© Hamamatsu Photonics
Created by Talley Lambert
In this simulation, "pixel size" corresponds to the size of the height and width of the physical photodiodes on the camera chip in microns. Magnification from the objective lens is irrelevant, and the "ground truth" can be thought of as the magnified optical image at the camera chip. The most important observation here is that, for a constant sample intensity, the number of photons per pixel increases as the pixel size increases, increasing the signal to noise ratio (while decreasing spatial resolution).
The quantum efficiency of a camera depends on the wavelength of light being detected. In this simulation, changing the wavelength will set the quantum efficiency according to the characteristics of the currently chosen Hamamatsu camera.
Once a camera chip has converted incoming photons into photoelectrons, these electrons (the "charge packet") must be measured and digitized. This read out process is inherently error prone and introduces noise into the measurement referred to as read noise. There are several on-chip sources of noise that can affect an image captured with a CCD or sCMOS. Manufacturers usually combine these sources and express the noise as a number of electrons root mean square (RMS). Read noise is independent of exposure time and usually increases as the read speed of the camera increases (read speed is not represented in this simulation).
CCD chips have a single read amplifier, therefore all pixels have the same amount of read noise. sCMOS cameras, by contrast, have an amplifier on every photodiode. This results in non-homegenous read noise across the chip. In other words, some pixels are effectively "more precise" than other pixels. The average pixel read noise, however, is often much lower on an sCMOS like the Flash 4.0 than with a CCD, such as the ORCA-R2.
Quantum efficiency is the probability that a photon incident on the camera chip will be converted into an electron. While in reality, quantum efficiency is a wavelength-dependent property of a camera, for the sake of this simulation, QE is constant and represents a single (arbitrary) wavelength. Please note that for the camera presets, the QE value represents the maximum QE reported by that camera, and does not address the width of the QE peak.
Full well capacity refers to the maximum number of photoelectrons that any one pixel can accumulate over the course of a single exposure before becoming "saturated" (represented in red in the simulation). After a given pixel has accumulated the maximum number of photoelectrons, it can no longer respond to additional photons (in other words, increasing light levels in the sample will no longer result in increasing intensity values in the recorded image), and the quantitative capabilities of the camera have become compromised.
Offset is the value (in gray levels) "added" to intensity of every pixel in the image. An offset greater than zero is important in order to capture the noise inherent in the camera read process. With a very low offset, some pixels may end up with a gray value of zero as a result of read noise. This means that the pixel was "undersaturated" and information about the noise in the image has been lost. (Note however, that noise is still present in areas of the image with brighter signal... you are just unable to measure it without offset).
On a CCD, every pixel shares the same amplifier and therefore the same offset. In an sCMOS camera, because every pixel has an amplifier and every column of pixels has an amplifier, the offset is actually slightly variable across the chip.
Dark current results from thermally generated electrons that build up over time in all of the pixels on a CCD. As a result, the effect of dark current is most noticeable for long exposure times. Dark current is greatly reduced by cooling the chip. At the exposure times most commonly used in fluorescence microscopy (usully less than a second), the dark current present on most cooled CCDs and sCMOS is generally so low that it tends to be inconsequential. A notable exception is with EMCCDs, where any spontaneously liberated electrons may become amplified by the electron multiplication process. EMCCDs are often cooled to even lower temperatures for this reason.
Note: Like read noise, dark current is also non-homegenous across the chip at sCMOS cameras. But, this simulator assumes that dark current is homogeneous. In the future, non-homegenous dark current will be implemented.
When the electric field in the silicon is changed for charge transfer during the CCD readout, additional electrons can be liberated in a small proportion. They are called CIC. By careful manipulation of clock waveform amplitudes and edges, manufacturers can minimize CIC, which is normally estimated to produce only one electron per pixel in approximately 100 vertical transfers. Since the CIC is one electron, even in high-performance low readout noise CCDs the CIC is negligible. However, in the EMCCD at high gain settings, the CIC is generally treated as an additional component of dark-related signal.
Bit depth refers to the number of bits ("ones and zeros" in a computer) used to record each pixel's intensity value. A bit depth of "12" means that there are a total of 212 = 4096 intensity values between the dimmest possible pixel (intensity value of 0) and the brightest possible "saturated" pixel (intensity value of 4095).
Analog Gain (not to be confused with EM gain on an EMCCD) refers to the relationship between electrons going in to the read amplifier and the voltage coming out. Increasing gain results in greater voltages (and therefore intensity values) for a given number of photoelectrons. Every amplifier has an intrinsic gain, but camera companies often allow the user to further manipulate the gain. Note however, that while increasing the gain does increase the intensity value in the image that corresponds to a given light level, it also increases the noise in the image: so the signal to noise ratio is generally not improved. Furthermore, high gain levels can cause pixels to reach the maximum intensity value (saturation) well before the full well capacity has been reached, which effectively decreases the dynamic range. However, if light levels are low enough that saturation is not a risk, increasing analog gain is acceptable and may slightly increase SNR when limited by noise introduced by the analog-to-digital conversion process.
On a CCD, every pixel shares the same amplifier and therefore the same gain. In an sCMOS camera, because every pixel has an amplifier and every column of pixels has an amplifier, the gain is actually slightly variable across the chip. This "pixel response nonuniformity" is most easily seen in images with even illumination across broad regions of the image, such as images dominated by high background levels (or the "square" sample in the simulator).
Electron Multiplication (EM) gain can amplifies the signal electrons and increase electrons converted by “charge to voltage amplifier” called floating diffusion amplifier (FDA). From this process, the readout noise generated in FDA is reduced by the magnitude of EM gain respectively. On the other hand, the EM gain adds the signal electrons the noise component which is caused by the gain fluctuation called “Excess noise”. The EM gain increases SNR of the low light signal the readout noise is dominant, however it reduces SNR of the high light signal the readout is negligible.
This slider sets the intensity (brightness) of the sample in the simulation. The number corresponds to the flux of photons (photons/µ2/second) at the brightest discrete point in the sample. Note: if the camera pixel is larger than a single square micron - as is the case for most cameras - then it is likely that the the brightest pixel will be lower than would be predicted based on this intensity setting (because the object may not completely fill that pixel with even intensity). Use the numbers in the statistics tab to get more information about the number of photons/electrons in each pixel. Don't forget that the exposure time setting in the exposure tab will also affect the number of photons per pixel!
This slider sets the intensity of the background in the image. In this simluation, background is treated as an even flux of photons across the entire field of view.
This determines the duration of the exposure in the simulated image. It will affect the number of photons collected from both signal and background (and will also affect the number of electrons generated due to dark current).
Camera binning is a feature that grants flexibility in the effective pixel size of a CCD camera. Binning causes photoelectrons in neighboring pixels to be combined prior to the (noisy) readout event, thus increasing signal-to-noise ratio (SNR) at the expense of spatial sampling. For example, a binning of "2" would combine the electrons from 4 pixels (2x2).
Note: Binning only increases SNR on a CCD. On an sCMOS camera, because each pixel has its own read amplifier, binning is performed after amplification and therefore does not increase the SNR any more than could be acheived by simply averaging neighboring pixels through downsampling in post-processing. In this simulation, binning on an sCMOS simply sums the post-digitization intensity values of neighboring pixels, whereas binning on a CCD combines the photoelectrons from neighboring pixels prior to amplification and digitization.