Shane Elen
Jan 4th 2010 - in process
The wheels of progress turn slowly, thanks for your patience, I am getting closer. This little project is far overdue because of limited funds and limited used market supply of specialty equipment required to perform this testing. The equipment "want list" is basically fulfilled although I may still upgrade some of the items. There was a major setback with the UV-Vis-NIR source which seems to be almost resolved. I am currently working out the bugs and hope to have some relative spectral sensitivity curves in the near future.
After capturing my first digital ultraviolet and infrared images I was interested to determine how ultraviolet and infrared wavelengths were distributed into the R, G and B channels. I felt that understanding this might be helpful in some situations. Although some sensor QE curves could be obtained, they typically showed monochromatic response and not individual R, G and B responses from a Bayer matrix sensor.
Knowing the response of the individual R G and B colour channels to a specific wavelength and, the sensitivity relative to adjacent wavelengths measured in exposure values (EV), may be of value to the photographer. The response of each colour channel to a specific wavelength can indicate how the DLSR sensor interprets UV and IR wavelengths, which may be of help during image capture or post processing. The relative sensitivity of adjacent wavelengths in EV can indicate the practical cut-off wavelength for UV sensitivity of the DSLR under ideal conditions.
My earlier monochromatic testing on the Nikon D70 and Fuji S3Pro (non-UVIR was a crude attempt to determine how ultraviolet and infrared wavelengths were distributed into the R, G and B channels. Although I achieved this, the testing was performed without consideration of optical power, hence there is no correlation of relative spectral sensitivity from wavelength to wavelength
The equipment required to measure relative spectral sensitivity is basically the same as that required for measuring a QE curve however the difference is primarily in the degree of complexity of data collection and reduction which is significantly greater for the QE method.
The efficiency of a sensor for detecting light as a function of wavelength can be presented as a quantum efficiency (QE) curve or a spectral responsivity curve. They are related and in effect similar measurements.
The QE of a sensor is defined as the ratio of the number of photogenerated electrons produced by a pixel to the number of photons incident on the pixel. When plotted as a function of wavelength, a QE curve can be generated in which the y-axis represents the efficiency (%) of the photon conversion process and the x-axis represents the wavelength. Therefore QE describes the precise response of the sensor to different wavelengths of light.
Spectral responsivity of a sensor is the relative efficiency of light detection as a function of wavelength and is determined by measurement of the current produced by the sensor for a given power and wavelength of incident light. When plotted as a function of wavelength, a spectral responsivity curve can be generated in which the y-axis represents the current produced per watt of incident light (A/W) and the x-axis represents the wavelength.
As mentioned earlier, the equipment required to determine response of the RGB channels and relative spectral sensitivity of the DSLR CCD are basically the same as that required for measuring QE and Spectral Responsivity curves.
There are two basic approaches to measuring QE, the diode method and the integration method.
The diode method is generally the preferred method as it is faster and more accurate. This method treats the CCD as a photodiode and requires direct access to the CCD connector pins in order to measure induced photocurrent. The QE is obtained by simple ratioing of the CCD photocurrent and the photocurrent of a calibrated reference photodiode. Obviously this method is not practical unless the DSLR is completely disassembled in order to reach the CCD pins.
The integration method is much slower and less accurate but is useful for CCDs that cannot be tested in diode mode i.e. the DSLR. In this method the CCD is presented a timed exposure to a calibrated light source and an image is recorded. QE is obtained by determining the CCD signal in electrons per second from the digital image and ratioing it to the light intensity in photons per second. Precision timing is required in order to accomplish this.
I propose to use a variation on the integration method by utilizing the shutter timer on the camera for a timed exposure. While this introduces an unacceptable timing error for scientific applications, it should still be sufficient for use by a photographer.
The system comprises a broadband UV-VIS-IR light source, a monochromator, 2nd order sorting and neutral density filters, an integrating sphere, a beam splitter, a calibrated UV-VIS-IR reference photodiode, and an optical power meter. This system will be used to deliver small stepped increments of order sorted monochromatic light with known optical power to the CCD of the DSLR. This is the same approach that I had used in earlier testing of the Nikon D70 and Fuji S3 Pro but with the addition of optical power measurements and timed exposures. The exact procedure will be detailed when the first RGB sensitivity curve is posted.
Copyright (c) Shane Elen 2006. Last updated Jan 4th, 2010.