This article describes how the CCD’s structure, operating parameters, and external signal processing circuitry affect the maximum brightness variation that can be captured by the CCD imaging hardware.
We’ve seen the general concept of dynamic range, and we’ve seen dynamic range as a performance specification in an imaging system. In this article, we will discuss the dynamic range of CCDs at the semiconductor and circuit levels. We will consider the following questions: What determines the range of brightness that can be recorded by a CCD and its associated signal processing circuitry?
Maximum signal and noise floor
In a CCD, the input signal is electromagnetic radiation, and the output signal is a charge packet whose number corresponds to the intensity of that electromagnetic radiation.
The dynamic range of the input signal is inherently infinite: brightness can vary from zero (meaning no photons at all) to the intensity produced by the brightest objects in the universe. However, the dynamic range of the output signal is limited.
Zero brightness does not produce zero charge due to noise. With full well capacity, an increase in brightness above a certain point will no longer produce a corresponding increase in charge. Therefore, the dynamic range of a CCD is the ratio of the full well capacity to the noise floor, which is the maximum output signal level a pixel can produce divided by the signal level it would produce even if the pixel had no incident light.
If you’ve been following image sensor technology, you know that the main sources of noise in CCDs are dark noise, photon noise, and read noise. However, when calculating the dynamic range, we only consider dark noise and read noise. I found two reasons: First, photon noise is not a characteristic of the CCD or accompanying readout circuitry – it is inherent in the nature of light and does not differ from one system to another. Second, the photon noise does not affect the minimum output signal level because when the incident light is zero, the photon noise is also zero. This leads us to the following formula:
where NSATURATION is the full well capacity (ie, the number of electrons at which the output signal is saturated), and NNOISE is the sum of dark noise and read noise in electron RMS. If you prefer to use decibels or stop losses instead of normal ratios, we have the following formula:
Full well capacity of CCD
An important factor affecting full well capacity is the area of the pixel or, if only a fraction of the pixel is sensitive to light, the area of the photodiode. A larger photosensitive element corresponds to a larger portion of silicon in which charge can be integrated. Therefore, we can extend the dynamic range by increasing the pixel size. Given that doing so without changing the overall area of the sensor results in a reduction in resolution, a larger sensor is required if the dynamic range is to be increased and resolution is maintained.
Physically larger photodiodes provide more room for the accumulation of photogenerated free electrons. CCDs currently on the market vary widely in full well capacity. For example, Oxford Instruments devices range from 25,000 electrons (pixel area = 100 µm2, dynamic range = 64 dB) to 510,000 electrons (pixel area = 676 µm2, dynamic range = 94 dB).
Effective full well capacity
Operating conditions may affect full well capacity. For example, ON Semi’s KAI-2020 CCD has a full well capacity of 20,000 electrons or 40,000 electrons. The actual physical capacity is closer to 40,000 electrons, but the output amplifier cannot handle 40,000 electrons at the full readout speed of 40 MHz. Therefore, when calculating the dynamic range, we need to consider the effective full well capacity, not just the electron capacity corresponding to the physical properties of the pixel. Likewise, the effective full well capacity of the KAI-2020 photodiode depends on the applied substrate voltage. A lower substrate voltage results in a larger full well capacity (and therefore a larger dynamic range), but also makes the sensor more prone to halo.
Dynamic range and analog-to-digital conversion
Although the dynamic range of a CCD depends on full well capacity and noise, we must remember that the camera must digitize the CCD’s data before processing it. The CCD signal goes through the ADC to the rest of the system, so we need to make sure that the analog-to-digital converter provides enough dynamic range.
If you pay for a high performance CCD with 90 dB dynamic range and then pair it with an 8-bit ADC, the final dynamic range of the image data is the ADC’s dynamic range, which is only 8-bit resolution, which is actually about 48 dB . You lose 42 dB. Fortunately, getting enough dynamic range from an ADC is usually not a major challenge. Many CCDs have a dynamic range of about 60–70 dB, and you can maintain this dynamic range performance with 12-bit ADC resolution. Analog Devices and Texas Instruments (TI) sell highly integrated CCD signal processors that support 12-bit A/D conversion.
We have seen that the dynamic range is determined by the full well capacity and noise characteristics of the CCD. However, one thing to keep in mind is that dynamic range is not always the most critical performance metric. Dynamic range helps cameras capture high-contrast scenes, but sometimes we may be more interested in well capacity or low noise, rather than pursuing maximum dynamic range and balancing noise performance with full well capacity.