CCD vs. CMOS, sensitivity in low light improvements with industrial CMOS image sensors and cameras

In the last decade or so, CMOS has been evolving much faster than CCD as a result of investments in the cell phone camera market/consumer market.

The change over in the industrial camera market from CCD to CMOS is now.

There are global shutter CMOS image sensors available with image quality that is now comparable to CCD.  The latest CMOS image sensors even perform better than CCD on key image quality parameters, such as sensitivity in low light.

When we talk about sensitivity of a camera, two sensor parameters are very (equally) important:

QE (Quantum Efficiency), which is a measure of how efficiently the sensor converts light (photons) to charge (electrons). The more electrons in a pixel during the integration period, the higher the output level of the sensor, so the more sensitive the sensor is for that specific wavelength of the light. A QE of 1 means that every photon generates (in average) one electron. Normally the QE is less than 1 (or 100%).

The Read Noise (RN) of the sensor, with is the equivalent noise level (in electrons RMS) at the output of the sensor in the dark and at zero integration time. The lower the read noise level, the lower the minimum number of signal electrons is that can be detected. A lower RN therefore results in a more sensitive sensor. These two combined give the overall sensitivity of the sensor as QE/Read Noise, or the minimum amount of light you can see.

To make the story a bit more complicated: the noise level at the output of the sensor is not only determined by the read noise. Also the Dark Current of a pixel contributes to the Dark Noise (as is it sometimes called). The dark current is dependent on the quality of the sensor, but also very dependent on the temperature and the integration time. At higher temperatures the dark current increases rapidly and the noise caused by the dark current will be higher than the read noise. In order to take the temperature effect in account, it is better to use the Dark Noise instead of the Read Noise.

Using QE/Dark Noise, the higher this value, the more sensitive the sensor is at a particular wavelength.

The figure below shows the QE/Dark Noise for two sensors at two temperatures: room temperature, and a higher temperature. The curves show that the CMOS sensor is more sensitive than the CCD sensor (at all wavelengths). Also, you can see that the sensors perform worse at higher temperatures than at room temperature, as expected.

QE/Dark Noise versus Wavelength

When we tested the latest generation CCD and CMOS image sensors in 2011, and looked at the sensitivity (here the QE/Read Noise), the CCD image sensor was significantly better.

Previously, CCD image sensors offered sensitivity advantages, especially in higher temperatures.  CMOS image sensors used to require additional functionality or even cooling to achieve the same noise performance and image uniformity at higher temperatures.  Now that CMOS image sensors have caught up or even surpassed in performance, it makes sense that almost 75% of industrial cameras are expected to be using CMOS image sensors by 2016 (according to a market survey from Framos).  The change over is happening rapidly, even in applications with challenging lighting conditions such as in defense and other global security systems.  More on this to come…