The mechanical and thermal design and analysis of the VISTA infrared camera

R. L. Edeson, B. M. Shaughnessy, M. S. Whalley, K. Burke, J. Lucas Space Science and Technology Department, Rutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire, OX11 0QX, United Kingdom

ABSTRACT
The infrared camera for the Visible and Infrared Survey Telescope for Astronomy (VISTA) sets many technical challenges for mechanical and thermal design. The flexion between optical subsystems must be minimised to maintain alignment in various camera orientations and meet performance requirements. Thermally induced stresses, atmospheric pressure and earthquake loads place high demands on structural components, some of which must also thermally isolate the cold (~70 K) detectors and optics. The success of the design hinges on the optimisation of heat flow to minimise thermal loads on the detectors whilst holding external temperatures very close to ambient to reduce misting and convective disturbances in the field of view. This paper describes the mechanical and thermal components of the design and discusses the analyses in detail. Keywords: VISTA, infrared camera, mechanical, thermal, modelling, cryogenic, vacuum.

INTRODUCTION

The VISTA infrared (IR) camera is a wide-field infrared imager which is an integral part of the VISTA telescope1,2,3. Observations will begin in 2006 at the European Southern Observatory site at Cerro Paranal in Chile. The camera will be built, aligned and tested at the Rutherford Appleton Laboratory (RAL) in the UK. It is currently entering its manufacturing phase. The mechanical and thermal design of the camera was performed largely at RAL, with input from project collaborators at the UK Astronomy Technology Centre and University of Durham. The various structural, thermal and optical requirements on the design necessitated an iterative approach between design and analysis models. Consequently, analysis models were created at an early stage in the design program, and continually updated as the overall design matured.

DESIGN OVERVIEW

2.1 Camera description The VISTA IR camera will be approximately three metres long and weigh around three tonnes (see Fig. 1). The outer cryostat vacuum vessel will be assembled from several fabricated aluminium alloy sections, and essentially comprises a tube with a large bulge in one side to accommodate a filter wheel. A fused silica window seals the end of this camera tube. A series of electronics boxes are mounted around the base of the camera for detector control and signal processing. A flange around the midsection constitutes the mechanical interface with the Cassegrain rotator on the telescope. Three two-stage cryo-coolers are used to maintain the camera at the required operating temperature (see Sec. 3).

Window

Baffles

Optical bench

Lens barrel

Interface flange Liquid nitrogen heat exchanger

Glass Fibre Reinforced Plastic (GFRP) trusses Cooler

Filter wheel Focal Plane Assembly (FPA)

Electronics box

Wavefront Sensor (WFS)

Fig. 1. Section view of the VISTA IR camera.

Positioned in the bore of the camera is a tubular assembly housing a series of ellipsoidal baffles. The base of this baffle tube is mounted on an optical bench, a large aluminium alloy structure which is also the support for the lens assembly, a liquid nitrogen heat exchanger (for cooldown of the camera to operating temperature), and the structure supporting the focal plane. Directly below the lens barrel is a pair of wavefront sensors. Filters, housed in a rotating filter wheel assembly, are positioned between the wavefront sensors and the focal plane. At the focal plane is an array of 16 detectors, each mounted to a molybdenum base, which are then in turn bolted onto a molybdenum detector plate (see

Fig. 2). This detector plate is supported from an aluminium superstructure by three titanium flexures, and located laterally by a spherically-ended molybdenum pin in a cylindrical hole. Thermal straps connect each detector to a thermally controlled plate. Beneath this focal plane assembly (FPA) is a cold electronics box, connected to those outside the cryostat vessel via vacuum feedthroughs in the vessel base.

Detectors

Molybdenum plate

FPA support frame (aluminium)

Molybdenum location pin

Titanium flexure

Fig. 2. Section view of the detector mounting plate.

2.2 The mechanical design model The mechanical design of the camera was performed mostly using the 3D CAD software Pro Engineer4. A main assembly model of the entire camera was created, made up of various sub-assemblies and parts which could be worked on by different designers at the same time. This approach successfully gave ownership of different parts of the model to different people, and enabled different areas to develop concurrently5. Project engineers and work package managers had visibility of the overall model at any point in time. This way they could monitor progress, allowing potential problems to be identified early, as well as extracting information such as dimensions and mass properties for hand calculations and analysis models. The overall CAD assembly model was also a valuable tool during technical meetings. Over time, the design has been iterated with other subsystems, such as thermal, structural, and optical. The collaborative nature of the project means that there have been several instances where geometry had to be imported from other CAD systems. In these cases, STEP format files were used as the intermediary. Although this approach was successful in bringing together complex assemblies of different formats, the imported geometry could not be modified and the files sometimes contained more detail than required in the overall model.

2.3 Design at different temperatures One issue with the cold parts of the assembly (the baffles, FPA, filter wheel and attached parts which operate at wellbelow ambient temperature) was whether to design them hot or cold; for some of the larger components, there are significant dimensional changes from room temperature to operating temperature. For instance aluminium (as used in the FPA) would shrink approximately 4 mm per metre during the cooldown. It was decided at an early stage to generate the initial CAD geometry using operating temperature dimensions. This is because the optical configuration had been defined at operating temperature, and it is the optical design that determines certain key dimensions for the layout of the camera. Only when the details of temperatures and materials have been determined is it possible to determine the geometry at other temperatures. Interface control documents with other subsystems were also defined cold. The thermal analysis (see Sec. 3) showed that much of the cold aluminium structure could be assumed isothermal, and therefore a constant scaling factor could be applied to obtain room temperature dimensions for manufacturing drawings. The baffle tube assembly, however, has a temperature gradient along its length. On cooling this results in a change in geometry rather than a uniform scaling of dimensions. This necessitated a more detailed examination of cooldown effects along its length. The same assembly shrinks about 6 mm in the axial direction on cooldown. Quite apart from any possible issues with material stress or movements of sensitive optical assemblies, this required an examination of clearances between close-fitting parts to ensure that assembly would still be possible at room temperature, and conversely, that no unwanted interferences or gaps would occur during cooldown. A related issue is with loss of preload on bolted joints due to the different coefficients of thermal expansion (CTE) of bolt and joint materials during cooldown. In particular this could have a detrimental effect on baffle temperature performance, which relies on good thermal conduction across bolted interfaces. Therefore provision was made during the design for the addition of low-CTE spacers or spring washers to counteract the differential contraction and therefore maintain preload.

2.4 Design for assembly, installation and maintenance The mechanical design of the camera was carried out with assembly sequences in mind. There are parts of the camera that are very complex and will be difficult to handle due to their size and weight. The cryostat vessel is designed around a central midsection with good internal access when the main cryostat tube and lower section are removed. Assembly will be conducted with the camera in a horizontal position, and with the midsection supported by dedicated handling equipment at its Cassegrain rotator interface. CAD models were used to aid design for access and installation, and mechanism models run to ensure that the full range of movement of the camera (pitch and roll) was possible.

THERMAL DESIGN

3.1 Thermal requirements The temperature requirements for camera internal components are defined from requirements on radiative flux reaching the focal plane (see table 1). The camera must operate, without misting of the window, in ambient temperatures of 0 to 15 C and relative humidity up to 70%. Further functional temperature ranges have been defined, for which the humidity for operation without window misting must be determined. There are also requirements on maximum allowable temperature difference between the cryostat and the ambient air, in order to minimise the effect of convective disturbances in the field of view. The operational requirements for external surfaces are given in Table 2.

Item Baffles as seen by detectors Correction lens mounting Optics Region between detectors and lower lens Filters Infrared detectors

Temperature, K < 190 < 170 < 190 < 150 < 150 77 5 (a,b)

(a) passive equilibrium temperature must be < 70 K to demonstrate active control can be achieved. (b) gradients within any one detector must not exceed 0.5 K, and gradients between any two detectors must not exceed 0.5 K. Table 1. Summary of internal temperature requirements.

Item Window Cryostat above Cassegrain rotator interface Cryostat below Cassegrain rotator interface

Temperature requirement, C 0 C Ambient 15 C Ambient Maximum Minimum Maximum Minimum 1.5 -4.6 16.5 10.0 1.5 -3.0 16.5 12.0 1.5 -5.0 16.5 10.0

Table 2. Operational temperature requirements for camera external surfaces.

3.2 Summary of thermal design The design principle is to thermally isolate the camera internals from the warm cryostat vacuum vessel. In general, internal components viewing the warmer cryostat have a low emissivity finish to minimise radiative heat transfer, whereas surfaces viewing the detectors are black-painted to minimise stray-light. The external surfaces, however, must be well-coupled to the ambient environment to minimise temperature differences. The key drivers for the thermal design are the strict temperature requirements defined in Sec. 3.1. The cryostat is cooled by three two-stage Leybold 5/100T cryo-coolers. Analysis has shown that requirements are very nearly achieved with only two coolers, therefore the third cooler provides a considerable margin against uncertainties in heat-lift. The first stage of each cooler is coupled to the optical bench by high-purity copper links. Considerable design effort was required to develop a link design that allows for contractions during cooldown and is also relatively straightforward to assemble. The second stage of each cooler is coupled to the detector thermal plate (see below) by copper straps. The external surface temperature of the cryostat is maintained by software controlled heaters (~40 W maximum). It is decoupled from the cold internal components by virtue of a low-emissivity internal finish and polished stainless steel radiation shields, which form a second skin within the cryostat. The edge of the window is fitted with a kapton film heater (~90 W maximum) to provide additional heat to prevent misting. Attached to the optical bench is a liquid nitrogen heat exchanger, used for initial cooldown of the camera from ambient temperature, and thermal links to the cryo-coolers. A heater is also fitted for warm-up of the cryostat. To minimise the parasitic heat loads the optical bench is mounted to the cryostat using a series of low-conductance glass fibre reinforced plastic (GFRP) trusses which restrict the heat flow from the structure to about 1 W. Baffle components are manufactured from aluminium and all interfaces are required to be firmly bolted to ensure that temperature requirements are achieved. The baffles absorb stray-light in the near infrared, however, as they are cold,

they must have a low emissivity in the thermal infrared to decouple them from the window. Therefore the windowfacing surfaces are covered with a selective absorber. The opposite faces are black-painted to absorb stray light. The detectors, and molybdenum plate to which they are mounted, are conductively isolated from the rest of the internals by the three titanium flexures. Thermal straps connect each detector to a separate, high-conductivity, thermal plate that is cooled via straps to the second stage of the cryo-coolers and controlled via a 10 W heater.

3.3 Models Detailed thermal mathematical models of the camera have been constructed using the European Space Agency standard software ESATAN and ESARAD6. ESATAN uses a thermal network representation. Nodes are defined and can be allocated a thermal capacitance to enable transient calculations. The thermal network is established by specifying conductive, radiative and convective couplings between nodes. The user may incorporate further routines to describe, for example, the operation of the cryo-coolers, or fluctuations in power dissipation. The radiative couplings were calculated using ESARAD, a geometric modelling package.

3.4 Predictions The steady-state predictions for the nominal operating conditions are summarised in Table 3. Fig.3 shows the ESARAD geometric model with temperatures overlaid for the 0 C ambient case. All temperature requirements are met with substantial margin, demonstrating that the camera thermal performance is compliant with the technical specifications. A sensitivity study was undertaken to assess the impact of key parameters on temperature predictions of items without active temperature control. With the exception of the baffle assembly, the uncertainty was predicted to be within 10 C. The maximum uncertainty in the baffle temperatures was -12 C / +22 C and is due to the large number of bolted interfaces. The uncertainty is lower for baffles closer to the interface with the optical bench.

Fig. 3. Predicted temperature profile within the cryostat in the 0 C ambient case (temperatures in C).

Temperature Prediction 0 C Ambient 15 C Ambient Maximum Minimum Maximum Minimum Internal Items, K Baffles Lens assembly Filters Detectors(a) External items, C Window outer surface Cryostat above Cassegrain rotator interface Cryostat below Cassegrain rotator interface 147 94 81 48.6 1 0 0 108 80 77 48.5 -3 -1 -1 169 108 91 43.5 16 15 15 123 90 85 43.4 11 14 14

(a) To demonstrate that the detector can be cooled below 70 K, the detector plate heater is inactive. A further simulation has shown that the detectors may be driven to about 90 K by applying the 10 W heater, demonstrating that detectors can be controlled within requirements. Table 3. Raw temperature predictions for the nominal operating cases.

STRUCTURAL ANALYSIS

4.1 Load cases The structural requirements for the IR camera required verification by analysis for a number of different loading states. The results of interest included flexion of the focal plane under different conditions, stresses and strains in structural components, and interface loads for the sizing of bolts. A summary of the considered effects and resulting load cases is given in Table 4. The load cases are divided into four categories: operating, short term accidental loading (STAL), survival loading and flexion.
Short Term Accidental Loading 6 7 8 9 10 11 Survival Loading 12 13 14

Operational Loading Structural Considered Effects 1 2 3 4 5

Flexion 15 16 17

Atmospheric and internal X X X X X X X X X X pressures loads Gravity at 0, 45, 90 X X X X X X X X degrees altitude angle X X X X X X X X X Operating thermal Cool-down / warm-up X non-equilibrium thermal X Cryogenic failure thermal X X Earthquake OBE X Earthquake MLE X Slewing decelerations X Survival wind speed X X Handling X Transport Gravity at 20, 45, 88 X degrees altitude angle Table 4. Summary of applicable load cases for each of the 17 load cases, the considered effects are marked with an X.

Atmospheric pressure loading was taken at sea level as 101,325 Pa. This pressure load will be seen during testing activities at RAL, however in operation, atmospheric loading will be about 75% of this value. Gravitational acceleration was taken as 9.81 ms-2. An Operating Basis Earthquake (OBE) is defined as an earthquake of moderate size but with a high probability of occurrence during the lifetime of the observatory. A maximum likely earthquake (MLE) is an earthquake with a large magnitude but a lower probability of occurrence. Some data on the levels was available in the form of ground-response acceleration spectra for the region. Slewing decelerations are rotational decelerations which come about during braking of the telescope. Handling and transport loads involved loads around special lifting features in the camera, and potential drop-loads during transport. 4.2 Structural requirements Margins of safety on failure for static loads must be above zero for compliance and were calculated as: MOS = (maximum allowable load) (actual applied load safety factor) -1 (1)

Safety factors were required on ultimate tensile strength (UTS), yield and fatigue strengths as given in Table 5. For slipping at bolted friction-grip interfaces, the yield safety factors were used. Operational Loading Short Term Accidental Loading Survival
Table 5. Safety factors.

UTS 4 4 2.5

Yield 1.5 1.5 1.2

Fatigue 2 N/A N/A

4.3 Model philosophy The cold mass and cryostat vessel were de-coupled for ease of analysis. Both structures are similar in terms of the modelling complexity required, and structurally the only link between the two is the series of eight thermally insulating GFRP trusses. For the cryostat model, the cold mass was treated as a point mass and from the cold mass point of view, the cryostat was a rigid body. 4.4 Model generation The Finite Element Analysis (FEA) package used was ANSYS 6.17. An import/export feature between ANSYS and Pro Engineer was used. Both models were necessarily complex, involving a combination of solid, shell, beam and point mass elements. The generation of geometry from scratch in ANSYS would be too time consuming, and would give rise to possible nonconformances with the CAD model. Conversely, the Pro Engineer model was far too complex to easily de-feature, and would have necessitated much re-assembly work to ensure that the removal of unwanted features (i.e., tapped holes and small fillet radii) did not alter the overall assembly geometry. Instead, a hybrid approach was used for generating the initial ANSYS geometry. Complex 3D solid parts were defeatured in Pro Engineer and directly imported to ANSYS for assembly and meshing. For thin-walled structures, an approach was taken to build simplified volume models in CAD using primitive geometric entities. The exteriors of these CAD solids were modified slightly so they would lie in the mid-plane of the resulting shell elements (see Fig. 4). In areas where shell elements (with six degrees-of-freedom per node) were attached to solid elements (with only three degrees-of-freedom) per node, a thin mesh of shells was generated over the solid elements to transmit the rotational degrees-of-freedom and avoid hinge-type effects. A similar strategy was used with beam elements.

Material properties were input over a range of temperatures from room temperature to absolute zero. For most relevant properties, such as Youngs Modulus and Poissons Ratio, there was little change over this range. Thermal contraction, however, varies appreciably for many of the materials used, so had to be defined as a series of data points in the model.

Original CAD model

Solid model generated in CAD, with exterior surfaces at midplanes of true geometry

Final meshed FEA model

Fig. 4. Generation of FEA shell geometry from CAD model, with FEA results leading to further design iteration.

4.5 Loading Boundary conditions for the outer cryostat were defined simply by constraining a circle of nodes around the Cassegrain rotator interface flange corresponding to mounting bolt positions. These constraints were moved to the lifting point interface locations for analysis of handling loads. For earthquake loads, it was decided that a quasi-static acceleration would suffice if it could be demonstrated that there was little or no dynamic response of the structure below a certain frequency (thereby avoiding any coupling with

telescope and building resonances). Gravity was applied combined with these, giving a resultant acceleration vector. There were a large number of permutations when combining different gravity vectors required for analysis (due to changing telescope orientation) with an effectively arbitrary choice of direction for assumed quasi-static earthquake accelerations. To simplify this, four harsher load cases were identified which enveloped the possible loads. This was done by finding the maximum possible resultant acceleration magnitude, and applying it in three mutually orthogonal directions simultaneously. This approach yields substantially conservative results. Temperature loads were taken from the thermal analysis model. Average temperatures for the various sub-assemblies were applied at several nodes around the centre of the sub-assembly in the FEA model. For the baffle tube, maximum and minimum temperatures were used at the top and bottom of the baffle tube respectively. Then an analysis was run in ANSYS which generated interpolated temperatures at every other node in the model. The output from this run can be used as a load input for further analyses. Other loads analysed were air pressure and wind loading on the exterior of the cryostat vessel, handling loads (where the camera is supported at certain handling points), transport loads, and rotational decelerations. 4.6 Solution To simplify the running of a large number of analyses, and to help file management and archiving, ANSYS loadsteps were used. A loadstep is a particular loading case for the FEA model defined in a separate file. A number of loadsteps can be solved sequentially in a single run, with all results being written to a single file. 4.7 Results In general, stresses in the outer cryostat vessel were dominated by atmospheric pressure while stresses in the cold mass were dominated by thermo-elastic effects (see Table 6).
Detector plate and Pin OB structure Electronics box Al 0.4 0.3 0.3 0.4 0.2 3.5 3.5 3.4 3.4 3.4 3.7 11.7 18.1 17.6 21.5 1.7 2.3 0.6 0.5 0.4 0.7 0.9 25.4 25.3 25.3 25.3 25.1 25.3 FPA frame and subframe Baffle tube WFS Plate Cryostat Vessel Structure Al 58.8 9.5 10.3 10.5 9.9 10.7 16 58.7 58.6 60.4 58.9 59.2 56.3 0.1 0.4 FW Hub Al 1.8 1.5 0.6 1.3 0.7 6.8 6.8 6.7 6.7 6.3 6.1 -0.6 0.0 Thermal plate Cu 12.7 11.7 8.8 14.2 8.7 548 543 546 551 548 553 0.8 2.9 Flexures Ti 7.4 13.8 19.7 15.9 22.6 71.4 71.6 71.1 70.8 55.4 48.6 0.1 0.3

Material: Loading: Atmospheric Gravity -

Al 0.3 0.7 0.7 0.8 0.8

Al

GFRP 19.2 15.5 7.4 13 7.2 74.4 70.6 74.6 80.3 74.1 80.2 -0.1 1.5

Trusses

Baffles

Al -

Al 0.8 0.8 1.1 0.7 0.6 11.9 11.9 12 12.1 11.9 12 4.7 5.9

Al 3.1 2.4 3.1 3.6 4.9 62.1 62 62 62.1 62.1 62.1

Mo

0.6 1.3 1.4 1.3 1.4 2.1 2.4 2.4 2.4 2.8 2.6 23.6 28.8

38.3 12.5 17.4 14.9 20.5 49.5 50.6 50.8 50.3 50.2 49.4 0.4 0.6

Operating thermal Combined

1.6 1.6 1.6 1.6 1.6 1.6

Handling MOS on UTS MOS on yield

42.0 51.1

Table 6. Von Mises stress results in MPa in different parts of the model under operational loading, along with margins of safety on Ultimate Tensile Strength (UTS) and yield strength.

For operational loading results, there were initially negative margins of safety at the GFRP trusses and the titanium flexures. These are both areas where the accommodation of large deflections is necessary, without compromising stiffness. A low but positive MOS can therefore be seen as a sign that the design is near-optimal in meeting these conflicting requirements. In the case of the titanium flexures, it was shown that halving their thickness would halve stress levels on cooldown to an acceptable level, without impacting on tilts or translations at the FPA appreciably. Similarly, it was possible to reduce the stiffness of the GFRP trusses in the radial direction (the direction of large deflection) and maintain compliance with natural frequency and deflection requirements. For the STAL and survival load cases, margins of safety were positive except for several of the earthquake scenarios. In these particular cases, the earthquake acceleration state was decomposed and more realistic loads were applied. The resulting stress levels were acceptable (see Sec. 4.5). Results for translations and rotations at the focal plane were fed into the optical image quality analysis. Other results of interest were reaction loads at mounting positions, which were used in sizing mounting bolts. Modal and buckling analyses both gave results compliant with requirements. 4.8 Submodels There were a number of areas which required special analysis; one such area was the cryostat window. Being a brittle material, failure would be through crack propagation in areas of tension in the material. A seating arrangement was designed whereby silicone rubber pads prevent glass-to-metal contact occurring once the sealing O-ring had been compressed. This problem was examined using a 2-dimensional axi-symmetric model with contact elements allowing sliding with frictional forces between glass and seal (see Fig. 6).

Pressure

Window

O-ring Silicone rubber pad

Aluminium seat

Fig. 6. Contact analysis of O-ring seal at window, showing silicone rubber pads preventing glass-to-metal contact.

CONCLUSION

The VISTA IR camera is a complex design which has required an iterative approach between CAD, structural FEA and thermal analyses. The design process has benefited greatly from the specialised software packages available, as well as expert users. A number of technical challenges have been overcome through close cooperation between the relevant disciplines. The bulk of the mechanical design work on the camera has been performed at RAL using powerful CAD tools. Throughout the design phase, careful attention has been given to assembly and integration issues, as well as the mechanical effects of the low operating temperatures. The thermal design has matured with the mechanical design, and requirements on thermal flux at the focal plane have been met. Finite element analysis has also been an aspect of the design process. Requirements of structural integrity under many environmental conditions have been verified through analysis, and movements of optical components have been assessed. Having been verified successfully by analysis, the design has recently passed a final design review, and the project has now entered its manufacturing phase.

ACKNOWLEDGEMENTS
The VISTA IR team gratefully acknowledges help and advice received from University of Durham and the UK Astronomy Technology Centre. VISTA is funded by a grant from the UK Joint Infrastructure Fund, supported by the Office of Science and Technology and the Higher Education Funding Council for England, to Queen Mary University of London on behalf of the 18 University members of the VISTA Consortium of: Queen Mary University of London; Queen's University of Belfast; University of Birmingham; University of Cambridge; Cardiff University; University of Central Lancashire; University of Durham; University of Edinburgh; University of Hertfordshire; Keele University; Leicester University; Liverpool John Moores University; University of Nottingham; University of Oxford; University of St Andrews; University of Southampton; University of Sussex; and University College London.

REFERENCES
[1] [2] [3] [4] [5] [6] [7] Emerson, J. P. et al., The Visible and Infra Red Survey Telescope for Astronomy: Overview, Proc SPIE Int. Soc. Opt. Eng 4836, p.35-42 (2002). McPherson, A. et al., The VISTA Project: a Review of its Progress and Lessons Learned Developing the Current Programme, Proc SPIE Int. Soc. Opt. Eng. 5489-46 (2004). Dalton, G. et al., The VISTA IR Camera, Proc SPIE Int. Soc. Opt. Eng. 5492-34 (2004). Parametric Technology Corporation, Pro Engineer, http://www.ptc.com/ Caldwell, M. et al., Aspects of Concurrent Design During the VISTA IR Camera Detailed Design Phase, Proc SPIE Int. Soc. Opt. Eng. 5497-06 (2004). Alstom Power Technology Centre, Software Products - Overview, http://www.techcentreuk.power.alstom.com/ ANSYS, Inc., ANSYS FEA software, http://www.ansys.com/