopted full-frame, they released a whole raft of FF models. So having taken a long time to accept 100 base ISO, they went even lower with the D810.

Sony's designers or marketing seem not to be very interested in a low base ISO, and I suspect it's not a very marketable feature. The same is true of every company except Nikon (Nikon is right on this one, BTW).

As for the 12 versus 14 bit, at that time on the Sony Exmors 14 bit was really slow. You were limited to 2.5 FPS on the D300 and 1.8 FPS on the D3X. At the time there was a lot of discussion whether 14-bit really brought any tangible advantage, so it wasn't a matter that was a clear choice. I suspect that Sony Imaging specced their new camera with 12-bit, Nikon specced theirs with 14 bit, and Sony Semiconductor managed to meet the needs of both with a sensor that had a 14-bit mode, but slow. It wasn't until a sensor generation later that Sony SS developed a new ADC design which allowed them to do 14 bit much faster.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

lattesweden
lattesweden Veteran Member • Posts: 4,190
Re: Nikon Sensor Development
1

bobn2 wrote:

lattesweden wrote:

bobn2 wrote:

the D850 is a Sony designed sensor with Sony BSI pixel design. In any case, there is no real difficulty in making a sensor have sufficient saturation capacity to work down to a base ISO of 64, it's just a case of increasing the pixel capacitance. The penalty is a bit of a rise in read noise, but if you have Sony's dual gain architecture, where the capacitance is switchable, that isn't an issue.

First thanks for pointing out the Wiki page for Sony sensors in another post as an answer to my question. It gives a little better view and understanding of what Sony has made sensor wise, but unfortunately not what the others manufacturers have done and the most interesting one - which camera body uses which sensor, though there where some examples. I think many would like that sort of list with camera model, sensor model and maybe a small comment if needed. I don't have the knowledge to do it, so I have to hope someone else puts such a list together.

I try to learn more about sensors in general, so I find it an interesting topic and I'll throw in two more questions I think you might know the answer to:

If Sony sensor division can make 64 ISO sensors, why don't Sony imaging use that in their cameras as well? Is there a drawback to use 64 ISO vs 100? Or why settle for 100?

Same question goes for an earlier discussion about Nikon using the 14-bit read out mode and Sony stuck with the 12-bit one in their camera using the same sensor. Same question here, is there a drawback with using the 14-bit mode vs the 12-bit one? If not, why settle for 12?

Thanks!

First thing, a sensor does not have an 'ISO', so there is no such thing as a 64 ISO or 100 ISO sensor. What a sensor has is a saturation exposure ultimately determined by the 'full well capacity', the sensor's quantum efficiency and the size of the pixel. In deciding the lowest ISO to have on your camera the decision is how much of that saturation exposure to use, and not all design teams come to the same conclusion. The closer you get to the saturation exposure the more non-linearity you get, so it's not a free lunch. Also, the raw saturation level is usually somewhat higher than the JPEG 100% level which is what determines the ISO rating. Different designers have different views on how much raw headroom to leave. Then there is marketing input. Some marketing teams think that a high ISO rating is more attractive than a lower one. Nikon used a 200 base ISO for years when competitors using the same sensors were using 100. Olympus went on with 200, and over a stop of raw headroom, even though the smaller sensor could have really benefitted for a bigger exposure. In Nikon's case, once they abandon a practice, they often swing over in the other direction. For years DX was the best solution, when they finally adopted full-frame, they released a whole raft of FF models. So having taken a long time to accept 100 base ISO, they went even lower with the D810.

Sony's designers or marketing seem not to be very interested in a low base ISO, and I suspect it's not a very marketable feature. The same is true of every company except Nikon (Nikon is right on this one, BTW).

As for the 12 versus 14 bit, at that time on the Sony Exmors 14 bit was really slow. You were limited to 2.5 FPS on the D300 and 1.8 FPS on the D3X. At the time there was a lot of discussion whether 14-bit really brought any tangible advantage, so it wasn't a matter that was a clear choice. I suspect that Sony Imaging specced their new camera with 12-bit, Nikon specced theirs with 14 bit, and Sony Semiconductor managed to meet the needs of both with a sensor that had a 14-bit mode, but slow. It wasn't until a sensor generation later that Sony SS developed a new ADC design which allowed them to do 14 bit much faster.

Amazing well written answer. Thank you very much!

-- hide signature --

Best regards
/Anders
----------------------------------------------------
Mirrorless, mirrorless on the wall, say which is the best camera of them all?
When I put my camera in Manual mode, why don't I get any instructions?
Some images:
https://www.dpreview.com/forums/post/64169208
https://www.dpreview.com/forums/post/64221482

BGD300V1
BGD300V1 Senior Member • Posts: 2,757
Re: Nikon Sensor Development
1

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

So that is the gist of it. Most of the sensors that Nikon uses are either solely designed by Sony SS, or designed by Sony SS to Nikon specifications, possibly with Nikon adding specific CFA and microlens layers. This article fails to make that differentiation and suggests that all Nikon's sensor design work is done in house, which is false. My suspicion is that Nikon didn't lie. Instead Etchells was fed weasel words, and let the weasel bite him.

That's cute.

Try this Wiki Page and click on the "utilizing devices" on the right hand side.

There are plenty of places where Nikon shares production model sensors with other manufactures.

There are others though which ONLY show Nikon use.

For example the IMX321 is shown only as used in the Nikon D500, D7500 and Z50. In all of the production numbers it is the only model showing a 5568 x 3712 20.9 MP configuration.

The IMX309 is shown only as used in the D850 and Z7. More detail shows that there is an A and B version. It is a 8256 x 5504 45.4 MP configuration.  I cannot find that configuration on any other camera.

It would appear in at least these cases that Nikon has contracted to Sony to fabricate a Nikon designed sensor.

There are other instances where a particular part appears only in Nikon and Sony cameras to the exclusion of others.

it would appear that there is cross-licensing there of some sort without knowing who the actual designer was.

As another example, the Z6 sensor is shared by Sony, Panasonic, Sigma, and Nikon as an example, perhaps, of a commercial sensor model.

-- hide signature --

I'm a photographer, Jim, not a graphic artist!
My photo blog: http://birdsnbugs.com
RF Stock Portfolio - http://www.dreamstime.com/resp129611

 BGD300V1's gear list:BGD300V1's gear list
Nikon 1 V2 Nikon 1 J5 Nikon Z7 Nikon Z6 Nikon Z50 +12 more
bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
9

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

So that is the gist of it. Most of the sensors that Nikon uses are either solely designed by Sony SS, or designed by Sony SS to Nikon specifications, possibly with Nikon adding specific CFA and microlens layers. This article fails to make that differentiation and suggests that all Nikon's sensor design work is done in house, which is false. My suspicion is that Nikon didn't lie. Instead Etchells was fed weasel words, and let the weasel bite him.

That's cute.

Try this Wiki Page and click on the "utilizing devices" on the right hand side.

There are plenty of places where Nikon shares production model sensors with other manufactures.

There are others though which ONLY show Nikon use.

For example the IMX321 is shown only as used in the Nikon D500, D7500 and Z50. In all of the production numbers it is the only model showing a 5568 x 3712 20.9 MP configuration.

The IMX309 is shown only as used in the D850 and Z7. More detail shows that there is an A and B version. It is a 8256 x 5504 45.4 MP configuration. I cannot find that configuration on any other camera.

It would appear in at least these cases that Nikon has contracted to Sony to fabricate a Nikon designed sensor.

No it doesn't. Point one, Sony SS does not offer an image sensor foundry service, apart from seeing out the Nikon D5 sensor, that was being fabbed on the Toshiba line that Sony bought. The part number indicates the designer. If a foundry service was used, then there will be a foundry part identifier somewhere on the die, but the company that designs the chip gets to decide the part number. If it is an 'IMX' number, it indicates that it came from the Sony design team, the number after the 'IMX' is simply the job number of the design team. What these chips being exclusive to Nikon means is that Nikon commissioned the design to its specifications, it was designed to those specifications by the Sony team and Nikon has secured exclusivity.

Nikon's practice is not to use Sony fabrication if it designs a sensor, the whole point is to maintain independence. They used the Renesas line for their sensors up to the D4S. Sony bought that line. If they were happy to continue using it, and Sony offered the service, they could have continued to use it, but they didn't, they moved to a Toshiba line. Moving fab providers is not straightforward, so they would not have done it had they not had a reason. Sony subsequently bought the Toshiba line. That's why it's a fair guess that Nikon will move to TPSCo. If they had a running arrangement where Sony fabbed Nikon designed sensors, they would not even be thinking of it, would they? And somehow, the sensors that we know for sure were designed by the Nikon design group are designed completely differently from the ones apparently designed by the same team and fabbed by Sony, yet somehow when they design for Sony fab they design exactly the same as the Sony team. It makes no sense whatsoever.

It also makes no sense to use Sony if you are not going to use its design expertise. Sony has no special sauce in the fabrication of sensors, what it has is loads of leading edge IP which it has built up with years of R&D. If you're not going to use it, why use Sony?

And then, read through the full list. The IMX 109 is used only by Olympus, so Olympus designed it? The IMX159 is used only by Z CAM. Z CAM designed it? The IMX211 is used only by Phase One. Phase One designed it? The IMX269 is used only by Xiaomi. Xiaomi designed it? The IMX183 is used only by DJI. DJI designed it? The IMX277 is used only by GoPro. GoPro designed it? The IMX299 is used only by Panasonic. Panasonic designed it? The IMX322 and IMX286 are used only by Huawei. Huawei designed them? The IMX240, IMX320, IMX333, IMX345, IMX374 and IMX260 are used only by Samsung. Samsung designed them? The IMX356, IMX503, and IMX372 are used only by Apple. Apple designed them?

There are other instances where a particular part appears only in Nikon and Sony cameras to the exclusion of others.

it would appear that there is cross-licensing there of some sort without knowing who the actual designer was.

If it is a Sony IMX sensor, it is a Sony design. Simple as that. If that's too simple for you, the Sony chips employ design patterns that are completely different from the practice of the Nikon designers.

As another example, the Z6 sensor is shared by Sony, Panasonic, Sigma, and Nikon as an example, perhaps, of a commercial sensor model.

It's also in the Sony catalogue, anyone can buy it.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

BGD300V1
BGD300V1 Senior Member • Posts: 2,757
Re: Nikon Sensor Development
3

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

So that is the gist of it. Most of the sensors that Nikon uses are either solely designed by Sony SS, or designed by Sony SS to Nikon specifications, possibly with Nikon adding specific CFA and microlens layers. This article fails to make that differentiation and suggests that all Nikon's sensor design work is done in house, which is false. My suspicion is that Nikon didn't lie. Instead Etchells was fed weasel words, and let the weasel bite him.

That's cute.

Try this Wiki Page and click on the "utilizing devices" on the right hand side.

There are plenty of places where Nikon shares production model sensors with other manufactures.

There are others though which ONLY show Nikon use.

For example the IMX321 is shown only as used in the Nikon D500, D7500 and Z50. In all of the production numbers it is the only model showing a 5568 x 3712 20.9 MP configuration.

The IMX309 is shown only as used in the D850 and Z7. More detail shows that there is an A and B version. It is a 8256 x 5504 45.4 MP configuration. I cannot find that configuration on any other camera.

It would appear in at least these cases that Nikon has contracted to Sony to fabricate a Nikon designed sensor.

And then, read through the full list. The IMX 109 is used only by Olympus, so Olympus designed it? The IMX159 is used only by Z CAM. Z CAM designed it? The IMX211 is used only by Phase One. Phase One designed it? The IMX269 is used only by Xiaomi. Xiaomi designed it? The IMX183 is used only by DJI. DJI designed it? The IMX277 is used only by GoPro. GoPro designed it? The IMX299 is used only by Panasonic. Panasonic designed it? The IMX322 and IMX286 are used only by Huawei. Huawei designed them? The IMX240, IMX320, IMX333, IMX345, IMX374 and IMX260 are used only by Samsung. Samsung designed them? The IMX356, IMX503, and IMX372 are used only by Apple. Apple designed them?

There are other instances where a particular part appears only in Nikon and Sony cameras to the exclusion of others.

it would appear that there is cross-licensing there of some sort without knowing who the actual designer was.

If it is a Sony IMX sensor, it is a Sony design. Simple as that. If that's too simple for you, the Sony chips employ design patterns that are completely different from the practice of the Nikon designers.

As another example, the Z6 sensor is shared by Sony, Panasonic, Sigma, and Nikon as an example, perhaps, of a commercial sensor model.

It's also in the Sony catalogue, anyone can buy it.

Of course it is. That's why three other companies use it.

Here is the Sony catalog. The other two sensors are NOT in that catalog.

https://www.sony-semicon.co.jp/e/products/IS/camera/product.html

Show me the Z7/D850 sensor on that.

Sony buys or licenses a lot of IP.

When they wanted to make Consumer cameras they bought Minolta

When they wanted to make image sensors on a large scale they bought Toshiba

Nikon worked with Renesas to produce their sensors for some cameras.  Sony bought Renesas.

Nikon developed the Nikon 1 Sensors with Aptina, Sony licensed it.

BTW, Olympus and Panasonic (among others) have designed their own sensors in the past.

Kodak made the original micro 4/3 sensors although the later ones were made by panasonic.

Sony's history of innovation is written in their checkbook.

There are plenty of IMX designated sensors that were not designed by Sony.

-- hide signature --

I'm a photographer, Jim, not a graphic artist!
My photo blog: http://birdsnbugs.com
RF Stock Portfolio - http://www.dreamstime.com/resp129611

 BGD300V1's gear list:BGD300V1's gear list
Nikon 1 V2 Nikon 1 J5 Nikon Z7 Nikon Z6 Nikon Z50 +12 more
bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
15

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

So that is the gist of it. Most of the sensors that Nikon uses are either solely designed by Sony SS, or designed by Sony SS to Nikon specifications, possibly with Nikon adding specific CFA and microlens layers. This article fails to make that differentiation and suggests that all Nikon's sensor design work is done in house, which is false. My suspicion is that Nikon didn't lie. Instead Etchells was fed weasel words, and let the weasel bite him.

That's cute.

Try this Wiki Page and click on the "utilizing devices" on the right hand side.

There are plenty of places where Nikon shares production model sensors with other manufactures.

There are others though which ONLY show Nikon use.

For example the IMX321 is shown only as used in the Nikon D500, D7500 and Z50. In all of the production numbers it is the only model showing a 5568 x 3712 20.9 MP configuration.

The IMX309 is shown only as used in the D850 and Z7. More detail shows that there is an A and B version. It is a 8256 x 5504 45.4 MP configuration. I cannot find that configuration on any other camera.

It would appear in at least these cases that Nikon has contracted to Sony to fabricate a Nikon designed sensor.

And then, read through the full list. The IMX 109 is used only by Olympus, so Olympus designed it? The IMX159 is used only by Z CAM. Z CAM designed it? The IMX211 is used only by Phase One. Phase One designed it? The IMX269 is used only by Xiaomi. Xiaomi designed it? The IMX183 is used only by DJI. DJI designed it? The IMX277 is used only by GoPro. GoPro designed it? The IMX299 is used only by Panasonic. Panasonic designed it? The IMX322 and IMX286 are used only by Huawei. Huawei designed them? The IMX240, IMX320, IMX333, IMX345, IMX374 and IMX260 are used only by Samsung. Samsung designed them? The IMX356, IMX503, and IMX372 are used only by Apple. Apple designed them?

There are other instances where a particular part appears only in Nikon and Sony cameras to the exclusion of others.

it would appear that there is cross-licensing there of some sort without knowing who the actual designer was.

If it is a Sony IMX sensor, it is a Sony design. Simple as that. If that's too simple for you, the Sony chips employ design patterns that are completely different from the practice of the Nikon designers.

As another example, the Z6 sensor is shared by Sony, Panasonic, Sigma, and Nikon as an example, perhaps, of a commercial sensor model.

It's also in the Sony catalogue, anyone can buy it.

Of course it is. That's why three other companies use it.

Here is the Sony catalog. The other two sensors are NOT in that catalog.

https://www.sony-semicon.co.jp/e/products/IS/camera/product.html

Show me the Z7/D850 sensor on that.

Yes, there is no disagreement about whether or not that sensor is Nikon exclusive.

Sony buys or licenses a lot of IP.

When they wanted to make Consumer cameras they bought Minolta

When they wanted to make image sensors on a large scale they bought Toshiba

Nikon worked with Renesas to produce their sensors for some cameras. Sony bought Renesas.

Sony didn't buy Renesas. They bought one fab line that Renesas had no use for. Renesas had already announce its intention of moving to a fabless model.

Nikon developed the Nikon 1 Sensors with Aptina, Sony licensed it.

Sony did not licence the Nikon 1 sensors. They did a patent swap agreement with Aptina, which applied to all of their sensors, primarily I suspect for access to the dual gain technology. Anyhow, what they gained was access to Aptina IP, not Nikon. There's also no evidence that 'Nikon developed the Nikon 1 sensors with Aptina'. It looks like a straightforward Aptina design, and those sensors were offered on the open market.

BTW, Olympus and Panasonic (among others) have designed their own sensors in the past.

Yes, but not the ones they bought from Sony. Panasonic spun off their sensor design and build operation. It became TPSCo, which is the company we are talking about as a probable fab partner for Nikon. Why would they need another fab partner if they were already getting fab services from Sony, which can already do FF stacked sensors?

Kodak made the original micro 4/3 sensors although the later ones were made by panasonic.

Kodak made and designed the original FT sensors, as Panasonic later designed and made them.

Sony's history of innovation is written in their checkbook.

Absolutely. That's why they can no do sensors better than any other company and they dominate the market.

There are plenty of IMX designated sensors that were not designed by Sony.

Nonsense. Sony SS's business model is based on added value of their sensor design capability. They do not operate as a foundry. Customers come to Sony SS because they can design and build, not for foundry services. Part of the business model is to design and build to order, which is why so many companies (such as Apple for instance) have contracted Sony to design and build a sensor to their requirements. Nikon is another such customer. Foundry companies have a business model that works completely differently, the support that they gives their customers is completely different and Sony doesn't off that kind of support.

The scenario that you are putting forward makes absolutely no sense. Sony SS spends loads of money developing and acquiring the technology. Nikon, which has a history of designing sensors according to its own practices, and using independent fab partners, decides that it's only going to use Sony as a fab partner, passing up all the Sony tech and design expertise, an as it does so its designers suddenly start designing like Sony engineers and including all kinds of Sony IP. And apparently, according to your loads of other companies are doing the same thing.

That's the disservice that Etchell's dreadful article has done to the service of truth, it has provided a bogusly authoritative source for Nikon fantasists to hang their hat on.

There is no need to make stuff up. What Nikon does with respect to sensors is extremely sensible. They buy what is the best sensor they can for the job, whether it is in-house, Sony or another vendor. They maintain a sensor design operation which has shown itself capable of building sensors which for some specialist purposes expand the state of the art. They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony. It's a very sound policy, and there is no need to dress it up by insisting Nikon designs what are very obviously Sony sourced sensors.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

lattesweden
lattesweden Veteran Member • Posts: 4,190
Re: Nikon Sensor Development

bobn2 wrote:

There are plenty of IMX designated sensors that were not designed by Sony.

Nonsense. Sony SS's business model is based on added value of their sensor design capability. They do not operate as a foundry. Customers come to Sony SS because they can design and build, not for foundry services. Part of the business model is to design and build to order, which is why so many companies (such as Apple for instance) have contracted Sony to design and build a sensor to their requirements. Nikon is another such customer. Foundry companies have a business model that works completely differently, the support that they gives their customers is completely different and Sony doesn't off that kind of support.

The scenario that you are putting forward makes absolutely no sense. Sony SS spends loads of money developing and acquiring the technology. Nikon, which has a history of designing sensors according to its own practices, and using independent fab partners, decides that it's only going to use Sony as a fab partner, passing up all the Sony tech and design expertise, an as it does so its designers suddenly start designing like Sony engineers and including all kinds of Sony IP. And apparently, according to your loads of other companies are doing the same thing.

That's the disservice that Etchell's dreadful article has done to the service of truth, it has provided a bogusly authoritative source for Nikon fantasists to hang their hat on.

There is no need to make stuff up. What Nikon does with respect to sensors is extremely sensible. They buy what is the best sensor they can for the job, whether it is in-house, Sony or another vendor. They maintain a sensor design operation which has shown itself capable of building sensors which for some specialist purposes expand the state of the art. They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony. It's a very sound policy, and there is no need to dress it up by insisting Nikon designs what are very obviously Sony sourced sensors.

Thanks for all the good info!

Below some questions that i put as statements. Are these true or not?

If I understand this correctly:

Anyone can go to Sony sensor division and get an off the shelf sensor that they (Sony sensor division) put together on chance and hope that someone will buy. Such a sensor would also be in their normal sales catalogue under an IMX number.

Also anyone can go to Sony sensor division and ask them to do a special sensor one then has exclusivity on. Such a sensor would still get an IMX number but not be in the normal Sony sensor division sales catalogue.

Can one also go to Sony sensor division and ask for a special designed sensor that one uses first exclusively, and that it later, after the agreed x amount of time has passed comes into the normal sales catalogue and one gets some sort of kick back if it then later also does sell to others?

The "Anyone" term above applies to both another camera maker as well as Sonys Camera division, who also is a customer at Sony Sensor Division (at least I have seen that claim in interviews with Sony Camera division managers).

If Sony manufacturs the sensor, they always also designs it. They don't offer any other business model.

Does all IP (patents) that Sony have apply for use for all their customers, meaning Sonys sensor division IP has no limitations with regards to whom it is sold to?

Can a customer that comes to Sony sensor division take with them their own IPs (patents) or IPs that they have bought for their own use and get those implemented in an for them exclusive made sensor?

There are some cases where the customer can modify/add on to an existing sensor, like having ones own microlenses etc if I got this right? Would that render a new IMX number, or how does that work?

-- hide signature --

Best regards
/Anders
----------------------------------------------------
Mirrorless, mirrorless on the wall, say which is the best camera of them all?
When I put my camera in Manual mode, why don't I get any instructions?
Some images:
https://www.dpreview.com/forums/post/64169208
https://www.dpreview.com/forums/post/64221482

Dibyendu Majumdar Contributing Member • Posts: 842
Re: Upcoming Nikon Z9 Sensor by TowerJazz (now Semiconductor)??

I believed that Nikon would use Tower for the newest pro model - and the reason they continued D5 sensor into D6 was that their new sensor tech wasn't ready.

Not based on any insights into relationships but simply that Nikon has since beginning preferred to have a lot of input into the sensors used in D-single digit series.

But reading between the lines of the announcement - there is a conspicuous absence of 'Nikon designed' in the statement. I think that it likely in this instance they will use a Sony sensor.

bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
5

lattesweden wrote:

bobn2 wrote:

There are plenty of IMX designated sensors that were not designed by Sony.

Nonsense. Sony SS's business model is based on added value of their sensor design capability. They do not operate as a foundry. Customers come to Sony SS because they can design and build, not for foundry services. Part of the business model is to design and build to order, which is why so many companies (such as Apple for instance) have contracted Sony to design and build a sensor to their requirements. Nikon is another such customer. Foundry companies have a business model that works completely differently, the support that they gives their customers is completely different and Sony doesn't off that kind of support.

The scenario that you are putting forward makes absolutely no sense. Sony SS spends loads of money developing and acquiring the technology. Nikon, which has a history of designing sensors according to its own practices, and using independent fab partners, decides that it's only going to use Sony as a fab partner, passing up all the Sony tech and design expertise, an as it does so its designers suddenly start designing like Sony engineers and including all kinds of Sony IP. And apparently, according to your loads of other companies are doing the same thing.

That's the disservice that Etchell's dreadful article has done to the service of truth, it has provided a bogusly authoritative source for Nikon fantasists to hang their hat on.

There is no need to make stuff up. What Nikon does with respect to sensors is extremely sensible. They buy what is the best sensor they can for the job, whether it is in-house, Sony or another vendor. They maintain a sensor design operation which has shown itself capable of building sensors which for some specialist purposes expand the state of the art. They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony. It's a very sound policy, and there is no need to dress it up by insisting Nikon designs what are very obviously Sony sourced sensors.

Thanks for all the good info!

Below some questions that i put as statements. Are these true or not?

If I understand this correctly:

Anyone can go to Sony sensor division and get an off the shelf sensor that they (Sony sensor division) put together on chance and hope that someone will buy. Such a sensor would also be in their normal sales catalogue under an IMX number.

Pretty much everyone. Sony SS doesn't sell retail, so you need to be a company, and if you're a small company, you'll go through a distributor, rather than direct. This is the normal practice in the semiconductor industry. That is, Sony never gets to say 'yes' or 'no', because the sales go through the distributor. You can't get the data that you'd need to design a camera using the sensor without some kind of agreement with the distributor, and most distributors will do a bit of vetting to check you're a serious buyer and not a hobbyist.

Also anyone can go to Sony sensor division and ask them to do a special sensor one then has exclusivity on. Such a sensor would still get an IMX number but not be in the normal Sony sensor division sales catalogue.

Exactly. You need to make doing the R&D for the special sensor worthwhile for Sony. It's less than you might think, because Sony has big subsystem libraries of subcomponents that they know work. It's mostly parameterised, so it's not a case of an electronics engineer laboriously laying out each pixel. They select a standard pixel design, lay out an array of the number of pixels you want, put in stripes of ADC and so on. Basically you get a menu of what they can offer, you tell them what you want and they do it. It's usually a bit iterative, that is, as a customer you get approvals along the line. It strikes me that could be called 'designing' the sensor. What you're actually doing is assembling it from Sony's kit of parts, and the actual design work will be done by their engineers.

In terms of what it costs, you might pay them something for the design work, and then buy the sensor at standard prices, or you might guarantee a minimum purchase, where the development costs are loaded into the price you pay.

Can one also go to Sony sensor division and ask for a special designed sensor that one uses first exclusively, and that it later, after the agreed x amount of time has passed comes into the normal sales catalogue and one gets some sort of kick back if it then later also does sell to others?

Deals like that have happened. For instance, Nikon had a year's exclusive on the IMX094 in the D800 before Sony imaging had use of it, and it eventually became a commodity sensor. This kind of deal will cost you less, because you're leaving Sony SS with more ways of making money from the sensor.

The "Anyone" term above applies to both another camera maker as well as Sonys Camera division, who also is a customer at Sony Sensor Division (at least I have seen that claim in interviews with Sony Camera division managers).

If Sony manufacturs the sensor, they always also designs it. They don't offer any other business model.

Not for sensors. They do run a MEMS foundry, where you design the chip and they make it. But for sensors, there would be no point. Thera are quite a few image sensor foundries. What you get with Sony is access to their design team and IP.

Does all IP (patents) that Sony have apply for use for all their customers, meaning Sonys sensor division IP has no limitations with regards to whom it is sold to?

Not all IP is patents. In fact, most of it isn't.

Can a customer that comes to Sony sensor division take with them their own IPs (patents) or IPs that they have bought for their own use and get those implemented in an for them exclusive made sensor?

I think that is a misunderstanding of what IP is. For the most part, it's not patents and not even copyright. It's just accumulated knowhow, which is why you end up signing NDAs to protect it. As an example, one of the major advances made with the Exmor sensor was the column ADC. This wasn't a Sony patent, in fact it was an IBM one, which was old enough that it had timed out. No-one had actually made a workable commercial column ADC. Sony did. There were no radical new ideas in it, it was just a case of refining the basic design until it worked. That's why non-Sony engineers don't get to work with the Sony subsystem libraries, they don't want the knowledge base represented by all those refinements getting out, because the only thing protecting it is secrecy.

That is the usual case with technical equipment. Patents represent bright ideas, but generally much more important is knowhow, just the kind of knowledge that comes from lots of trial and error about the best way to do things.

Anyhow, I suppose it is possible for a company to come along with its own bit of IP and ask for Sony's design team to incorporate it. How feasible it would be depends on the nature of the IP. If it was special sauce in the microlenses or CFA, that would be reasonably easy, since those are organic layers on top of the silicon. In fact Fujifilm does just that with its XTrans CFA, but then since they supply Sony SS (and most of the sensor industry) with its CFA film, it's easy for them to do. It's very plausible that the Z sensors have PDAF microlens layers incorporating Nikon IP, in the same way. Nikon is very good at microlenses, and putting its own onto Sony silicon would provide some differentiation.  Some piece of IP deep in the silicon might be harder, because it would have to be compatible with Sony's libraries. As soon as it requires real engineers doing ground-up design, it becomes much more expensive, and also limits the saleability of the product to other customers. And also, there is not much point, because Sony's libraries are very, very good. It's how they came to dominate the market. It's unlikely that another company has some piece of IP which is going to make the sensor perform radically better than Sony's design.

There are some cases where the customer can modify/add on to an existing sensor, like having ones own microlenses etc if I got this right? Would that render a new IMX number, or how does that work?

See the above. As to the part number, It's not always 'IMX', it can be 'ICX' or 'ISX' depending on the sensor. The 'I' is image' The 'C' is 'CCD', the 'M' is 'MOS', the 'S' is 'system on a chip', meaning the sensor has image processing and the like built in. Not sure what the 'X' is. Then the rest is just what would be 'drawing office numbers', standard engineering practice. Each new project gets a new number. What constitutes a 'new project' isn't always clear.

Nikon's design team uses 'NC81xxx' numbers. I suspect the 'NC' stands for 'Nikon CMOS', I don't know what the '81' means. it might be a stocking or department code in Nikon's organisation. The D5 sensor, which was a co-production with Toshiba's design team was a 'T4K54'. I've no idea how that came about. It's also not a standard Toshiba number.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

lattesweden
lattesweden Veteran Member • Posts: 4,190
Re: Nikon Sensor Development

bobn2 wrote:

lattesweden wrote:

bobn2 wrote:

There are plenty of IMX designated sensors that were not designed by Sony.

Nonsense. Sony SS's business model is based on added value of their sensor design capability. They do not operate as a foundry. Customers come to Sony SS because they can design and build, not for foundry services. Part of the business model is to design and build to order, which is why so many companies (such as Apple for instance) have contracted Sony to design and build a sensor to their requirements. Nikon is another such customer. Foundry companies have a business model that works completely differently, the support that they gives their customers is completely different and Sony doesn't off that kind of support.

The scenario that you are putting forward makes absolutely no sense. Sony SS spends loads of money developing and acquiring the technology. Nikon, which has a history of designing sensors according to its own practices, and using independent fab partners, decides that it's only going to use Sony as a fab partner, passing up all the Sony tech and design expertise, an as it does so its designers suddenly start designing like Sony engineers and including all kinds of Sony IP. And apparently, according to your loads of other companies are doing the same thing.

That's the disservice that Etchell's dreadful article has done to the service of truth, it has provided a bogusly authoritative source for Nikon fantasists to hang their hat on.

There is no need to make stuff up. What Nikon does with respect to sensors is extremely sensible. They buy what is the best sensor they can for the job, whether it is in-house, Sony or another vendor. They maintain a sensor design operation which has shown itself capable of building sensors which for some specialist purposes expand the state of the art. They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony. It's a very sound policy, and there is no need to dress it up by insisting Nikon designs what are very obviously Sony sourced sensors.

Thanks for all the good info!

Below some questions that i put as statements. Are these true or not?

If I understand this correctly:

Anyone can go to Sony sensor division and get an off the shelf sensor that they (Sony sensor division) put together on chance and hope that someone will buy. Such a sensor would also be in their normal sales catalogue under an IMX number.

Pretty much everyone. Sony SS doesn't sell retail, so you need to be a company, and if you're a small company, you'll go through a distributor, rather than direct. This is the normal practice in the semiconductor industry. That is, Sony never gets to say 'yes' or 'no', because the sales go through the distributor. You can't get the data that you'd need to design a camera using the sensor without some kind of agreement with the distributor, and most distributors will do a bit of vetting to check you're a serious buyer and not a hobbyist.

Also anyone can go to Sony sensor division and ask them to do a special sensor one then has exclusivity on. Such a sensor would still get an IMX number but not be in the normal Sony sensor division sales catalogue.

Exactly. You need to make doing the R&D for the special sensor worthwhile for Sony. It's less than you might think, because Sony has big subsystem libraries of subcomponents that they know work. It's mostly parameterised, so it's not a case of an electronics engineer laboriously laying out each pixel. They select a standard pixel design, lay out an array of the number of pixels you want, put in stripes of ADC and so on. Basically you get a menu of what they can offer, you tell them what you want and they do it. It's usually a bit iterative, that is, as a customer you get approvals along the line. It strikes me that could be called 'designing' the sensor. What you're actually doing is assembling it from Sony's kit of parts, and the actual design work will be done by their engineers.

In terms of what it costs, you might pay them something for the design work, and then buy the sensor at standard prices, or you might guarantee a minimum purchase, where the development costs are loaded into the price you pay.

Can one also go to Sony sensor division and ask for a special designed sensor that one uses first exclusively, and that it later, after the agreed x amount of time has passed comes into the normal sales catalogue and one gets some sort of kick back if it then later also does sell to others?

Deals like that have happened. For instance, Nikon had a year's exclusive on the IMX094 in the D800 before Sony imaging had use of it, and it eventually became a commodity sensor. This kind of deal will cost you less, because you're leaving Sony SS with more ways of making money from the sensor.

The "Anyone" term above applies to both another camera maker as well as Sonys Camera division, who also is a customer at Sony Sensor Division (at least I have seen that claim in interviews with Sony Camera division managers).

If Sony manufacturs the sensor, they always also designs it. They don't offer any other business model.

Not for sensors. They do run a MEMS foundry, where you design the chip and they make it. But for sensors, there would be no point. Thera are quite a few image sensor foundries. What you get with Sony is access to their design team and IP.

Does all IP (patents) that Sony have apply for use for all their customers, meaning Sonys sensor division IP has no limitations with regards to whom it is sold to?

Not all IP is patents. In fact, most of it isn't.

Can a customer that comes to Sony sensor division take with them their own IPs (patents) or IPs that they have bought for their own use and get those implemented in an for them exclusive made sensor?

I think that is a misunderstanding of what IP is. For the most part, it's not patents and not even copyright. It's just accumulated knowhow, which is why you end up signing NDAs to protect it. As an example, one of the major advances made with the Exmor sensor was the column ADC. This wasn't a Sony patent, in fact it was an IBM one, which was old enough that it had timed out. No-one had actually made a workable commercial column ADC. Sony did. There were no radical new ideas in it, it was just a case of refining the basic design until it worked. That's why non-Sony engineers don't get to work with the Sony subsystem libraries, they don't want the knowledge base represented by all those refinements getting out, because the only thing protecting it is secrecy.

That is the usual case with technical equipment. Patents represent bright ideas, but generally much more important is knowhow, just the kind of knowledge that comes from lots of trial and error about the best way to do things.

Anyhow, I suppose it is possible for a company to come along with its own bit of IP and ask for Sony's design team to incorporate it. How feasible it would be depends on the nature of the IP. If it was special sauce in the microlenses or CFA, that would be reasonably easy, since those are organic layers on top of the silicon. In fact Fujifilm does just that with its XTrans CFA, but then since they supply Sony SS (and most of the sensor industry) with its CFA film, it's easy for them to do. It's very plausible that the Z sensors have PDAF microlens layers incorporating Nikon IP, in the same way. Nikon is very good at microlenses, and putting its own onto Sony silicon would provide some differentiation. Some piece of IP deep in the silicon might be harder, because it would have to be compatible with Sony's libraries. As soon as it requires real engineers doing ground-up design, it becomes much more expensive, and also limits the saleability of the product to other customers. And also, there is not much point, because Sony's libraries are very, very good. It's how they came to dominate the market. It's unlikely that another company has some piece of IP which is going to make the sensor perform radically better than Sony's design.

There are some cases where the customer can modify/add on to an existing sensor, like having ones own microlenses etc if I got this right? Would that render a new IMX number, or how does that work?

See the above. As to the part number, It's not always 'IMX', it can be 'ICX' or 'ISX' depending on the sensor. The 'I' is image' The 'C' is 'CCD', the 'M' is 'MOS', the 'S' is 'system on a chip', meaning the sensor has image processing and the like built in. Not sure what the 'X' is. Then the rest is just what would be 'drawing office numbers', standard engineering practice. Each new project gets a new number. What constitutes a 'new project' isn't always clear.

Nikon's design team uses 'NC81xxx' numbers. I suspect the 'NC' stands for 'Nikon CMOS', I don't know what the '81' means. it might be a stocking or department code in Nikon's organisation. The D5 sensor, which was a co-production with Toshiba's design team was a 'T4K54'. I've no idea how that came about. It's also not a standard Toshiba number.

Fantastic, that cleared up a lot of my wondering. Thank you so much for taking the time to answer these questions!

-- hide signature --

Best regards
/Anders
----------------------------------------------------
Mirrorless, mirrorless on the wall, say which is the best camera of them all?
When I put my camera in Manual mode, why don't I get any instructions?
Some images:
https://www.dpreview.com/forums/post/64169208
https://www.dpreview.com/forums/post/64221482

12HT75 Regular Member • Posts: 128
Re: Nikon Sensor Development

bobn2 wrote:

See the above. As to the part number, It's not always 'IMX', it can be 'ICX' or 'ISX' depending on the sensor. The 'I' is image' The 'C' is 'CCD', the 'M' is 'MOS', the 'S' is 'system on a chip', meaning the sensor has image processing and the like built in. Not sure what the 'X' is.

X for Exmor? (Just guessing!)

 12HT75's gear list:12HT75's gear list
Pentax K-r Pentax K-5 IIs Pentax smc DA 18-135mm F3.5-5.6ED AL [IF] DC WR HD Pentax DA AF 1.4X AW Rear Converter Pentax FA 150-450mm F4.5-5.6
Jeff Foong New Member • Posts: 3
Re: Nikon Sensor Development

Interesting topic, but the question nobody is asking is why.  Why would Nikon want to start new at Tower?  We know the market is shrinking.  Unless this move gives them an edge in performance or/and price, it is a high risk move.  And it better be a SIGNIFICANT improvement in performance and/or price to move the marketshare needle in any way.  Just throwing this in the mix.

bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
1

12HT75 wrote:

bobn2 wrote:

See the above. As to the part number, It's not always 'IMX', it can be 'ICX' or 'ISX' depending on the sensor. The 'I' is image' The 'C' is 'CCD', the 'M' is 'MOS', the 'S' is 'system on a chip', meaning the sensor has image processing and the like built in. Not sure what the 'X' is.

X for Exmor? (Just guessing!)

It predates Exmor.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
5

Jeff Foong wrote:

Interesting topic, but the question nobody is asking is why. Why would Nikon want to start new at Tower? We know the market is shrinking. Unless this move gives them an edge in performance or/and price, it is a high risk move. And it better be a SIGNIFICANT improvement in performance and/or price to move the marketshare needle in any way. Just throwing this in the mix.

It's not a high risk move, since they always have the backup of going to Sony. Also, it isn't 'starting new'. Neither TPSCo (as opposed to Tower) or Nikon is new to the production and design of image sensors in commercial quality. It's an arrangement that works for them all. TPSCo gets Nikon support to upgrade their capability to FF, BSI and stacking, all of which helps them in their market. Nikon keeps a second source for sensors, and maintains its ability to have a unique sensor for its top model. It keeps Sony honest, because they always have to compete for a Nikon design-in. And as we have seen, Nikon having its own sensor capability is emotionally very important to quite a lot of people. For some people it's the mark of a proper camera company, Canon and Sony both have in-house sensor capability, so to be at the same table they think that Nikon must too.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

Chris Mak Senior Member • Posts: 2,558
Re: Nikon Sensor Development

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

bobn2 wrote:

BGD300V1 wrote:

This might explain a few things.

https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab

A clip from the article

That article is the source of so much of the nonsense which goes around about the origins of Nikon's sensors. Take it with a good pinch of salt.

Perhaps you would point out the specific nonsense.

So that is the gist of it. Most of the sensors that Nikon uses are either solely designed by Sony SS, or designed by Sony SS to Nikon specifications, possibly with Nikon adding specific CFA and microlens layers. This article fails to make that differentiation and suggests that all Nikon's sensor design work is done in house, which is false. My suspicion is that Nikon didn't lie. Instead Etchells was fed weasel words, and let the weasel bite him.

That's cute.

Try this Wiki Page and click on the "utilizing devices" on the right hand side.

There are plenty of places where Nikon shares production model sensors with other manufactures.

There are others though which ONLY show Nikon use.

For example the IMX321 is shown only as used in the Nikon D500, D7500 and Z50. In all of the production numbers it is the only model showing a 5568 x 3712 20.9 MP configuration.

The IMX309 is shown only as used in the D850 and Z7. More detail shows that there is an A and B version. It is a 8256 x 5504 45.4 MP configuration. I cannot find that configuration on any other camera.

It would appear in at least these cases that Nikon has contracted to Sony to fabricate a Nikon designed sensor.

And then, read through the full list. The IMX 109 is used only by Olympus, so Olympus designed it? The IMX159 is used only by Z CAM. Z CAM designed it? The IMX211 is used only by Phase One. Phase One designed it? The IMX269 is used only by Xiaomi. Xiaomi designed it? The IMX183 is used only by DJI. DJI designed it? The IMX277 is used only by GoPro. GoPro designed it? The IMX299 is used only by Panasonic. Panasonic designed it? The IMX322 and IMX286 are used only by Huawei. Huawei designed them? The IMX240, IMX320, IMX333, IMX345, IMX374 and IMX260 are used only by Samsung. Samsung designed them? The IMX356, IMX503, and IMX372 are used only by Apple. Apple designed them?

There are other instances where a particular part appears only in Nikon and Sony cameras to the exclusion of others.

it would appear that there is cross-licensing there of some sort without knowing who the actual designer was.

If it is a Sony IMX sensor, it is a Sony design. Simple as that. If that's too simple for you, the Sony chips employ design patterns that are completely different from the practice of the Nikon designers.

As another example, the Z6 sensor is shared by Sony, Panasonic, Sigma, and Nikon as an example, perhaps, of a commercial sensor model.

It's also in the Sony catalogue, anyone can buy it.

Of course it is. That's why three other companies use it.

Here is the Sony catalog. The other two sensors are NOT in that catalog.

https://www.sony-semicon.co.jp/e/products/IS/camera/product.html

Show me the Z7/D850 sensor on that.

Yes, there is no disagreement about whether or not that sensor is Nikon exclusive.

Sony buys or licenses a lot of IP.

When they wanted to make Consumer cameras they bought Minolta

When they wanted to make image sensors on a large scale they bought Toshiba

Nikon worked with Renesas to produce their sensors for some cameras. Sony bought Renesas.

Sony didn't buy Renesas. They bought one fab line that Renesas had no use for. Renesas had already announce its intention of moving to a fabless model.

Nikon developed the Nikon 1 Sensors with Aptina, Sony licensed it.

Sony did not licence the Nikon 1 sensors. They did a patent swap agreement with Aptina, which applied to all of their sensors, primarily I suspect for access to the dual gain technology. Anyhow, what they gained was access to Aptina IP, not Nikon. There's also no evidence that 'Nikon developed the Nikon 1 sensors with Aptina'. It looks like a straightforward Aptina design, and those sensors were offered on the open market.

BTW, Olympus and Panasonic (among others) have designed their own sensors in the past.

Yes, but not the ones they bought from Sony. Panasonic spun off their sensor design and build operation. It became TPSCo, which is the company we are talking about as a probable fab partner for Nikon. Why would they need another fab partner if they were already getting fab services from Sony, which can already do FF stacked sensors?

Kodak made the original micro 4/3 sensors although the later ones were made by panasonic.

Kodak made and designed the original FT sensors, as Panasonic later designed and made them.

Sony's history of innovation is written in their checkbook.

Absolutely. That's why they can no do sensors better than any other company and they dominate the market.

There are plenty of IMX designated sensors that were not designed by Sony.

Nonsense. Sony SS's business model is based on added value of their sensor design capability. They do not operate as a foundry. Customers come to Sony SS because they can design and build, not for foundry services. Part of the business model is to design and build to order, which is why so many companies (such as Apple for instance) have contracted Sony to design and build a sensor to their requirements. Nikon is another such customer. Foundry companies have a business model that works completely differently, the support that they gives their customers is completely different and Sony doesn't off that kind of support.

The scenario that you are putting forward makes absolutely no sense. Sony SS spends loads of money developing and acquiring the technology. Nikon, which has a history of designing sensors according to its own practices, and using independent fab partners, decides that it's only going to use Sony as a fab partner, passing up all the Sony tech and design expertise, an as it does so its designers suddenly start designing like Sony engineers and including all kinds of Sony IP. And apparently, according to your loads of other companies are doing the same thing.

That's the disservice that Etchell's dreadful article has done to the service of truth, it has provided a bogusly authoritative source for Nikon fantasists to hang their hat on.

There is no need to make stuff up. What Nikon does with respect to sensors is extremely sensible. They buy what is the best sensor they can for the job, whether it is in-house, Sony or another vendor. They maintain a sensor design operation which has shown itself capable of building sensors which for some specialist purposes expand the state of the art. They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony.

Was that a trap, or did Canon simply rest on their laurels and refuse to invest in innovation I wonder. I used the 7DII so got bit by Canon's poor raw performance. I use Nikon now. Apparently Canon either finally saw the urge and decided to invest in sensor innovation a few years ago, or they spied on Sony how it's done.

I don't know, but I find Canon's trailing sensor performance one of the black pages in digital photography, if only compared to where they were with lens design years ago already.

It's a very sound policy, and there is no need to dress it up by insisting Nikon designs what are very obviously Sony sourced sensors.

 Chris Mak's gear list:Chris Mak's gear list
Sony a7R II Nikon D500 Nikon AF-S Teleconverter TC-14E III Zeiss Loxia 25mm F2.4 Nikon 500mm F5.6E PF
bobn2
bobn2 Forum Pro • Posts: 68,333
Re: Nikon Sensor Development
2

Chris Mak wrote:

bobn2 wrote:

They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony.

Was that a trap, or did Canon simply rest on their laurels and refuse to invest in innovation I wonder. I used the 7DII so got bit by Canon's poor raw performance. I use Nikon now. Apparently Canon either finally saw the urge and decided to invest in sensor innovation a few years ago, or they spied on Sony how it's done.

I don't know, but I find Canon's trailing sensor performance one of the black pages in digital photography, if only compared to where they were with lens design years ago already.

I think it was a trap. Semiconductor lines are expensive, even for the second-hand ones used to make image sensors. Canon really never had the sensor volume to fund the lines. They have two lines, both 8" wafers, one at 500nm and one at 180nm. They have to outsource some parts of the process to other lines which have better capability (they use Fujitsu, apparently). Sony has five lines, mostly 12", they go down to 65nm and it's investing $1 Bn in a brand new line. That's not hard when its sensor business is making $1.3 Bn profit each year (which is one reason that the camera tail isn't wagging the huge sensor dog). Despite that capacity, Sony still produces a good proportion of its image sensors at TSMC's foundry service (which is why they aren't offering a foundry service themselves - their lines are more than occupied by their own production). For comparison, the operating profit of the complete Canon corporation is $350 M year on year. The consequence of not having the finer geometries is that more advanced logic functionality can be added to the sensor chips. Canon's old processes are fine for making pixel arrays, but when it comes to high speed ADCs and sequencing circuitry, is problematic. Yet they have so much capital tied up in the plants, that it's hard to just abandon them. If you're making image sensors, and you aren't Sony, fabless is a much more sensible model. It is used by all the other main image sensor companies, including On semiconductor, Omnivision and Foveon.

The cost of maintaining a sensor design team, as Nikon does, is quite small. Apart from the photography, it allows them to produce specialist sensors for their other businesses, for which they likely wouldn't get the volume to make a Sony design and build practicable, and it keeps them with the option of having unique sensors for their cameras when they want them. Even so, their sensors have to depend on chunks of bought-in IP, because a modern-day sensor is much more than just a pixel array.

-- hide signature --

Is it always wrong
for one to have the hots for
Comrade Kim Yo Jong?

Chris Mak Senior Member • Posts: 2,558
Re: Nikon Sensor Development

bobn2 wrote:

Chris Mak wrote:

bobn2 wrote:

They've avoided the trap Canon put itself in, where it was dependent on in-house sensors and found itself incapable of keeping up with Sony.

Was that a trap, or did Canon simply rest on their laurels and refuse to invest in innovation I wonder. I used the 7DII so got bit by Canon's poor raw performance. I use Nikon now. Apparently Canon either finally saw the urge and decided to invest in sensor innovation a few years ago, or they spied on Sony how it's done.

I don't know, but I find Canon's trailing sensor performance one of the black pages in digital photography, if only compared to where they were with lens design years ago already.

I think it was a trap. Semiconductor lines are expensive, even for the second-hand ones used to make image sensors. Canon really never had the sensor volume to fund the lines. They have two lines, both 8" wafers, one at 500nm and one at 180nm. They have to outsource some parts of the process to other lines which have better capability (they use Fujitsu, apparently). Sony has five lines, mostly 12", they go down to 65nm and it's investing $1 Bn in a brand new line. That's not hard when its sensor business is making $1.3 Bn profit each year (which is one reason that the camera tail isn't wagging the huge sensor dog). Despite that capacity, Sony still produces a good proportion of its image sensors at TSMC's foundry service (which is why they aren't offering a foundry service themselves - their lines are more than occupied by their own production). For comparison, the operating profit of the complete Canon corporation is $350 M year on year. The consequence of not having the finer geometries is that more advanced logic functionality can be added to the sensor chips. Canon's old processes are fine for making pixel arrays, but when it comes to high speed ADCs and sequencing circuitry, is problematic. Yet they have so much capital tied up in the plants, that it's hard to just abandon them. If you're making image sensors, and you aren't Sony, fabless is a much more sensible model. It is used by all the other main image sensor companies, including On semiconductor, Omnivision and Foveon.

The cost of maintaining a sensor design team, as Nikon does, is quite small. Apart from the photography, it allows them to produce specialist sensors for their other businesses, for which they likely wouldn't get the volume to make a Sony design and build practicable, and it keeps them with the option of having unique sensors for their cameras when they want them. Even so, their sensors have to depend on chunks of bought-in IP, because a modern-day sensor is much more than just a pixel array.

Thanks, that gives me a better understanding, though it also implies that Canon will never really be able to keep up with Sony sensors. So I assume then, that a stacked sensor for the R1 is unlikely.

 Chris Mak's gear list:Chris Mak's gear list
Sony a7R II Nikon D500 Nikon AF-S Teleconverter TC-14E III Zeiss Loxia 25mm F2.4 Nikon 500mm F5.6E PF
Keyboard shortcuts:
FForum MMy threads