Teledyne Photometrics KINETIX, Руководство пользователя, Страница: 23 / 48

14
58-723-004 Rev A01
Digital Binning
CCD image sensors are capable of charge binning (i.e., combining adjacent pixels into one
super pixel). This is accomplished as part of the charge transfer process and has the
advantage of increasing signal to noise in read-noise limited situations, at the expense of
spatial resolution.
The lack of a charge transfer process in CMOS devices means true charge binning is not
available in current scientific CMOS sensors. Even so, co-adding pixels is a convenient
means to reduce image data, or increase signal by 4x and improve SNR by 2x as the noise
from each pixel adds in quadrature.
The Kinetix system includes symmetric and asymmetric on-camera digital binning of up to 4
x 4 pixels.
This mathematically combines signal from adjacent pixels and adjusts the sum so
that the bias offset is only added one time.
Sensor Clearing
In order to capture the highest signal-to-noise ratio possible, it is important that scientific
cameras minimize any signal that’s not derived from the sample. One contribution to this
background signal is the buildup of charge prior to an exposure, which includes any light still
reaching the sensor as well as thermally generated sensor dark current. To eliminate this
pre-acquisition charge accumulation, most CCD and CMOS cameras clear the sensor one or
more times prior to exposing the sensor to light. This can be done using a “fast” readout that
is subsequently discarded.
Unlike CCDs, there is limited benefit to performing multiple pre-exposure clearing cycles
with CMOS, because each pixel is reset as part of the normal readout process, and the
charge transfer registers that can hold residual signals are not present.
Bias Offset
Settings
Scientific cameras produce a fixed artificial signal offset known as bias offset. This grayscale
offset is present in all pixel values, even when no light is falling on the sensor and the
exposure time is set to zero. Bias offset prevents clipping of the camera read noise and
preserves quantitation even down to signals of a few electrons per pixel.
The factory default bias level is approximately 100 ADU in Dynamic Range Mode and
Sensitivity Mode. The default bias level is approximately 20 ADU in Speed Mode and
approximately 1000 ADU in Sub-Electron Mode.
If supported by the software application being used, the bias level can be altered. However,
we do not recommend changing this value because the preset values for the Despeckle Pixel
Noise Filters are set with this value.