Range of Thermal Cameras - Current Market Overview from a Professional Perspective

Sensor types, frame rates, detector readout, detector cooling

Thermographic devices suitable for contactless temperature measurement (infrared cameras with thermographic capabilities) have undergone rapid development in recent years. Considering that these devices appeared just 50 years ago, they have now grown into one of the most well-known and versatile inspection tools. Therefore, it is not the lack of suitable types that poses a problem for a customer planning to purchase a thermal camera, but rather the overwhelming variety in the market (manufacturers, types). It is time to review the development and types of these instruments from a professional perspective and organize the current offerings based on some important technical parameters. The measurement technology implemented in the cameras and the available accessories determine the device's application area, as well as the expected measurement accuracy and the achievable thermal image quality.

Scanning Thermal Cameras - the "vanished" peak technology of the beginnings

The very first commercially available thermal cameras suitable for temperature measurement were primarily produced in scanning (point-scanning) design. These cameras use only a single-element ("point") detector to convert the infrared radiation and scan the object to be measured with a mechanical (mirror or lens) system. Since this imaging principle requires a high-speed (photon) detector and high-precision mechanics, its production is quite expensive, requires cooling, and due to the mechanical components, has a limited lifespan. However, it has a significant advantage over all other methods: each signal corresponding to each image point is mapped by the same detector. Thus, data from every point in the thermal image are created under perfectly uniform conditions, resulting in very good image homogeneity (and up to 10mK thermal resolution). The slowness of image acquisition (typically only one image per second) and the other disadvantages listed earlier have led to the fact that this thermal camera technology is now only available as used equipment at best.

Figure: schematic structure of scanning thermal cameras [source: Infratec] (1 detector, 2+5 lenses, 3 horizontal deflecting mirror, 4 vertical deflecting mirror, 6 object, 7 measurement surface)

Matrix Detector Thermal Cameras - the "common" structure of current thermal cameras

In matrix detector thermal cameras, thousands of individual sensors are arranged in a matrix-like manner to detect the thermal radiation to be measured "simultaneously," eliminating the need for a mechanical deflecting unit. This makes the camera mechanically simpler, smaller, lighter (and cheaper). Although the optical path seems surprisingly simple, the devil is in the details: one main problem is that each individual sensor converts each pixel of the thermal image, and while its characteristics may be very similar to its neighbor, they are still measurably different. Compensating for this lack of uniformity requires a significant amount of real-time image processing, yet the image homogeneity achieved by scanning systems is still not attainable. However, with modern matrix detector thermal cameras - depending on the sensor technology used - achieving a thermal resolution of 30mK (or even 20mK) is now possible, which is sufficient for most applications, leading to the discontinuation of scanning thermal camera production.

Figure: schematic structure of matrix detector thermal cameras [source: Infratec] (1 detector, 2 lens, 3 object)

Sensors of Modern Matrix Detector Thermal Cameras

Basically, two fundamental types - thermal sensors and photon detectors - are distinguished. Thermal types are based on the principle that they heat up due to infrared radiation (energy of electromagnetic waves), causing a change in one of their physical (electrical) parameters, from which the necessary electrical signal can be extracted. Photon detectors, on the other hand, provide an electrical signal proportional to the number of photons, but they require deep cooling (-150°C ... - 200°C) to operate. (Without cooling, disordered electron movement would hinder the occurrence of the exploitable physical effect.) Basic sensor technologies

• Thermal detectors (e.g., microbolometers)
  • do not require cooling (allowing for lightweight, low-energy consumption thermal cameras)
  • suitable for mass production, hence cost-effective
  • lower thermal sensitivity, resolution, and accuracy

Figure: operation of thermal detectors [source: PIM]

Figure: schematic structure of microbolometer [source: Infratec]

• Photon detectors (quantum detectors)
  • require cooling
  • high thermal sensitivity and resolution
  • high stability and accuracy

Figure: Structure and operation of photon detectors [source: PIM]

Each sensor technology has detectors for various wavelength ranges, depending on the material used. However, bolometers / microbolometers can only be made for the long wavelength range due to their weak thermal sensitivity. (Sufficiently high radiation intensity can only be expected in this range.) The following figure provides an overview of the technical possibilities.

Figure: Infrared sensor wavelength ranges according to the materials of the detectors [source: Infratec]

It is important to know that the sensor's wavelength range (spectral sensitivity) significantly influences the application areas of thermal cameras. (As a reminder: thermal cameras in different - limited - wavelength ranges are necessary due to the atmospheric transmission properties. Short, medium, and long wavelength thermal cameras are made for the so-called atmospheric windows.) While low-temperature objects (e.g., -80°C) cannot be measured with medium-wave 3 ... 5 µm thermal cameras, it is impossible to detect the thermal radiation of objects behind glass with long-wave 7.5 ... 14 µm thermal cameras.

There are additional application-related limitations regarding large (several hundred meters) measurement distances: these can only be achieved with long-wave thermal cameras. On the other hand, the detection of flame temperatures in combustion processes is mostly possible only with medium-wave thermal cameras, but the reverse task - detecting object temperatures through flames without sensing the flame temperature - can be achieved with long-wave thermal cameras. For many applications (such as sensing the temperature of thin foils, detecting gas leaks, measurements through special measurement windows (e.g., vacuum chamber windows, furnace measurement windows), suitable wavelength range thermal cameras and appropriate infrared filters must be selected based on their materials. This task requires special knowledge and experience, and to avoid costly mistakes, it is advisable to entrust it to a professional.

Thermal Camera Frame Rate (Image Refresh Rate)

Thermal cameras with matrix sensors based on microbolometers are available with image refresh rates of 9, 15, 30, 50, 60, 120 Hz, or even 240 Hz - whether they are installed or portable (mobile) thermal cameras. Significantly higher - 850, and even 6000 or 9000 Hz - image refresh rates can be achieved with photon detector thermal cameras. The required image refresh rate depends on the time constant of the temperature changes of the object to be measured, the speed of motion, or even the camera's movement speed. The time constant of the temperature changes of the object (rate of change of temperature) - or more scientifically, the frequency of the temperature process - imposes serious requirements due to the operating principle of thermal cameras: thermal cameras (like all digital signal processing measuring systems) must also comply with the basic sampling rule - the Shannon theorem. The Shannon theorem requires that the sampling rate for digitizing the highest frequency component of the process being measured should be at least twice its frequency. If this rule is not followed, undersampling occurs, which would, for example, lead to the apparent slow progression of the recorded temperature process in the case of a periodic temperature fluctuation, as shown in the following figure. This would lead to completely erroneous conclusions in many cases!

Figure: Consequence of violating the Shannon theorem resulting in undersampling error [source: PIM]

Based on the above, the frame rate of thermal cameras is critical for any task involving the examination of temperature changes. If the recorded change has a period of 1/10 second, a minimum of 20 Hz (preferably 25 Hz) frame rate is required. In the case of power electronics devices, temperature rises with frequencies as high as 300 Hz are not uncommon, requiring a frame rate above 600 Hz for recording (which can only be achieved with photon detector thermal cameras)! Further examples of the need for exceptionally fast photon detector thermal cameras include detecting tool and workpiece heating in machining technologies, observing the surface temperatures of car airbags, researching the temperatures of pyrotechnic processes, or investigating sudden mechanical impacts...

The previous list could go on for a long time, but this should not lead to the mistaken conclusion that in the case of slow (or even steady-state) thermal processes, the frame rate of the thermal camera cannot be a critical parameter for the feasibility of the measurement. Because in the case of moving objects or a moving thermal camera, it is equally important for the thermal camera to be fast enough. For microbolometer thermal cameras, the integration time that determines their frame rate limits the speed at which moving objects can still be correctly detected. The maximum speed of motion is reached when during the integration time, the surface of the object detected by an individual detector extends so much in the direction of motion that this detection surface "runs off" the object surface during the integration time.

Sample calculation: If we want to detect a 15 mm wide object with a thermal camera with a 30 Hz frame rate (typically with an integration time of about 25 ms), 2 mrad geometric resolution from a distance of 1 m, then the maximum speed between the thermal camera and the object (parallel to the object surface) can be calculated as follows: 2 mm + 25 ms * x m/s < 15 mm, where x is the maximum speed. Therefore, based on the above equation, the maximum speed is 0.52 m/s, which is only 1.87 km/h.

Figure: Blurring of thermal image due to fast object motion - enlarged view of running legs [source: PIM] (slow-moving body + left leg on the ground --> sharp, hands and right leg in fast motion --> blurred)

Serious problems arise even when trying to capture detailed thermal images or perform long-distance measurements with a handheld thermal camera. It is a known fact in photography that a practiced - steady-handed - photographer can take motionless photos even at a 1/60 shutter speed (without a tripod), while an "amateur" with shaky hands may occasionally result in blurred images at a 1/125 shutter speed. These shutter speeds correspond to 17 ms and 8 ms exposure times. What skill is required to capture motionless thermal images handheld with a 30 Hz or even just 15 or 9 Hz thermal camera! To achieve this, the thermal camera would need to be held motionless for up to 30 ... 40 ms, which is practically impossible. In other words, handheld thermal cameras capable of capturing motionless images must have an integration time shorter than 15 ms. This is generally provided only by thermal cameras with a 50 Hz or faster frame rate; slower thermal cameras are unsuitable for handheld shots.

Figure: Blurring of thermal image due to camera movement (e.g., hand tremor) [source: PIM]

Detector Readout Methods

For moving or rotating objects or a thermal camera in motion relative to the object, the metrological applicability of thermal cameras depends not only on the discussed frame refresh rate but also on the method of pixel data readout. Two common methods are implemented: line-by-line readout (applicable for both thermal and photon detectors) and the so-called "Snap-Shot" readout. The latter is exclusive to certain photon detectors due to their special properties, as the slowness of thermal detectors (e.g., microbolometers) with integration times of up to 6 ... 20 ms renders this technology completely meaningless. Line-by-line readout: Taking an average 320x240 pixel matrix sensor as a basis, this represents 78,600 individual sensors. It is obvious that it is not practical to use the same number of samplers and AD converters for digitalizing the analog electrical output signal of each pixel (due to space and energy requirements and costs). Therefore, only a circuit with 240 samplers - AD converters corresponding to a single line is used to "read out" the 320 rows of the sensor successively (stepping through one by one). First, we zero the "signals" of the sensors in the first row and start their measurement (integration) time, then shortly after, we do the same with the second, third, and subsequent rows. Meanwhile, the integration time of the sensors in the first row elapses, so we can read out their measurement data. Then, individually moving on, we do the same with the rest of the rows until we reach the last one. In the meantime, the restart of the integration time of the first rows for the next "readout" cycle has already taken place. This process can practically be described as if the sensors are continuously integrating, and we interrupt this with a readout and zeroing out line by line.

Figure: Timing diagram for line-by-line readout [source: PIM]

The consequence of line-by-line readout is that the representation of moving objects is distorted, as illustrated in the following figure. (The faster the motion, the greater the degree of distortion.) The reason for this is that the measurement data per line were not generated at the same time but only successively - similar to a multiplexed multi-channel metrological measurement system.

Figure: Distortion of moving objects representation due to line-by-line readout effect [source: PIM]

"Snap-Shot" Technology

The issue related to detecting moving or rotating objects can be addressed with "Snap-Shot" technology. However, this method is meaningful only with sufficiently fast (even as short as 10 µs integration time) photon detectors. In contrast, thermal detectors (e.g., microbolometers) that are orders of magnitude slower would blur the representation of moving objects anyway (due to the long integration time). Photon detectors equipped with "Snap-Shot" capability perform measurements (signal integration) simultaneously for each pixel, and then the measured values for the pixels are "frozen" at the same time. Subsequently, as with line-by-line readout, the values are read out and A/D-converted line by line. Therefore, instead of using thousands of readout-digitalization circuits, only as many as needed for reading out each row are employed. Despite this, the representation of moving objects is not distorted because the signal measured by each individual sensor originates from the same time (moment). From a metrological perspective, this is a system of simultaneous sampling.

Figure: Timing diagram for "Snap-Shot" technology (gray = frozen value) [source: PIM]

With the most advanced thermal cameras featuring "Snap-Shot" technology, up to 450 pieces of 320x256 pixel thermal images can currently be recorded per second. However, the time required for line-by-line readout recognizable in the timing diagram can be comparable to the integration time of the photon detectors, and may even exceed it. The maximum image readout frequency is therefore mostly limited by the readout. To overcome this limitation and achieve even faster image acquisition, the processing of so-called SubFrame partial images can be applied, which unfortunately results in displaying fewer details due to fewer pixels. Special thermal cameras equipped with this image processing technology are capable of capturing up to 4500 images per second at a resolution of 160x128 pixels. Note: the detector still performs signal integration and value freezing simultaneously on all pixels. Simply put, readout and digitization are limited to the selected area.

Figure: Examples of SubFrame solutions in "Snap-Shot" technology [source: PIM]

Figure: Fan thermal image - left: with microbolometer serial readout, right: with photon detector "Snap-Shot" procedure [source: InfraTec]

Cooling Technologies for Photon Detectors

Currently, a wide variety of uncooled thermal detectors (e.g., microbolometers) are available. However, the most accurate and fastest measurement capabilities, as well as short- and mid-wave thermal cameras, can only be made with photon detectors, which exclusively require cooling. Instead of liquid nitrogen solutions, highly reliable miniature cooling compressors (Stirling coolers) have become more common for ensuring their cooling. For some detector types, an additional option is the use of thermoelectric (Peltier) cooling, although this does not achieve such low temperatures (thus narrowing the range of detector designs and materials.) Stirling Cooling Stirling cooling is based on the CARNOT thermal cycle, where a gas (helium) is compressed (causing the gas to heat up), then cools down by releasing heat to the surroundings. During the subsequent expansion (in another cylinder), the gas cools to a very low temperature, enabling it to absorb heat energy from the surroundings (in our case, the detector). This whole process occurs as a closed cycle. The two-piston micro-compressor used for this purpose in thermal cameras allows for their application in any position, ensuring measurement reliability and accuracy over a wide operating temperature range (with good efficiency). However, a drawback is that these cooling compressors have a significant size and weight, making it impossible to create lightweight and compact thermal cameras using this technology. An even bigger issue (especially for continuous applications) is that Stirling coolers are a mechanical system with a limited lifespan. For the most modern devices, this limit can reach up to 8000 - recently 12000 - operating hours (maintenance-free!).

Figure: Stirling cooling principle and an actual Stirling cooler [source: InfraTec]

Peltier Element Cooling

Peltier element cooling (also known as thermoelectric cooling) is usually implemented in the form of a 3-stage Peltier element cascade to achieve the deep temperatures required. Its advantage over Stirling cooling is that it has no mechanical (moving and thus wearing) parts, resulting in practically no limit to its lifespan. However, it consumes more energy and can only achieve moderately low temperatures (approximately -150 °C) at the cost of higher energy consumption, which may not be sufficient for the operation of all types of photon detectors.

Figure: 3-stage Peltier cooling principle / MCT-Sprite photon detector with Peltier cooling [source: InfraTec]

Rahne Eric (PIM Ltd.) pim-ltd.com, engineeringexpert.com  

Contact Us

The content of this publication is protected by copyright. Any (even partial) use, electronic or printed re-publication, is only permitted with the indication of the source and the author's name, and with the author's prior written permission. Violation of copyright (Copyright) will result in legal consequences.