A.

Filters design

The system includes two dedicated infrared filters, mounted in a field-separation concept close to the focal plane. They allow to measure the temperature of the clouds using stereovision algorithms. In Fig. 3 is shown the two filters set, a mosaic that separates spatially the FOV in two regions that observe the scene in two spectral bands. The spectral response of each filter is also given.

Fig. 3.

IRCAM filters mounted on the microbolometer detector, and spectral response

A technological problem to be remarked is the presence of a non-useful area for detection in the central part of the microbolometer detector. In this area there is an overlap of the two wavelengths λ1 and λ2 and not a “pure” λ1 or λ2. This effect is mainly due to the physical interface between the two coated filters on the window, and to the optical effect of the incident rays emphasized by the fast optics. This problem can be solved in the post-processing, after acquiring the images, and will be minimized in the final design improving the filters assembly.

B.

Breadboard model

A breadboard model has been manufactured by INTA to test the optical performances of the system. But in addition to the performance, assembly and integration procedures have been also validated. The breadboard lenses have been mounted in the same way they will be assembled in the flight model, and the tolerances and optomechanical processes have been successfully tested and verified. Pictures of the optics breadboard are shown in Fig. 4. Its mechanical support was designed to interface with the FEE prototype.

Fig. 4.

Optical system breadboard, showing the lenses (left), the filters set (right) and the mechanical support.

A.

TEC temperature control

The TEC control achieves a thermal stability better than 10 mK, as required. It is based on a current regulator implemented by a buck converter and controlled by two closed loops (current and temperature). The temperature can be set between 10 and 40°C, and it is only possible to heat up due to the fact that the temperature of the FPA (environment) will be always lower than the microbolometer setting temperature.

B.

FEE control via FPGA

The FEE control is implemented by means of a radiation tolerant Proasic FPGA, with the following functions:

C.

Detector data acquisition

The electronics in the FPA are mainly in charge of providing low noise-high bandwidth amplifier adaptation of the microbolometer outputs to differential inputs and performing the analog to digital conversion, to maximize signal to noise ratio. The microbolometer provides data through each of the two output channels at 10Msamples/s and therefore, the analog acquisition chain has been designed to cope with this data throughput.

D.

FEE Prototype

The Spanish company SENER S.A. has manufactured a FEE prototype (FEEP), with components selected to be the commercial noise performance equivalent parts to the flight components, to verify the IRCAM requirements in the laboratory. The prototype implements the functionalities of the FEE; however, it is not representative of the flight model neither in form nor in radiation characteristics. In the prototype the FEE electronic assembly and the FPA have been implemented in a single Printed Circuit Board (PCB) for simplicity. Both parts are electrically isolated in the PCB and can be connected by a harness simulating the flex connector in the flight design, or by direct electrical connections using jumpers. Further details about the design and implementation of the prototype are included in [10]. Fig. 6 depicts the FEEP.

Fig. 6.

IRCAM Front End Electronics Prototype, including FPA (left part) and FEE (right part).

A.

Calibration of temperature response

The detector response to temperature will map the level of counts generated in the detector with a given object temperature under specific conditions [13]. The camera response to scene temperatures is measured in the laboratory using the precision blackbody as the object temperature. The object temperature is scanned from 243K to 333K, increasing the temperature range used in [13]. The images obtained are preprocessed to remove image bias and to reduce the noise by frame averaging. This experimentally obtained data is then fitted to the theoretical response curve of the detector. The theoretical response of the camera is linear with the radiance of the object observed. In the case of a blackbody, the radiance depends on its temperature. Additionally, the optical filters used in the camera determine the wavelength range. The radiance of the blackbody for a certain frequency range is expressed as [14]:

In this equation L(γ, T) is the spectral radiance of the blackbody, γ is frequency, T is the temperature of the blackbody. The integral included in (1) can be approximated using the numerical method proposed by Widger [15] in order to compute the final curve fitting. The curve fitting process will allow us to extrapolate the response curve to the whole temperature range required for IRCAM (down to 200K), which cannot be fully recreated in the laboratory.

The detector response depends directly on some configurable and environmental parameters. Accordingly, these parameters must be fixed during the test. The conditions were the following, a reference scenario for IRCAM:

Fig. 8 depicts the camera response as a function of the target temperature. The response has been evaluated in two sub-windows of the microbolometer, each one corresponding to a narrow band filter. Measurements have been fitted to the theoretical model and extrapolated to the final temperature range. The measured values are displayed using the star symbol. The curve slope, that is the camera response factor in terms of A/D converter units per K (ADUs/K), has been evaluated at different target temperatures. The results are shown in Table 2.

Fig. 8.

IRCAM response curves to scene temperature with the different narrow band filters.

Table 2.

IRCAM response factor in terms of A/D converter units per K.

B.

NETD characterization

The Noise Equivalent Temperature Difference (NETD) is a figure of merit for IR detectors which represents the temperature change that gives an output signal equal to the rms noise. NETD can be expressed as [17] :

Where V_n is the noise signal, ΔT a temperature variation, and ΔV_s is the signal value corresponding to ΔT in the detector. In an infrared camera, the NETD value is a key characteristic that defines the minimum resolvable temperature increment. For IRCAM, this value is one of the mission requirements and must be <3K, which correspond to a 500m in cloud altitude. The final NETD include different sources of noise such as detector and ADC noise, calibration accuracy, thermal stability and retrieval algorithms accuracy. In the laboratory test, the IRCAM telescope assembly containing the FEEP, the microbolometer FPA and the optical system (narrow band filters included) is characterized. The subsystem requirement has been fixed to a maximum NETD of 1K for a 300K target, without frame averaging processing.

There are different procedures [18] to measure the noise Vn of an infrared camera required to obtain the NETD value. One procedures to accurately measure this parameter is the temporal/histogram method. However the temporal/histogram method did not offer very accurate results in the IRCAM set up. IRCAM FEEP transmits an image every 17s, so a RS422 link at 921600 bps is used to transmit the image. Using this link, the maximum frame transfer rate is approximately 0.1 fps. Therefore, if a long series of images is acquired during the characterizations tests, a long time interval exists between the images. This leads to low frequency temperature drifts in the test setup components (environmental temperature, bias image temperature, blackbody, etc.), which are considered part of the camera noise when the histogram/temporal NETD method is used. Consequently we have used an alternative method to compute the noise, which is not so dependent of low frequency temperature drifts of the setup. This procedure is based in a spatial noise variance measurement, acquiring two subsequent images subtracted in order to eliminate the detector bias. The noise measured using this procedure is about 2.24 ADUs, and the results obtained using the histogram/temporal NETD procedure with different series length varies from 2.52-2.88 ADUs. The NETD measured in this test does not include the noise reduction techniques that will be used in the final system. Accordingly, the final NETD value can still be reduced, and the improvement will depend on the number of accumulated images (N). For N = 4, the preprocessed image (including the subtraction of an averaged set of N offset images) will reduce the noise in a factor √2.

The NETD values are calculated for a target temperature of 300K in order to compare the results with the IRCAM telescope assembly requirements, which define that the value shall be <1K at 300K. The NETD value is different depending on the optical filter. The measured results are shown in Table 3. They are slightly worse than those in [13] due to the extended temperature range and, therefore, longer characterization duration (temperature drifts); but the results comply with the requirement. The NETD values using the commercial INO camera of the EUSO-balloon project [19] are also shown for comparison purposes.

Table 3.

IRCAM measured NETD value at 300K.

The results in Table 3 have been calculated using sub-windows in a central part of the detector. It must be remarked that due to the low f-number and the focal distance of the optical system (f/1, f=19mm), non-negligible vignetting effects appear in the images. These effects generate an irradiance pattern where the irradiance decreases with the distance to the optical axis. The non-homogeneity in irradiance can be compensated using Non Uniformity Correction in order to obtain a homogenous image when the target temperature is homogenous. However, the lower irradiance in the corners of the image causes an increase in the NETD value that cannot be compensated. The reason of this behavior is that if the image is corrected to increase the response in the corners where the irradiance is lower, the noise in these sections will also be increased so that the NETD value remains constant. The NETD degradation caused by the optical effects in IRCAM has been evaluated using theoretical models [20] and laboratory images showing similar results. These results indicate that the NETD value in the corners of the image, the less favorable sections of the detector, is lower than 150% of the NETD value in the central part of the image.