Thermal image sensor with integrated lens
Nowadays image sensors are important appliances for supervision and control of various industrial processes.
Circuit boards, for example, have in the past been examined under a magnifying glass to expose soldering defects. Today special cameras are invariably used. Cameras in a visible field are – in some cases – not the ideal solution for inspection. Specific sensors enable contact-free temperature monitoring of processes via thermal radiation. As a matter of principal, the temperature of a test object can be measured contactless with a spot-pyrometer. For this purpose, the spot to be measured is fixated with a laser beam. Larger objects, where the temperature distribution is required to be determined, are more complex. Every position to be measured must be located. The thermal radiation of each position is determined with the pyrometer. For larger equipment, therefore, many measurements must be undertaken. Coincidentally, errors may occur, as certain regions are not noticed or cannot be determined accurately.
A thermal image sensor reduces this effort. Similar to an image sensor – composed of many pixels – it works essentially on the same physical basis as a pyrometer.
Due to the higher number of pixels, the temperature distribution of larger objects can be determined in only one measurement process. The matrix- shaped arrangement of the pixels generate a thermal image of the monitored object.
Similar to a regular camera, a picture of an object is taken in the infrared scope. Particular algorithms improve image recognition.
Synchronously, certain colours are attributed to respective radiation intensities. The main objective is not an exact temperature measurement, moreover the temperature of partial objects in a certain grid must be recognized. Temperature disparities to neighbouring objects should be detected equally. Videos can also be recorded with such a thermal image camera. Hereby temperature changes in erroneous processes can be detected.
Examples of use
Thermal image cameras based on this principle have become widely-used in various fields of application in the last couple of years. A few examples illustrate the operating areas:
Building industry, tenements: :
- control of heat insulation
- early warning of mildew in rooms
Police, fire departments:
- remote detection of burglars in empty spaces
- detection of hot spots and individuals in smoke-filled buildings
- detection of circulatory disorders and inflammation
- detection of circulatory disorder, caused by zero gravity in aerospace medical science
- driver assistance systems for detection of animals and humans crossing the road at night or in fog
- error checking of power train
- fire monitoring
- detection of inflammation in animals
- detection of animals during harvest
- punctual detection of errors in driving engines of industrial plants
- control of power distribution and transformer stations and their safe circuit points – especially for high voltage facilities a safe distance is inevitable
- controlling temperature distribution during production methods in high temperatures (semiconductor production, soldering processes, injection moulding)
- monitoring pipelines regarding discharge of gas or liquids
- quality control for PCBs concluding assembly
- control of hardware, that is heated quickly, on cracks in leadfree soldering
Logistics, waste disposal:
- early warning for conflagration of packages possibly containing hazardous goods (lithium batteries)
- detection of smouldering fires on waste heaps
Thermal image sensors with a high pixel concentration are expensive. A great deal of the expenses is due to the high value lens. The MLX90621 by Melexis offers an ideal solution
|Number of pixels||Matrix 16 x 4|
|Optics||Objektive integrated lens|
|FOV||0° x 15°, 40° x 10°, 120° x 30°|
|Calibration of object temperature||-20° to 300°C|
|Measurement accuracy||+/- 1°C within 0° bis 50°C|
|Framerate||programmable 0.5 to 512Hz
Synchronisation via external driver
|Operating voltage||2.5 bis 3.3V|
The infraredsensor provides an image with 64 pixels in a matrix of 16x4 pixels. An additional lens is not mandatory. The lens is already integrated within the sensor.
The sensor is available in 3 different versions with different viewing angles. Depending on distance and angle of view, zones of different size can be monitored.
Every pixel measures the thermal radiation of one specific area. The data is collected via a low-noise amplifying circuit and fed to an AD converter with an 18 bit resolution. The hereby ascertained data is stored in the internal RAM. The sensor’s ambient temperature is measured with a PTAT-sensor and also stored in the RAM simultaneously. The MLX90621 comes as a factory pre-calibrated sensor, straight from Melexis. Relevant data is stored in the EEPROM. The MLX90621 is connected to the controller via the I²C Bus.
With the thermal radiation, the ambient temperature and the calibration measurement, the exact object temperature can be calculated for each individual pixel.
The computational procedure is documented extensively in the datasheet.
The first conducted test
In order to develop an effective application, experience can be gained by conducting tests. For this, the Evaluationboard EVB90621 – which is available at Dacom West – can be utilized.
With the freely available software, heat image data can be displayed without much effort. The EVB is connected to the PC via the USB interface. Subsequently the software is launched and the object in question is targetted by the EVB.
In the next step adjustments are made for the desired performance. The software enables the display of current heat images or videos. For this purpose the number of pixels displayed can be chosen from 4x16 to 32x128. The sensor has 4x16 pixels. Nonetheless the software can generate a realistic display with 32x128 pixels. This is possible by the means of mathematical algorithms of bilinear interpolation. To understand the principle quickly, we can use a metal plate as an example.
The plate is heated in one specific area. The combustion point, where the beam hits the plate, is the hottest. The rest of the plate does not remain cool - the heat distributes itself equally from the combustion point.
For two known measuring points that are not too far apart, a gradient can be measured through interpolation. The interpolation is an approximation method, meaning that an approximated value is calculated between both measuring points. In bilinear interpolation, the principle of measurement is extended to a square field. Hereby the temperature distribution within the field can be determined with four measuring points.
The data, processed by the image sensor cannot only be used for image presentation. The data can be used for further evaluation methods.
For this purpose adjustments to the data, which is to be stored, can be made with the button „Application“. An analysis is possible in Excel or other data processing software.
The MLX90621 is not solely meant for generating thermal images, but can also be used for driving technical surveillance systems. By interpretation of data with mathematical software, algorithms can be developed for automatic drivers.
Subsequently, after the experimental measurements, we can start developing the circuitry and the appropriate software.
The hardware composition is relatively basic. The MLX90621 is connected via an I²C Bus with an arbitrary microcontroller. The choice of controller is dependent on the planned software.
In the case where the temperature is measured for each pixel and a warning signal is emitted, when a threshold value is exceeded, a standard 8-Bit controller is sufficient.
More complex algorithms – for instance for thermal images with a higher pixel concentration – require a high-performance controller. This way, images for thermal image videos with a higher number of pixels can be calculated in real-time.
The I²C protocol s mentioned extensively in the datasheet.
The software development to process measured data is complex. For security applications, sufficient checking is necessary after development.