Detecting light#
Light detectors convert incoming photons into electrical signals that can be digitized to produce images.
It is important to understand few characteristics of the digitized signal.
background level
offset
dark counts
noise (multiple sources/types)
signal/noise (SNR) and signal/background (SBR) ratio
shot noise
electronic gain applied
Single-pixel detectors#
Laser-scanning microscopy (confocal, two-photon, near-field super-resolution and other types) relies on point detectors for high sensitivity and low-noise detection. Commonly we use PMT and GaAsP detectors and move the sample or position of illuminated point to achieve 2D imaging. PMT detectors work either in photon-counting mode or photon frequency mode, and your applications will require specific type. You can read more about PMT types from Hamamatsu
Camera-based sensors#
2D sensors combine multiple pixels to detect light coming from various points within field of view. Currently sCMOS technology is prevalent for scientific imaging.
When picking sCMOS detector we can choose from high density of pixels for high-resolution imaging or smaller number of larger, more sensitive pixels for low-signal applications. Speed of readout defines maximum frame rate and can be increased by selecting specific ROI (specifically number of lines of pixels).
Spectral detection#
Both single-pixel detectors and cameras can be used to acquire spectral information from the sample. To achieve fast, one-shot acquisition linear detectors are often used where spectral information is spread across multiple pixels. When using 2D sensors, data can be multiplexed to acquire spatial and spectral information at the same time.
Two-dimensional sensors with Bayer filter allow detection of spectral information and decoding it can be used to extract spectral information.
Signal averaging#
When signal is low we might want to perform averaging of measurement to improve SNR. It can be done in multiple ways:
temporal averaging: increase exposure (when using camera), average images together, repeat and average measurement from single-pixel detector
spatial (digital) averaging: bin pixels together (camera) or average pixels after acquisition
spatial (optical) averaging: reduce magnification factor of the system
Apart from decrease in spatial/temporal resolution, such averaging carries another drawbacks. For example, electronic binning will not decrease readout noise as binning happens after pixels were digitized. Longer exposure can lead to accumulation of thermal noise.
Power meters / light meters#
Power meters are devices that convert light hitting the sensor into electrical signal. Some power meters detect change in sensor temperature, others directly convert photons into electrons using photodiodes. Room (ambient) light and temperature fluctuations affect reading of power meters. Not all sensors are compatible with all light sources (e.g. pyroelectric sensors will only work with pulsed light sources)
It is important to measure light power to ensure that excitation light will not damage sensor devices (cameras) or biological tissues, and to report it as part of methods. Decrease in light power over time can mean degradation of light source or change in optical alignment. Power meters usually report “average power” while light-induced damage happens due to high peak power or high pulse energy. Here is nice documentation from Thorlabs
Power meters are not spectrometers. They are calibrated for a wide range of wavelengths but ultimately don’t “know” about wavelength of the incoming light — only about the photons hitting sensor. By changing wavelength in the power meter interface you cannot survey the spectral parameters of the incoming light, only change the scaling coefficients.
Luminocity meters (lux meters) measure amount of light falling on the detector but weighted by the wavelength according to human vision (luminocity function). These devices can be used to measure light power for monochromatic sources (laser) but conversion is non-trivial.