Six theses on sensitivity.
The modern video surveillance market, designed for a wide range of consumers, has been formed over the past 10–15 years. Moreover, this happened during the period of active destruction of state enterprises and the abolition of control institutions.
In the Soviet Union, any product was designed at state enterprises, it had strictly verified technical documentation, inspection authorities strictly monitored the compliance of the supplied products with its requirements. Imports were carried out according to the same principle, i.e., each product imported from abroad met the approved technical requirements. In addition, it should be noted that any technical devices were manufactured on the basis of or taking into account the level of knowledge created by academic science in academic institutes. Therefore, the technical documentation for the products, of course, corresponded to the existing level of technology and almost completely corresponded to its real characteristics.
Unfortunately, the modern video surveillance market very often sins by indicating rather vaguely interpreted parameters in technical documentation; in advertising articles, some authors try to characterize certain products using some scientific terms, describing some technologies. Upon detailed discussion, it turns out that they do not know what these technologies include, but simply repeat opinions heard somewhere, valid for products of a different class, manufactured using a different version of this technology.
As a result, an installer who needs to build a video surveillance system with certain parameters based on equipment offered by the modern video surveillance market should trust only his own experience and create video surveillance systems based on equipment that he knows well.
In order to try to make the installer's task a little easier, we will try to analyze the parameters of video cameras, as well as pay attention to some subtleties that may help in assessing the real parameters of this or that equipment.
Video camera sensitivity
Photosensitive elements are characterized by such parameters as integral current sensitivity, integral volt sensitivity, monochromatic sensitivity.
Current sensitivity Si determines the value of photocurrent created by a single radiation flux and is measured A/lux or A/Watt depending on the units in which light is measured: in light (lux) or energy (watts). It is necessary to draw the reader's attention to the fact that 1 lux is the illumination of a surface of 1 sq. m with a luminous flux falling on it equal to 1 lm.
Volt sensitivity Su characterizes the signal value in volts, related to the unit of incident radiation. Current and volt sensitivity is called integral if it characterizes the sensitivity to the integral flux of the light source, and monochromatic if it characterizes the sensitivity to monochromatic radiation.
The methods for determining these parameters have been clearly and unambiguously defined by GOSTs and standards for many years and do not cause any doubts or discussions. For example, integral sensitivity is determined by measuring the current or voltage coming from a photodetector connected according to a certain electrical circuit. In this case, the photodetector is illuminated from a type A source with a luminous flux of about 1000 lumens or 2000 lumens, located at a distance of about 30 cm from the photodetector in a special installation that provides a certain solid angle. This source imitates sunlight. Note that a conventional 100 W incandescent lamp at 220 V creates a luminous flux of about 1300 lm.
The integral current sensitivity is calculated using the formula:
Si = Iф/(E x A) , where
Iф is the photocurrent,
E is the luminous flux incident on the photodetector,
A is the area of the photodetector.
Let me repeat that sensitivity has the dimension of A/lx. The luminous flux in this case falls, of course, on the photodetector itself, and not on the lens. At the same time, as follows from the formula, the higher the photocurrent, the higher the sensitivity, the better and higher quality the photodetector.
There are two more parameters that are important for security television equipment.
Detectivity D*, cm x Hz ½ W -1 characterizes the possibility of using the photodetector to detect extremely small optical signals and is determined using the formula:
D* = Uф x (A x f ) 1/2/Uш x Ф, where
Uф is the photosignal voltage,
Uш is the photodetector noise voltage,
f is the frequency band,
A is the area of the photodetector.
The threshold sensitivity of the photodetector Ф th. determines the power level of the light flux at which the signal is equal to the noise. The threshold sensitivity and detectability are related to each other by the ratio:
Ф th. = А1/2/D*
However, in the Russian market, practically all descriptions of video surveillance cameras include such a parameter as sensitivity. However, this parameter is not specified by any of the current GOSTs or any standard. This allows equipment manufacturers to interpret this very sensitivity as they please.
Firstly, very often this parameter is measured in lux [lx] = [lm] x [area unit], i.e. illumination is given, and the term “minimum illumination” is often found in descriptions. Sensitivity, as noted above, has a different dimension. And this is not just nitpicking.
Secondly, it is most likely assumed that at such illumination some signal appears at the output. The level of this signal also remains, apparently, somewhere in the depths of the manufacturer's documentation, usually inaccessible to the installer. The assumption that we are talking about the threshold sensitivity at which the signal is equal to noise leads to the sad conclusion about the absence of a useful signal at this illumination and the inoperability of the system.
It should be noted that in some descriptions sensitivity is interpreted as illumination in lux, at which the signal level is, for example, 50 IRE. This is also, of course, not sensitivity, but still some more specific information. I would like to repeat that I draw the reader's attention to the term «sensitivity» not for the sake of criticism, but because it is really important. We will dwell on this in more detail below. For clarification, I will note that in analog video paths, the reference electrical level corresponding to white is determined by the value of 100 IRE (a unit that got its name from the Institute of Radio Engineers — a professional organization created in 1912 in New York). But the value of the reference black level in different countries and TV systems may differ. In the European PAL system and the Japanese version of NTSC, black is always defined as 0 IRE, in the USA and Canada — 7.5 IRE.Thirdly, the sensitivity specified in the manufacturers' descriptions was apparently measured for an optical-electronic system: a photosensitive element with a lens. Therefore, the sensitivity specified in the descriptions is only valid for a camera with a lens of the specified aperture ratio. With lenses of a different aperture ratio, it will be different. Considering that CCTV cameras are supplied mainly with removable lenses, and sensitivity values are specified by different manufacturers for lenses with different photographic powers, the installer must have significant experience to draw any conclusions about the value specified by the manufacturer versus the term «sensitivity». Thus, if the descriptions provide a value for the minimum illumination recorded by the camera with two different lenses with aperture ratios of F1.2 and F1.4, it must be taken into account that the ratio of the relative aperture values must be squared to obtain the corresponding attenuation of the light flux. Another thing is that:
the light transmittance coefficient of the lens is also important;
the actual value of the relative aperture may differ from the declared one as F1.2 from F1.4, unfortunately, more often in the smaller direction.
Fourthly, sensitivity is sometimes measured with different exposure times, i.e. with the shutter open from fractions of a second to several seconds. Some manufacturers increase sensitivity by about 10 times by increasing the exposure time from 1/25 s to 1/8 s. In this case, the minimum illumination recorded by the camera naturally increases with increasing exposure time. However, the image received from the camera may be blurred, especially if the camera is aimed at fast-moving objects. In this regard, the following should be explained.
To take into account the inertial properties of photodetectors in classical photoelectronics, such characteristics as pulse integral sensitivity, maximum operating frequency (light modulation frequency at which the photoresponse amplitude decreases to 0.7 of the maximum), photosignal time constant (determined by the photosignal pulse rise time to 0.63 of the maximum for a rectangular light pulse), and phase shift between the input light and output electrical signals are used.
If a rectangular light pulse falls on the photodetector (Fig. 1a), then, due to the inherent inertial properties, the current at the photodetector output will appear and disappear after some time д = W2/2D p, where:
W is the thickness of the photodetector base,
D p is the diffusion coefficient.
In this case, the front and the tail of the photocurrent pulse will be blurred due to both the recombination of nonequilibrium charge carriers created by light and the fluctuations in the thermal velocities of the charge carriers (Fig. 1b). If the duration of the intervals between light pulses is much greater than ≈ d, the photocurrent will look like pulses separated from each other. As the frequency of light pulses increases, the duration of the intervals between them decreases, and at high frequencies, the next level of photocurrent begins when the previous one has not yet ended (Fig. 1c). As a result, at a high frequency of light pulses, the photodetector does not have time to respond to each light pulse. Therefore, the photocurrent pulses merge with each other and become constant. A video surveillance camera operating in this mode will not produce a clear image. The limiting frequency is the frequency at which I f = 0.7 I f max.Fifthly, as mentioned above, according to GOSTs, sensitivity in its classical sense is measured under irradiation from a standard light source, usually a type A source. Some video camera suppliers proudly say that the sensitivity values given in their descriptions were measured under exposure to near IR radiation in addition to visible spectrum radiation.
This is motivated by the fact that all natural objects emit IR rays. At the same time, it is often forgotten that natural objects are characterized by long-wave IR radiation, to which the photodetectors under study are not sensitive.
Sixthly, as a rule, it is not specified at what distance from the photodetector the illumination measurements were made, at which an acceptable signal was received from the video camera. It is clear that this can be either on the photodetector itself or at a distance of several meters from it.
All of the above aspects explain the importance of using the term sensitivity in its classical sense, otherwise it is very difficult to understand what is meant by the given parameters. But there is another important argument.
Let us return to some of the basics of semiconductor physics, which determine the parameters of cameras.
Let us remember that as a result of the action of light in a semiconductor, electron-hole pairs are formed, which cause the generation of an electric current, i.e. a non-electrical signal is converted into an electrical signal.
Dependence of photoconductivity on the luminous flux value [lm]. This dependence is linear at low light intensities, then the characteristic enters the saturation region, i.e. as the light intensity increases, the photoconductivity remains constant. Taking this into account, the actual photocurrent at low luminous fluxes can be orders of magnitude lower than the value approximated from the photocurrent value measured under illumination by a standard source, due to the nonlinearity of the photocurrent dependence on the luminous flux value. Therefore, it is important to draw the attention of installers once again to the need to clarify with the camera manufacturer at what actual illumination values the sensitivity was measured, whether the specified value is the result of theoretical approximations that do not take into account the described physical processes.
The useful signal or photocurrent is equal to the total current minus the dark current caused by noise of various natures (thermal, generation-recombination, shot) and minus the current caused by background radiation.
Obviously, in order to improve the sensitivity of photodiodes, it is necessary to reduce the dark and background currents and increase the photocurrent.
In the literature on video surveillance, there is such a parameter (apparently, this has developed historically) as the signal-to-noise ratio. Although the developers of photoelectronic matrices most often use such a parameter as dark current.
At high illumination levels, when the useful signal for silicon photodiodes (namely, silicon photodiodes convert visible light into an electrical signal) is microamperes, and the dark current is nanoamperes, there are no problems with the conversion of the optical signal. As the illumination decreases, the photocurrent value drops. The dark current in such cases remains the same. Obviously, in the case when the photocurrent value is commensurate with the value of the dark and background currents, there can be no talk of any clarity, contrast and stability of the image. In this case, it is necessary to take into account the nonlinear dependence of the photocurrent on the luminous flux.
As a result, the useful signal may be completely lost. That is, in a situation where the photocurrent decreases with decreasing illumination, and noise signals retain a constant value, the requirements for the quality of the photodetector increase, which, of course, leads to its rise in price. There are photodetectors with a low level of dark currents (on the order of picoamperes, etc.). Attempts to solve this problem other than by choosing a better quality, differently designed photodetector are self-deception.
In this regard, it is also necessary to note the following point, which is sometimes not quite correctly interpreted by installers. When the area of each individual photosensitive element (pixel) increases, the overall signal does increase somewhat. However, the sensitivity does not increase, since when calculating the sensitivity, the area value is located in the denominator of the formula. In addition, with an increase in the area of the photosensitive element, the value of dark currents increases. So, increasing the area of the photosensitive pixel is not a solution to the problem, if not to say the opposite.
There is also such a parameter as the size of the matrix of the photosensitive element. Historically, for the first television cameras, the actual matrix size was not specified, but the diameter of the exit pupil of the lens. Therefore, usually, not the physical size of the matrix is specified, but the diagonal size in inches. 1 inch is 2.54 cm. The size can be 1/2 ”1/3” 1/4 ”. The more pixels are located on the matrix, the higher the resolution. Therefore, there is an opinion that the larger the matrix size, the better. However, creating identical pixels over the entire area of the matrix is a very difficult task; there will never be an ideal match. The larger the matrix size, the more difficult it is to achieve identical pixels. Elements located in the center of the matrix will differ from elements located at the edges of the matrix.
Now we are talking about the production of matrices with pixels of ever smaller sizes.
In this regard, it should be noted that in order to create smaller pixels, it is necessary to work using technological photolithographic processes with a higher degree of resolution. Sometimes it is simply a different level of equipment, differing in price by orders of magnitude. At the same time, the equipment itself and the organization of the technological process at nm cost millions of dollars. Accordingly, the transition from one technology to another leads not only to an improvement in quality, but also to a significant increase in cost.