The Sensor Game Tightens

(news & commentary)

Reuters is reporting that Sony is in discussions to purchase the Toshiba sensor business for 20b yen (US$165m), effectively removing a competitor from the camera sensor supplier scene. 

Toshiba recently got into trouble in Japan for an accounting scandal dating back to 2008/2009. It’s assumed that selling off the sensor business is part of the expected restructuring that was promised by Toshiba’s new CEO, appointed after the scandal broke.

It’s also not particularly unusual for Toshiba and Sony to make deals. Hitachi, Sony, and Toshiba merged their liquid-crystal display operations in a government-backed merger in 2011. But an outright sale is something new. Sony would get the Oita sensor manufacturing plant, and Toshiba would retain its other semiconductor activity and get out of image sensor manufacturing completely.

The Nikon D5500 and D7200 use Toshiba sensors, though these sensors have a very high similarity to Sony Exmor sensors, to the point where they exhibit nearly identical low-level test scores, score differences that are well within the range of sample variation. The Fujifilm X20 and X30 compact cameras are also known to use Toshiba sensors.

What’s interesting to me is something I noted almost a decade ago: Nikon and Sony may have a behind-the-shoji coopetition agreement where they’re trying to unseat Canon. Nikon wouldn’t mind Sony being #2 in cameras as long as Nikon was #1. 

Note how Nikon is playing pollinator: the Aptina sensor used in the Nikon 1 series was the first to have a production quality phase detect on the image sensor. Aptina eventually cross-licensed this technology to Sony as Aptina underwent takeover discussions with a sensor company that specializes in industrial and automotive sensors. With Aptina no longer interested in making camera sensors, we saw Nikon pick up with a Sony version of the 1” sensor with phase detect for the Nikon 1 J5. Now we have Nikon having managing to build an Exmor clone via Toshiba, and Sony looking to take that over. 

But cycle back to when I wrote my original article about the Sony/Nikon coopetition thought: Nikon sensor designers had designed and shipped two CMOS sensors in the D2 generation. Interestingly, the D2x version was downsized by Sony and put into a Sony camera (R1), and there are strong hints of early Exmor in that design. Not long after that we saw changes in Sony-made sensors used by Nikon in the ADC capabilities. 

In essence, Sony has been acquiring both technology and now production capacity via Nikon’s fabless sensor activity. If that wasn’t planned or at least anticipated, Nikon is going to find the world ahead very tricky to navigate. If it was planned or anticipated, it may very well be that Nikon is quietly directing a lot of sensor R&D without being publicly visible while doing so. 

2016 will be particularly interesting in that regard. We have the D5, which should be a Nikon-designed, Nikon-produced sensor, and we should have a followup to the D810 and/or D750, and maybe a D400 if Nikon ever gets that worked out.

What I’ll be looking for in Nikon’s upcoming designs is any sense that Nikon is deploying slightly ahead of the curve in terms of something sensor-wise. The D800, for example, was the first instance of the high-resolution Sony full frame Exmors. As I’ve written, I don’t believe Nikon will pick up the 42mp version that Sony has deployed in the A7rII. (Short answer: it doesn’t give any real resolution boost, no real dynamic range boost, and most of the benefits are on the video side. Competitors like Pentax are picking up on that 36mp sensor, too, so Nikon needs to move further ahead.) 

So, a 20mp D5 with a Nikon sensor, continuing the D3/D3s/D4/D4s lineage. Maybe 54mp in a D850 with a next generation Exmor? That leaves DX, and even Sony seems to be struggling with that. Rumors are that the next generation of APS sensors for the Sony A5xxx/A6xxx/A7xxx had heat and yield issues, which is why we’ve seen no E-mount Sony’s lately (the full frame A7’s are FE-mount). Fujifilm will deploy a version of the existing Sony 24mp APS with the Fujifilm X-trans filtration in their upcoming X generation, so they don’t need whatever the special sauce is in the next generation of APS from Sony. Nikon probably would, for any D7200 followup or heaven forbid, any D400 model that shows up (too early for the D7200 followup, by the way). 

Overall, Sony is now clearly the Intel of dedicated camera sensors. Nikon seems to have a persistent role in that. In terms of volume, only Canon remains in the APS/full frame world as any type of viable competitor. Don’t underestimate Canon. A change in fabs would do them a world of good, and they have a 40% market share in ILC sensors. Still, Sony is making a PacMan-like scooping up of all the little dots in the sensor market. Canon needs a bigger and more visible response in the core technologies, and I think they need to move to at least a 250nm fab to do that. Update: a tech friend who’s well connected has pointed out to me that Canon has been making a few chips for cameras on a 180nm process fab. Also, to those of you who are reading mm instead of nm, those are two different things. When a fab spec is made in mm, they’re referring to the overall size of the die that can be addressed; when specified in nm we’re talking about the smallest feature size that can be created.

text and images © 2017 Thom Hogan
portions Copyright 1999-2016 Thom Hogan-- All Rights Reserved
Follow us on Twitter: @bythom, hashtags #bythom, #dslrbodies