814 
Temperature 
Windows 
-50 °C 
-20 °C 
0°C 
20 °C 
50 °C 
8-14 pm 
0.0112 
0.0071 
0.0055 
0.0045 
0.0035 
3-4 pm 
2.2 
0.36 
0.14 
0.059 
0.02 
Table 3. Theoretical noise equivalent difference temperature (NEDT) (K/Hz 1 ^ 2 ) at different blackbody 
temperatures for an ideal 1 -pun wide filter with a transmittance equal to unity 
Table 4 shows a comparison between theoretical and experimental noises in the broadband channel. 
The theoretical noise takes only into account the detector and its preamplifier, whereas the experimental 
"noise" accounts for all possible sources of fluctuations, one of them being the cavity temperature fluctuations. 
This appears as an increased experimental noise equivalent voltage. The difference between theoretical and 
experimental NEP and NEDT values is increased even more, due to the discrepancy found between the 
theoretical and experimental sensitivities, which is discussed in Section 4.2.1. 
Theoretical 
Experimental 
Voltage noise 
(nV/Hz 1 / 2 ) 
32.6 
63.7 
Radiant power 
noise 
NEP (nW/Hz 1/21 
0.27 
1.01 
Temperature nois 
NEDT (K/Hz 1 / 2 ; 
0.0014 
0.0052 
Table 4. Comparison between theoretical and experimental prototype noise at 23°C 
5 - CONCLUSIONS AND PROSPECTS 
Some points are still to be improved on the current prototype. The main insufficiency concerns the 
experimental value of the sensitivity of the broadband channel which is half of the theoretical one, 630 and 
1200 ct/pW, respectively. The major explanation for this is that the detector location has not been completely 
optimized due to the dimensions of the detector case. Another reason for this discrepancy lies in the 
uncertainty associated with the knowledge of the spectral transmittance of the instrument. This point appears 
clearly when comparing the sensitivities derived experimentally for the different channels of the instrument. 
These values would coincide if the spectral transmittances were accurately known. 
The measured voltage fluctuations of the instrument are twice the minimum voltage noise of the 
detector. The required accuracy, given in Section 2, of 0.05K for a 1-pm wide channel is reached with the 
current broadband channel of the prototype. Indeed, 0.0052 K/Hz 1 / 2 for the broadband channel corresponds to 
about 0.03 K/Hz 1 / 2 for a 1 -pm wide channel. However, the theoretical results of Table 3 show that 
measurements with the 3.7-pm window will involve a larger noise. Therefore, efforts are still being made to 
solve the problems described above. 
Further arrangements of the prototype are scheduled, such as installation of the internal blackbody and 
addition of a 3.7-pm window. A narrow FOV prototype is to be built. The questions of the instrument 
motorization for automated viewing, and of the environment for an airborne version remain to be tackled. 
Calibrations and intercomparisons between different prototypes as well as with other instruments are planned.