On the static and dynamic characterisation of microbolometer thermal camera

Author(s):
F. Romanò - IMAGINGLAB

Industry:
Industrial Controls/ Devices/ Systems

Products:
PXI/CompactPCI, LabVIEW

The Challenge:
Verify the static calibration, the geometrical resolution and the dynamic response of a microbolometer thermal camera

The Solution:
The solution is a NI HW + LabVIEW SW system based. In the testing has been used also a climatic cell to verify the camera response in the limit of its operating temperature range.

 

Short summary

Thermography cameras are spreading, as a tool for temperature measurement, due their continuous cost lowering and to the easy of use, mainly related to the non contact measurements made available. However, temperature detection relying on radiation measurements is a somehow difficult task, for a number of reasons, so a mean to evaluate the reliability of these measurements is a preliminary unavoidable step [1]. Apart form the well known problem of emissivity, which is responsible for systematic shifts, other additional performances are to be checked [2].

In addition to the more commonly investigated static performances of radiation cameras, also the dynamic ones are of concern, for instance in the case of temperature measurements on non stationary thermal fields. An assessment about the dynamic capabilities to correctly measure these varying temperature fields must therefore be a further goal.

 

Introduction

The present article is aimed at defining the performances of a semi-automatic procedure, for radiation cameras, to qualify these devices under both the static and the dynamic point of view, providing calibration tests under the most critical conditions. Some test devices have been proposed in literature, such as in [3], for the characterisation of “bare” bolometric detectors for development purposes; the approach followed in this article is, however, deeply different because the whole thermal camera is tested here, keeping into account the optics and the complete measurement chain, as it will be applied in the measurements. Together with the description of the designed test bench, some experimental tests are shown, as an example. The infrared camera this work deals with is of the micro-bolometer kind and is equipped with a 320x240 pixel focal plane array (FPA), with a spectral range of 8-14 mm. This kind of sensor is less expensive with respect to the photonic sensing technique, but it also exhibits some drawbacks due to the adopted technology [4].

The first stage of the work has consisted in verifying some facts related to a deeper insight on both the static calibration and the maximum allowable spatial resolution: while it is known that the sensing pixels interact with each other, smoothing the temperature gradients (crosstalk phenomena, [5]), the data available on the calibration certificate of the IR cameras are not always exhaustive.

Once the static response has been qualified, the dynamic performances of the thermal camera have to be checked. In the test device herein developed a number of rising and falling step inputs have been applied to the detector and the output time histories have been recorded; the analysis of the acquired time histories has allowed for the estimation of the system transfer function.

A fit-to-the purpose software, together with the suitable hardware, has been set to perform the experimentation, independently from the particular thermal camera under test, and allowing for the definition of a standard procedure.

During all the tests the thermal camera has been kept into a climatic cell in order to allow us to check the instrument performances under even extreme but controlled temperature environment, witch may affect the bolometer device behavior.

 

1) Static characterization

 

The reference input for all the tests is a black body equipped with a built-in Pt100 sensor for temperature reference; the uncertainty of the black body temperature is lower than 0.1 °C in the whole temperature range used in the tests, as given by the builder specification and by the calibration certificate.

 

1.1) Calibration check

In order to check the accuracy of the microbolometric sensor both in the mid part and in the border regions, the black body has been moved by means of a motorised support (figure 2) in order to put the “hot spot” due to the black body hole in 45 different positions in the field of view of the thermal camera, following a precise path in the sensor of the thermal camera. For each position the mean temperature of a 3x3 pixel region of interest (ROI) is measured: the ROI position is automatically set in the centre of the black body hole. The test was repeated changing both the climatic cell temperature (-10°C, ambient, +45°C) and the black body one (50°C, 110°C, 120°C). In the image 3 the temperature distribution measured by different parts of the bolometric detector under stationary conditions is shown as an example (data smoothed with splines).

 

For each test 45 temperature values are acquired, as described above, then the average and the standard deviation values are calculated, together with the black body reference temperature. In the last column an estimation of the systematic effect is shown.

 

1.2) Geometrical resolution measurement

Microbolometric detectors can not correctly measure high temperature gradients [6]; this phenomenon is known as “the slit response function” and it is due to the “crosstalk” between adjacent pixels [5]. The capability of the detectors to measure thermal gradients should be checked to avoid any eventual dangerous temperature underestimation in the presence of small “hot spots”.

The resolution of the detector was measured using a system able to produce areas with high and low radiation in the field of view of the thermal camera (figure 4). The system has the black body cavity (high radiation) partially hidden by two high absorption plates (low radiation) and the distance between the plates is accurately measured with an uncertainty on estimated uncertainty in thermal image units of 0.18 pixel.

The entity of the temperature underestimation is correlated as a function of the number of pixels heated by the black body hole. The IR camera under test produces reliable measurement if the high radiation area has a width of 3 or more adjacent pixels.

 

2) Dynamic tests

 

The goal of the dynamic tests is to extract the system capability and limits in measuring time-varying phenomena. For these tests the grabbing frequency has been set to the maximum value available with the IR camera under test (i.e. 50 Hz). The easiest way to qualify the system dynamic response it to analyse the output due to a known simple input signal, such as a step. In order to check whether a 1^st order model can be used for modelling the bolometric sensor dynamic response, a number of rising and falling steps have been applied as an input. This kind of input has been generated by means of a rotating disk interposed between the black body and the thermal camera (figure 5): since the disk is shaped, a sequence of low and high radiation levels impact on the bolometric sensor: the cycle time is related to the disk rotation speed and to the opening shapes.

 

While the disk has a temperature equal to the ambient one (25°C) and an emissivity is close to unity, the black body temperature has been set to 125°C for the dynamic tests; the step is then 100°C. In the centre of the bolometric sensor a square region of interest of 3 by 3 pixels has been selected; the use of the ROI allows us to perform an average over 9 pixels, increasing the signal to noise ratio. On the other hand it is necessary to keep in mind that, the larger is the ROI, the longer becomes the “transient” time between the low and the high input level. With the geometry of the used disk, this transient time is of the order of 0.5% of the square wave cycle time: a value that can be neglected for this application since the fluctuation in the estimated t values is larger than 1%, as will be shown later.

The time history of the ROI average temperature is similar to the response of a 1^st order system subjected to a square wave input. The time constant t has been estimated for each of the rising/falling step, by minimising the mean square distance between experimental points and the theoretical step response Y(t) of a 1^st order system at a given step input ,

 

                                            (1)

 

where t denotes the time, while Y_min and Y_max are the maximum and the minimum limits of the input steps.

 

Since it was observed that the detector behaves as a 1^st order system [7], the parameter to be measured is the time constant t. The values of the rising and falling step time constants have been estimated separately, because they are experimentally different; this is due to the different heat transfer conditions [7].

The average and standard deviation time constant values shown in Tab 1 have been estimated with a 45 test set.

 

 

Tab 1: Bolometric array time constant estimation

 

Once t is known, together with the static characteristics defined in the first part of the article, the system transfer function can be easily calculated through the equation (2):

 

                                                   (2)

 

where w is the circular frequency, and i is the imaginary unit.

 

For the thermal camera under test the time constant value t is, in the worst case, 15.11 ms: the 1^st order system cut-off frequency ([7]) is then about 10.5 Hz. The grabbing frequency of 50 Hz is then adequate to the frequency response of the sensor.

 

3) Concluding remarks

 

In the article a semi automatic system for the measurement of some static and dynamic thermal cameras features is presented. The system can be used for testing almost any type of thermal camera and, thanks to the limited and guided human role, the results are expected to be reliable and repeatable.

 

4) References

 

[1]  S. Rainieri, G. Pagliarini, “Data processing technique applied to the calibration of a high performance FPA infrared camera”, Infrared Physics & Technology, 43 (2002) 345-351  Elsevier

[2]  N. Horny, “FPA camera standardization”, Infrared Physics & technology 43 (2003) 109-119,  Elsevier

[3]  N. Liberatore, A. Pifferi, S. Perri, M. E. Marini, “Test bench for IRFPA based on CMT and microbolometer”, Infrared Physics & Technology 43 (2002) 291-296, Elsevier

[4]  A. Rogalski, “Infrared detectors: an overview”, Infrared Physics & Technology 43 (2002) 187-210, Elsevier

[5]  J. Ziegler, M. Bruder, M. Finck, R. Kruger, P. Menger, T. H. Simon, R. Wollrab, “Advanced sensor technologies for high performance infrared detectors”, Infrared Physics & technology 43 (2002) 239-243, Elsevier

[6]  Infrared Training Center, Methods for assessing the instrument performance, pg.3, “The slit response function”

[7]  E. O. Doebelin “Measurement systems, application and design” Fourth Edition Mc Graw-Hill International Editions, 1990

[8]  L. Tissot, “Advanced detector technology development at CEA/LETI”, Infrared Physics & technology 43 (2002) 223-228, Elsevier

Browse All Case Studies »