A bit on CCD and CMOS sensors

[cernobila]

Advertisement (gone after registration)

The more I read the more I learn. Panasonic and PanaLeica cameras often use one of these two sensors. For example, the V Lux 20 uses the CCD while the new V Lux 30 uses the CMOS.

 

Below is a summary from "how stuff works dot com". This helps me to understand why identical cameras may yield a different result.

 

CCD sensors, as mentioned above, create high-quality, low-noise images. CMOS sensors, traditionally, are more susceptible to noise.

Because each pixel on a CMOS sensor has several transistors located next to it, the light sensitivity of a CMOS chip tends to be lower. Many of the photons hitting the chip hit the transistors instead of the photodiode.

CMOS traditionally consumes little power. Implementing a sensor in CMOS yields a low-power sensor.

CCDs use a process that consumes lots of power. CCDs consume as much as 100 times more power than an equivalent CMOS sensor.

CMOS chips can be fabricated on just about any standard silicon production line, so they tend to be extremely inexpensive compared to CCD sensors.

CCD sensors have been mass produced for a longer period of time, so they are more mature. They tend to have higher quality and more pixels.

[Jared]

A few years ago, that summary would have been accurate. Now, there are a few pieces that I'd argue with. Specifically, the sections on noise...

 

There are several factors that feed into the overall noise characteristics of a chip (or, more accurately, a camera). Quantum efficiency, read noise, pixel size, and thermal noise all play a role in how noisy a given image looks.

 

CCD's can have an advantage in read noise, but they don't tend to unless the readout speeds are quite low--much slower than is typical in a consumer SLR, digicam, or rangefinder. I'd say this one is a wash.

 

Quantum efficiency is still generally higher in CCD's than in CMOS chips, but the adoption of microlenses has reduced this advantage significantly, and in smaller chips manufacturers are now moving towards back illumination designs which can push the quantum efficiency up near ninety percent in either design.

 

As pixels get smaller and smaller, quantum efficiency becomes more and more important since, with really small pixels, much of the "noise" you seen in an image is actually lack of signal rather than lots of noise. You can have a really low noise chip in terms of read noise and thermal noise, but if there isn't much signal coming in the image is still going to appear grainy because of the low signal-to-noise ratio. Every extra photon that is detected matters now when you are talking about 2-4 micron pixels--even in broad daylight.

 

Thermal noise differences generally don't matter much except in multi-second exposures, so it is critical for astrophotographers but not really for many other applications. If you are an astrophotographer, you are likely to choose a CCD chip and then cool it to reduce thermal signal.

 

All in all, I agree with power, fabrication, and production statements, but I think the differences in noise--particularly as implemented in most consumer digicams, rangefinders, and DSLR's--have eroded. You could make an argument that the advantage actually lies with CMOS now, based mostly on the processing algorithms applied to even RAW frames by Nikon, Canon, etc.. They have gotten awfully good at controlling the noise on CMOS chips without eroding resolution below what the anti-aliasing filter will pass.

[mjh]

Also the bit about “CMOS chips can be fabricated on just about any standard silicon production line” would not be upheld anymore. Yes, it’s still true, but that’s part of what had earned CMOS sensors the bad reputation they enjoyed, say, 10 years ago. If you are serious about high-quality CMOS sensor production you have to optimize the fabrication process for sensor production. “Any standard silicon production line” won’t do.

[cernobila]

Since the V Lux 30 came out I took a little interest in the difference in the two sensors used, CMOS in the 30 and CCD in the V Lux 20. One review web site (camera labs) compared two Lumix cameras with these two sensors and this is a short part of the summary......

 

"The Lumix TZ18 / ZS8 is equipped with a 14.1 Megapixel CCD sensor measuring 1/ 2.33in. This gives the TZ18 / ZS8 two extra Megapixels over its predecessor, allowing you to make prints approximately 1.5in bigger on the diagonal. Note the pricier TZ20 / ZS10 shares the same resolution and sensor size, but switched from CCD to MOS technology to support 1080i video and faster continuous shooting, albeit as we discovered, with reduced photo quality as a result."

 

Since this is a recent test, its likely that there is still some difference between these two sensors.

[zapp]

The advantages of CCD sensors are overcome by CMOS sensors these days - or are close to being eliminated plus CMOS offers many advantages over CCD sensors. Other placed discuss this quite well...

 

For Leica large CMOS sensors not only mean higher ISO, ..., but a way for for compatibility. Mounting R lenses again on Leica cameras would be possible. I would expect the M10 to be CMOS with electronic viewfinder in maybe 1-2 years time.

The X1 was probably the first step towards CMOS technology for Leica. People are expecting an M9 companion at lower price in the near future, and an X2 with interchangeable lenses, ..., I would assume that the X2 is coming out rather soon, allowing to mount M lenses as well as R lenses infront of the CMOS sensor. The smaller sensor dimenions and the less sophisticated viewfinder will place it cleary below the M9 even pricewise, but it may be used as entry into the M line of cameras, make R users happy, provide an answers to 4/3 and M4/3 and start an independent Leica APS-C line.