<center>![ccd.png](https://cdn.steemitimages.com/DQmU2AWJkPEcsSEsTMSzCMkPv6Asd6XHQqTU8Zo4WQRawRh/ccd.png)</center>
<center>*Figure 1. CCD bucket analogy.  Credit:  Howell 2006*</center>

Each pixel of a CCD sensor is like a bucket, but instead of collecting rain water, these buckets collect electrons that are generated by the incoming photons. &nbsp;Just like a bucket can only hold a finite volume of water, a pixel can only collect a finite number of electrons before it "overflows." &nbsp;When the bucket overflows, the measurement of &nbsp;the electrons is no longer linear. &nbsp;What does that mean? &nbsp;Consider in the water example, a certain number of rain drops will produce a certain change in volume of the amount of water in the bucket. &nbsp; This relationship holds up to the point where the bucket overflows and the incoming rain drops cannot be measured because all of the space has been taken up. &nbsp;The same is true for pixels in the CCD sensor: &nbsp;There is a linear response between the amount of incoming photons and the measured number of electrons per pixel up to a certain point where the relationship breaks down and the pixel "overflows" or "blooms" (see Figure 2 below).

<center>![bloom.png](https://cdn.steemitimages.com/DQmdniwGm8vPapn8wnWvRSRHCSWwmUJBTLkhtyNocHB2vfE/bloom.png)</center>
<center>*Figure 2. Example of blooming. Credit: [AOweb](http://astronomyonline.org/astrophotography/ccd.asp)*</center>

Some types of CCD sensors have what is called an Anti-Blooming Gate (ABG) to try to minimize the effects of blooming. &nbsp;However, since photometry is concerned with precise counting of the number of photons, it is better, but not absolutely necessary, to have a Non Anti Blooming Gate (NABG) to get an unadulterated measurement. &nbsp;The manufacturer's CCD specifications will include the Full Well Depth of the sensor, or the <i>expected</i>&nbsp;number of electrons that can be collected before the pixels bloom. &nbsp;In either case (ABG or NABG), it is best practice to actually test the camera to find out where the pixels saturate and if there is any non-linear behavior before the point of saturation. &nbsp;This will enable better photometry since the "overflow" point is well established and can be avoided.

<center>![linearity.png](https://cdn.steemitimages.com/DQmNXskNYFxWmRPAZwGDrUHRWUvsgcpTECkSU2N22tu3iyB/linearity.png)</center>
<center>*Figure 3. Results of linearity testing. The sensor displays linear behavior for roughly a minute before saturating. Credit: Author*</center>

The sensor clearly saturates at just over 60,000 ADU (Analog to digital units). &nbsp;However, after fitting a straight line to the data, there also appears to be some non-linear behavior before the point of saturation. &nbsp;It is not entirely clear from this data exactly where the linearity ends, but a reasonable estimate is about 50,000 ADU. &nbsp;This is a "deeper" full well depth than reported by the CCD manufacturer (SBIG). &nbsp;However, it is probably best that their estimate is conservative. &nbsp;In fact, it is smart to use the more conservative number to calculate exposure times for the targets of interest. &nbsp;For example, if the goal was to expose a target until it reaches half of its full well depth (a typical practice in photometry), it would be wise to use 40,000/2 = 20,000 ADU. &nbsp;When in the business of counting photons, it is always safer to err on the side of under-exposure in order to maintain the all important linear relationship between the incoming light and measured electrons.

**References**
1. AOweb: &nbsp;<a href="http://astronomyonline.org/astrophotography/ccd.asp">http://astronomyonline.org/astrophotography/ccd.asp</a>
2. Howell, S. B. 2006, Handbook of CCD Astronomy, Cambridge University Press, UK