In developing the first HD-ready CMOS sensor for camcorders, why did you set your sights on CMOS instead of CCD?
- Hirose
- CMOS has now caught up to and even surpassed CCD in pixel performance, but back when planning first got underway, CCD still had an edge over CMOS.
But the drawback of CCD is that there's only one output channel. No matter how fast images are read, we're limited to output at about 60 MHz. We can't capture the full HD resolution of 1920 x 1080 pixels.
It's not that CMOS was more useful; CMOS was the only way for HD. (Hirose)
So a CMOS sensor was more useful in achieving HD performance?
- Hirose
- It's not that CMOS was more useful; CMOS was the only way for HD.
CCD-based equipment continues to be used by broadcasters even today. These involve interlaced scanning, in which half the vertical lines of pixels are scanned and displayed at a time. Good vertical resolution is maintained through this alternating method of display. Reading half as many scanning lines cuts the time in half. In contrast, what we attempted was to apply an RGB color filter to individual sensors for separate production of each color. This system must be twice as fast as regular broadcast cameras. And because even broadcast cameras currently push the envelope of speed, HD is out of reach with CCD.
That's when we knew we had to develop an HD CMOS sensor.
- Ishizaki
- With CMOS (and I'm talking about SLRs in this case), we can receive signals from eight sources at once, as shown in the figure. Image acquisition is much faster than with CCD, which is slowed down because it's limited to the two-stage charge-transfer method.
Structural differences between CCD and CMOS
The advantage of CMOS is the faster read rates, then. Forgive the impertinence, but does CMOS have any shortcomings?
- Noda
- Yes, it does. CCD units have a simple structure of photodiodes, transfer gates, and CCDs. Photodiodes, which determine pixel performance, take up a lot of space.
But CMOS also requires MOS components (for transfer in individual pixels and for resetting photodiodes) and pixel amps, besides photodiodes. This leaves less space for photodiodes.
As a countermeasure, the read circuit is shared by a pair of pixels. By creating a structure with four transistors for two pixels, we increased the area available for photodiodes.
