2.1

Thermal

The active area of the detectors must be cooled down in order to reduce the parasitic noise. Apart from the packaging made of Kovar, the SWIR and VNIR detectors are different for many aspects such as the light spectrum range and thermal management.

For the SWIR detector, the operating temperature of the chip is 185K. The cooling down to this operative temperature is simply achieved by conduction through the backside of the package. Two transistors are glued into the package and are used as thermal sensors in order to read and / or control the temperature of the detector.

For the VNIR detector, the electro-optic technology is based on CCD. The chip is composed of an active and storage areas mounted on a ceramic layer. Two thermal sensors are also glued on this ceramic layer to read and / or control the temperature of CCD at -20°C in order to reduce the noise. Unlike the SWIR detector, the cooling down to this temperature is achieved by a two stages Thermal Electrical Cooler (TEC) directly integrated into the detector package. So the FPA shall cool down the package of the VNIR detector that will be hotter than the chip.

The FPAs are thermally coupled with the 3MI instrument through three thermal interfaces (see figure 2):

Figure 2.

Schematic representation of FPAs into 3MI instrument.

The radiative environment and baseplate I/F are identical in term of range and defined as imposed temperature covering hot and cold cases. A low emissivity surface (0.1) is specified for radiative environment.

The definition of the cooling I/F is more complex. Instead of an imposed temperature, the temperature of this I/F is defined as a function of the power rejected by the FPA. This function takes into account the performance of the radiator and the heat pipe in the case of VNIR FPA. Thanks to this method, the operation point of the FPA can be easily simulated without the need to iterate with thermal engineering at upper system level. As the function between temperature I/F and rejected power is linear, the thermal performance of the radiator can be modeled by a reference temperature defined for a null rejected power and a thermal resistance.

The following table sums up the value of the temperature I/F.

Table: 1

Summary of thermal I/F.

The FLEX cable between the FPA and the FEE also represents a thermal interface by mains of:

The thermal flux transmitted by this part will be managed by design in order to have a negligible impact of the temperature of the DU.

The maximum power dissipation of the detectors and the maximum power available for the thermal control are indicated in the following table.

Table: 2

Detector and thermal control maximal power.

2.2

Mechanical

The main mechanical requirements for both FPAs can be divided into the following items:

Stiffness, thermal regulation, dimensional stability and IRD are the main design drivers that will be discussed in details in this paper.

Random vibrations and shocks loads are also challenging but can be managed by iterations with mechanical engineering at upper system level thanks to notching approach or level update ([5]).

3.1

General considerations

The design of a FPA, like most of the opto-mechanical devices and in particular in space applications, is a complex task because the specifications are numerous, with a high level of accuracy (temperature and dimensional stability) that must face severe environments. These specifications, especially mechanical and thermal ones, lead to conflicting solutions. A tighten optimization in the choice of materials, architecture and dimensioning is needed to overcome any stalemate.

The project addresses the development of two FPAs. SWIR and VNIR specifications are identical or very close from a mechanical point of view but opposite in term of thermal specifications:

Despite these technical challenges and differences of specifications between SWIR and VNIR FPAs, the development of a specific solution for each of these two FPAs is to be avoided because it will lead to additional costs and time delays. Therefore, the strategy is to develop one FPA and adapt the resulting design to the other FPA. This approach during the developing phase is also beneficial for the following phases: common parts lead to reducing purchase costs and delays, similarity of product design is reflected on tools and methods for integration and testing.

SWIR FPA appears to be more dimensioning in comparison to VNIR FPA. Without any computation, the thermal specifications show that the design is out of the state of the art. For SWIR, the detector must be at a temperature lower than -88.15°C with a temperature of thermal sink at -100°C that represents a thermal gradient of 12°C that is lower than usual margin of 15°C (5°C for analysis uncertainties, 5°C for acceptance and 5°C for qualification as defined in [3] and [4]). On one hand, this small gradient implies to have a high thermal coupling. On the other hand, the thermal gradient with the baseplate is very high, up to 133°C, thus implying to have a high insulation to avoid too much thermal losses.

3.2

Materials

The choice of materials is a key step in the design process because it implies to take into account many compromises:

Silicon Carbide is widely used in FPA structures because of its dimensional stability: it’s a rigid material with low deformation under thermal loads thanks to a high thermal conductivity to reduce thermal gradient and also with a low CTE. The latter property is usually compatible with the detector material. The main drawback of this material is its fragile behavior under shock.

In addition, with the low temperature range specified for SWIR FPA, only a DSA made of Kovar, i.e. identical to the detector package, is suitable to avoid sliding at screws interface. Flexible items could be used to relax thermal elastic efforts but they are not compatible with high eigenfrequencies requirement, thermal coupling and also the lack of room allocated by the IRD.

For similar reasons, 6082 aluminum alloy is used for the baseplate and for the end of thermal strap in interface with the 3MI instrument. The thermal treatment is basically T6 but, to face the reduction of thermal conductivity at low temperature observed on aluminum alloys, a O temper is defined for the thermal end (see §3.6 for details on thermal strap).

The most critical part of the FPA remains the ISM because it must insure its thermal insulation and mechanical stiffness. These two functions can be realized by separated parts. For example, ceramic washers are widely used as local thermal insulators. But, after a detailed analysis, they don’t provide any advantage in comparison to shaped titanium parts: higher volume and mass at identical thermal resistance, fragile materials with risks of failure during assembly phase and environment loads, the need of additional washers for thermal-elastic compensation of screws tightening, an increase in the number of parts, an increase in the complexity of assembly phase. Some composite materials, fibers encapsulated into resin, have also been foreseen because of their low thermal conductivity and their relative stiffness but their stability with respect to humidity absorption and creeping over time storage of satellite before launch are not compatible with dimensional stability.

Finally, titanium alloy and more precisely TA6V is chosen because it represents the best material for all the constraints presented above. However, this choice must be completed by the geometry of the ISM to reach all the performances.

3.3

Thermal budget

The global thermal budget of the FPA can be generalized by the following figure where the temperature of the DSA is a function of temperature and thermal resistance at the two interfaces:

Figure 3:

Global thermal models of FPA.

This figure shows the main difference between SWIR and VNIR FPA:

In this model, the thermal flux brought back by the flex cable to the FPA can be neglected because:

Based on this model for the SWIR FPA, the following equation gives the relationship between the DSA temperature according to the cooling and losses thermal resistances, the interface temperature of the cold sink and the baseplate.

The power of radiative coupling with the environment is hand-computed and remains low if parts are treated with low emissive coating like gold or SurTec. The desired temperature of the DSA is computed from the detector characteristics (operating temperature, resistance of the package and its assembly and the dissipated power):

Taking into account the fact that equation Eq. 1 is applicable for hot and cold cases with, in particular, a null value of heating power in hot case, the values of the cooling and losses resistances can be determined and used as derived specifications for the design guide line:

The range of magnitude of these resistances confirms that the DSA must be highly coupled with the cold sink and highly isolated from the baseplate to limit the losses. It is due to the fact that the temperature of the DSA (-89°C) is very close to the cold sink in comparison to the baseplate. As the resistance of losses is very high, the power dissipated by the heater is directly injected to the cold sink, it can be observed that the resistance of cooling corresponds to the compensation of the temperature variation of the cold interface between hot and cold cases: 6.23K/W x 3W = 18.69°C ≈ 118°C – 100°C. It also means that the need of heating power can be reduced if the resistance of cooling is increased but with the drawback to also increase the losses resistance to maintain the equilibrium in hot case.

Each of these two resistances is also derived into two resistances:

3.4

Thermal and mechanical optimization of ISM

Form this primary analysis and specification, the ISM must respond to two opposite objectives:

As already discussed, Ta6V titanium alloy is identified as the best material for this type of application. The important challenge is now to find the optimal geometry to achieve these objectives. Thermal insulation requests to design parts with a reduced cross-section and an increased length while stiffness requirements lead to the opposite.

The DSA, as well as the detector, also play a role in the stiffness performance of the FPA because they represent the major part of vibrating mass, notably by the fact that they are made of Kovar. Nevertheless, these items cannot be thoroughly optimized. Firstly, the mass of the detector is an imposed input data. Secondly, the size of the DSA and then its mass is constrained by two parts in interface with it: the detector and the thermal strap.

However, a significant parameter is the location of the fixing of the thermal strap on the DSA. Basically, two options are possible: as the cold sink is located on the side of the FPA, either a part of the DSA passes through the ISM to join the thermal strap or it’s the thermal strap. During design tread-off, the first option has been chosen but the second one appears to be the best for many reasons:

In addition, as the thermal strap is a flexible part (see §3.6), its resonance frequencies are very low (around 50Hz) in comparison to stiffness requirement and its mass does not bring in account the effective mass of the FPA design.

Figure 4:

Final design of the SWIR DSA.

The VNIR DSA is adapted from SWIR DSA to take into account the difference of detector interface and electronic boards.

The method to optimize the thermal and mechanical design of the ISM is based on two steps:

The following figure shows the different architectures that have been envisaged.

Figure 5:

Some of the explored architectures for ISM.

The results of this optimization phase are analyzed by means of a Pareto diagram (figure 6). On this diagram, each iteration is represented by a point which stiffness is reported on the horizontal axis and the thermal resistance on the vertical axis. To be compliant, an iteration must have its representing point located in the white area: Rth > 200K/W and freq. > 500Hz. The points having the same color correspond to the optimization of one architecture.

Figure 6:

Pareto diagram of stiffness and thermal resistance objectives.

The Pareto diagram shows that the only architecture capable to fulfill the thermal and stiffness requirements is the number 11.1 corresponding to a hexapod composed by 6 shafts. This architecture is an extrapolation of a trivial hexapod used to sustain large structures like mirrors of telescope or other optical devices ([6], [7], [8]).

This architecture offers other advantages:

The direct drawback of this latter advantage is that angled surfaces must be manufactured with a high accuracy (50µm to 10µm) to limit the initial stresses during the mounting. Special tools are designed to secure the shaft, notably to counteract the torque during the tightening of the screws.

In the final optimization, buckling is also taken into account as dimensioning criteria because shafts are slim structures ([9]). Equivalent load to maximum RMS acceleration acceptable by the detector and a safety factor of 10 is considered to cover the uncertainty of linear buckling analysis.

On the figure 7, one can notice that the fixings of the shafts on the SWIR baseplate are aligned with the fixings of the baseplate on the 3MI instruments. It allows avoiding any contribution of the stiffness of the baseplate. Unfortunately, the application of this design rule is not possible for VNIR baseplate and we observed a lower stiffness of the VNIR FPA (620Hz) in comparison to SWIR FPA (798Hz).

Figure 7:

View of SWIR and VNIR hexapods with thermal strap.

3.5

Thermo-elastic dimensional stability

The in-plane stability can be easily managed if the whole structure is symmetric and aligned around the optical axis of the optical bench:

The fact that the optical axis of the detector is not aligned with its package creates an unwanted shift of the center of gravity of the vibration mass that penalizes the dynamic behavior by rotations of the DSA and then an increase of amplifications factors.

The out-of-plane stability is more complex because the architecture must be able to compensation the thermal expansion of the materials in order to maintain the detector at the same distance from the baseplate.

Some of the architectures presented in Błąd! Nie można odnaleźć źródła odwołania. are designed with an ISM having a ‘round trip’ shape along the thermal path. The vertical expansion of the forward path is compensated by the expansion of the backward path. If the lengths of the two paths are not identical, their cross-sections must be adjusted to distribute their respective thermal gradients with a ratio inversely proportional to the ratio of the lengths. With this architecture, the thermal gradient within the DSA must be negligible that implies the use of material like SiC instead of Kovar.

Figure 8:

Out-of-plane dimensional stability behavior of FPAs with ‘round trip’ principle.

This principle is not feasible with a hexapod at least within the room allocated by the IRD.

The vertical compensation of hexapod is obtained by a more complex kinematic. When the temperature increase, two phenomena occurs:

An optimization of the angle of the triangle forms by two shafts permits to adjust the two phenomena to obtain a perfect compensation.

The mean temperature variation of the shafts is the median of the variations of the baseplate and the DSA. As the thermal control is different between SWIR and VNIR FPA, the variations of temperature of the shafts are not the same for the two FPAs (see figure 9):

Figure 9:

Out-of-plane dimensional stability behavior of SWIR and VNIR FPAs with a hexapod.

However, in order to keep a common design between SWIR and VNIR, a correction is not complete and a compromise is established. The relative displacement of the DSA is:

for a temperature change of between median case and hot case.

3.6

Thermal strap

An important tread-off has been carried-out in order to realize the thermal strap with the following criteria:

The stiffness of the strap must be taken into account when dimensional stability is needed like for FPA devices. The relative displacement of the thermal strap produces a displacement of the DSA proportional to the ratio between the stiffnesses of the thermal strap and the ISM: 40 to 150N/m for transverse directions and 2500 to 3500N/m for direction along the length of the strap depending on SWIR or VNIR FPA. Tests on samples and hand-computations based on beam theory demonstrate that the flexibility for this latter direction is mainly tunable according to the shape of the strap: an important loop promotes more flexible bending deformations compared to traction/compression.

Three different thermal strap technologies are described hereunder:

Finally, copper braid technology is retained for its capability to be directly integrated into the structural parts of the FPA, without intermediate ends. It allows a reduction of volume, weight and no contact thermal resistance at connection. The connection between the strap and the structure is a key point, whatever is the retained technology. For 3MI FPA, a special technology is developed with FORISSIER company: a silver tube is crimped around the braid (see figure 10). It facilitates the insertion of the braids into the holes of the mechanical parts and permits a better control of the thickness of the joining material. In replacement of solder initially envisaged, a glue, specifically designed for cryogenic applications (CTE closed aluminum, low outgassing), is used to maintain the braids into the mechanical parts.

Figure 10:

VNIR braid with silver tubes at each end and assembly (DSA, braids and thermal rod) during AIT.

For thermoelastic reasons, the end of the strap in contact with the 3MI instrument (named thermal rod) must be made of aluminum. A special attention is paid to the choice of the grade and thermal treatment because one can observe a decrease of the thermal conductivity at low temperature for many of them ([10]). The best compromise found between thermal and mechanical performances is the grade 6082-O.

3.7

Gluing of detectors

Thermal resistances of detector package are indicated with in ICD but are only applicable if package is in thermal contact on the whole area of the back side of the package. In the early design, the detectors were only fixed on the DSA by 4 screws. In this case, the thermal exchanges between the DSA and the detector can only be achieved at the 4 corners. As a result, the assumption declared in the ICD was not obtained and the thermal resistance of the detector package was much more important, 27K/W instead of 0.9K/W, especially as the package is made of Kovar that is a low thermal conductive material. In addition, the same phenomenon occurs in the DSA because of the distance between the corners and the center where the thermal strap is fixed (see figure 11).

Figure 11:

Addition of glue to avoid gradients in detectors and DSA.

The solution to shorten the thermal paths consists by adding a Thermal Interface Material, TIM between the detector and the DSA. The criterion used to choose this TIM takes into account thermal performances but also mechanical aspects:

The loads criterion mentioned above concerns the screws used to fix the DU on the DSA and also those applied to the DU. Even if no specific limit is mentioned in the ICD of the DU, Sodern looks for TIM allowing a limitation of the efforts transmitted to the DU. In the sense, elastomer or soft metallic layers are rejected. The usage of a glue identical to the one retained for the thermal strap is the best candidate. Qualification process to verify the adhesion on gold coating is still in progress.

3.8

Electronics integration

In §0, the thermal resistance of the flex cable of SWIR detector must be higher the 200K/W. With respect to electronic signals performances, the flex are designed with the following guide lines:

The space left for the integration of the proximity electronic is very small because of the size of the detectors in comparison to the space defined in the IRD. The electronic boards and flexes are shaped to pass around the 6 shafts of the hexapod and fixed on the DSA.

4.1

Mechanical tests

AMM, Advanced Mechanical Model

Early mechanical models are realized in order to:

Notching is defined to limit the acceleration in random vibration and shock at detector interface, respectively 51gRMS and 1250g, half-sine during 0.5ms finally converted into SRS.

A ballast is designed and mounted on the DSA in order to represent the mass of the detector and the proximity boards.

Figure 12:

Views of SWIR and VNIR FPA AMMs (configuration for tests in X axis).

Tests performed with SWIR FPA AMM shows that the thermal strap has no influence on dynamic behavior of the FPA, whatever is in terms of stiffness or damping. For this reason, no thermal strap is mounted on VNIR FPA AMM.

The table 3, figure 13 and figure 14 show that the FEMs can correctly predict the mechanical behavior of the FPAs that are relative simple structures. The measured amplification factors are more important than expected: from 40 to 190 according to the frequencies modes and direction.

Figure 13:

Comparison of measured and computed LRSC in Z axis between AMM and FEM.

Table: 3

Eigenvalues and amplification factors measured on SWIR and VNIR AMMs.

Figure 14:

Comparison of measured and computed shock responses in Z axis between AMM and FEM.

Even if the output SRS levels exceed the maximum level of detector for frequencies close to the resonances of the FPAs (1000Hz), these levels are acceptable because they are below this limit for frequencies above 1500Hz corresponding to the first frequency resonance of the detector.

STM, Structural and Mechanical Model

The first aim of the STM is the validation of the thermal performances (see §4.2). For this reason, functional detectors are mounted on the FPA in order to have a full representation of their thermal behaviors: power dissipation of detector and proximity electronics, dissipation of the TEC for VNIR, losses by the flex cables, measures of temperature of sensing elements.

Mechanical tests on STMs confirm the dynamic behavior observed on AMMs.

The main interest of the STM mechanical tests campaign is the measurement of the dimensional stability with respect to the mechanical and thermal environments including operational and non-operational conditions. A set-up of these measurements is performed in order to quantify and reduce as much as possible the accuracy. 3D machine programing, positioning of the sample and statistical approach permits to reach an accuracy of ±2μm. Apertures in the cover of the FPAs are tooled in order to allow the access of the sensor to the DSA and the detector and the outlet of accelerometer cables.

Figure 15:

Views of SWIR and VNIR FPA STMs.

After each environment, a series of measurements with the 3D machine are performed. The mean value and the standard deviation of the series are computed and are reported on the following graphs:

Figure 16:

Results of dimensional stability measurements after random, thermal and shocks environments.

From these data, we observe that no permanent deformation or sliding occurs during vibration, shock and thermal tests:

4.2

Thermal tests

Thermal Vacuum Equipment

In comparison to standard thermal tests, the particularity of the FPA tests is the fact that the cold sink interface temperature is not fixed value but must be representation of the linear variation with rejected power (see table 1). The implementation of this specification is realized by the design of aluminum rod with the appropriate length and cross-section. Even with a 3D FEM analysis, the thermal resistance must be measured in order to recover uncertainties mainly due to the thermal conductivity decrease at low temperature. In addition, the residual error with the specification must be corrected with the adjustment of the detailed thermal model in order to determine the thermal performance of the FPA in specified environment.

Despite of the difficulties encountered to implement this specification, an advantage is the capability to measure the power rejected at the interface. It allows to validate the performances of the device in terms of heat fluxes in addition to temperatures.

For SWIR STM only, the disturbance of the thermo-couples must be taken into account because they create parasitic couplings between cold parts and room environment even if special care is done to reduce this effect: thermal loop, K type thermo-couple with insulating metals instead of copper. The influence of each thermo-couple is evaluated taking into account the conduction and the radiative exchange. This correction is unneglectable and represents 3.5°C.

Thermal Results

After corrections presented here above, the thermal performances of SWIR and VNIR FPAs are validated:

Figure 17:

Thermal gradients in FPAs for cold case.