3.1

Optical performance

Fig. 3 and fig. 4 show that the camera’s predicted performance is close to the diffraction limit. The wavefront has a peak-valley (PV) value of λ/3 with an MTF of 47%, at the Nyquist frequency of 71.4 lines/mm.

Fig. 3.

Predicted on-axis wavefront at 633nm.

Fig. 4.

On axis MTF at 633nm

3.2

Sensitivity and tolerances

Table 1 shows the sensitivity of the optical model wave front to arbitrary individual mirror perturbations. Clearly, the system performance is dominated by the stability of M1.

Table 1.

Optical system sensitivity

The complete end-to-end tolerance analysis is too detailed to include here, but the final allocated budget covers the effects on performance from the following areas:

The tolerance analysis was modelled using a Monte Carlo approach, resulting in a predicted in orbit MTF of 0.390±0.096. The goal of the tolerancing was to allocate a substantial part of the budget to uncompensatable errors within the mechanical system.

As a design aim, the structure must maintain the optical components within ±10μm and ±10” of their locked positions.

3.3

Alignment

An assembly strategy based on optical and mechanical tolerances alone is insufficient to produce an aligned camera by itself. With six degrees of freedom for each of the mirrors it is probable that any initial build configuration would be closer to any one of a number of local wavefront error minima rather than the overall global minimum. Consequently, the alignment strategy developed is based on fixing M1 to the structure then aligning M2 and M3 with respect to M1. Recently developed techniques are employed using high accuracy co-ordinate measuring machines (CMM), allowing optical alignment to start from a position closer to the final target. The final optical alignment is achieved using positive feedback between the interferometer and optical design code.

4.1

Primary structure

Stability requirements dictate the need for an optical bench that will remain dimensionally stable throughout the mission. To simplify the design, the structure incorporates low CTE (coefficient of thermal expansion) materials; rather than a compensating system.

The change in length of the instrument must remain within 10μm over the distance between M1 and M2 of 0.58m. The worst case temperature difference between the alignment environment and the predicted temperature during image acquisition is 10°C (ΔT_max). The maximum coefficient of thermal expansion (CTE) required to achieve this is given in Eq. 1.

The bulk of the structure is manufactured using carbon fibre reinforced plastic (CFRP). This material has high stiffness to mass ratio and is capable of achieving near zero CTE through careful lay-up selection.

Collaborating with composite manufacturers has enabled selection of a fibre / resin system and lay-up meeting our requirements (table 2). CTE measurements on a sample panel have shown that the material has adequate thermal stability [1].

Table 2.

Predicted Laminate Engineering Properties

A cyanate-ester resin system was chosen to minimise moisture expansion effects. An estimate of the change in length of the camera due to these effects is given in Eq. 2, using the coefficient of moisture expansion (CME).

This figure estimates the likely change in length between full saturation and no moisture content. In practice, the relative humidity environment of the camera is likely to vary between 0 (in vacuum) and 50% (exposure to ambient environment for alignment activities, vibration tests, etc) [2].

Fig. 5 shows the design of the optical bench primary structure, optimised through extensive finite element analyses. All of the structure panels are composed of CFRP skins (as specified in table 2), surrounding a 20mm thick aluminium 5052 honeycomb core. A fixed titanium support is mounted in conjunction with two titanium flexures to minimise deformations in the structure that could affect system performance.

Fig. 5.

CAD model of primary structure

In addition to structural analyses, the finite element model has been used to assess the impact of thermoelastic and mounting plane deformation effects (table 3). For example, a deflection applied to the flexures simulated linear thermal expansion of the satellite. The resulting translations and rotations of the mirrors were then analysed and directly imported into the optics model, so that optical performance could be evaluated. This cycle enabled the compliancy of the titanium flexures to be optimised.

Table 3.

Summary of Structural Analyses

The structure has been designed to be both mass efficient (total ~12 kg) and stiff enough to avoid dynamic coupling with the spacecraft under the launch environment. Fig. 6. shows the predicted fundamental frequency torsional mode shape.

Fig. 6.

Fundamental frequency mode shape

The panels are manufactured individually with core-filler and woven carbon fibre blocks used around areas that need to be strengthened (e.g. where the fastener inserts are located). The completed panels are assembled around a forming tool and bonded together using epoxy adhesive. The panels were joined together using a propriety technique developed and patented by QinetiQ.

The principal challenge in manufacturing the structure is achieving critical dimensions and tolerances between the mirror mounting surfaces. The main example is the distance (719.5mm) and parallelism (0.1mm) between M1 and FPA, each mounted on three aluminium inserts. In order to achieve this, important features such as inserts and mounting holes are accurately machined after final structure assembly (fig. 7).

Fig. 7.

Flight model primary structure

4.2

Mirror mounts

The mechanisms and techniques used to mount, align and finally lock the mirrors in position have been realised through an extensive testing and development program.

M1, the largest mirror (4.5kg), is bonded in position using three invar inserts potted into the rear surface, through injection of epoxy adhesive (Fig. 8). These inserts are joined to an Invar mount, incorporating flexures to compensate for thermal expansion effects and non co-planar mounting flatness. Designing the flexure geometry has been challenging, given the conflict between required compliancy and adequate stiffness to survive launch. The Invar mount is joined to the structure using a combination of fasteners and dowel pins.

Fig. 8.

Rear of M1 insert

M2, the smallest mirror, is bonded into an invar cell using RTV elastomer (fig. 9). The supporting cell is adjusted in 5 degrees of freedom (3 translations, 2 rotations) through the alignment process. Fig. 10 shows the spherical joint mechanism designed to both adjust and effectively lock the mirror following alignment. This is achieved through a combination of clamping, adhesive injection and insertion of taper pins. M3 is bonded in a similar manner to M1 and is also adjustable in 5 degrees of freedom.

Fig. 9.

M2 mirror cell

Fig. 10.

Rear of M2 mount in camera

The mount designs evolved through individual tests (fig. 11 shows an M2 test), devised to verify survivability and stability through representative launch loads. These vibration levels were derived using dynamic finite element analyses. Movements of the mount components through vibration were assessed by measuring the positions of tooling balls with a CMM.

Fig. 11.

M2 mount vibration test

6.1

Primary structure conditioning

Prior to alignment, the structure was baked at 60°C under vacuum to remove volatiles and dry out moisture from the CFRP skins. Several thermal cycles (-20 to +50°C) were also applied, to reduce the effects of microcracking [2]. Finally, the structure was allowed to settle under vibration, to minimise susceptibility to movement during subsequent vibration tests.

6.2

Mirror Manufacture

Due to cost considerations, traditional polishing techniques were chosen for mirror manufacture. Although the mirrors were thinned, no other lightweighting methods were used. The mirrors were successfully polished to <λ/15 PV surface form error. The design was re-optimised with the M2 parameters as compensators, following completion of M1 and M3.

The measured optical parameters, radii of curvature (R) and conic constants (k), were input into the optical model and mirror separations re-optimised.

Fig. 13.

Completed mirrors

6.3

Mirror metrology

After the off-axis segments were machined from their parents, the position of the poles and axes were no longer well defined. To recover the accuracy, the optical surfaces were characterised with a CMM, deriving axes and poles with respect to tooling balls bonded to the mirror edges.

6.4

Initial mechanical build

The system was assembled, with the mirror positions determined solely by mechanical tolerances and adjusters set at their mid-positions.

At first, a full pupil interferogramme could not be displayed due to severe aberrations. However, after a simple adjustment, the interferogramme in fig. 14 was obtained.

Fig. 14.

Interferogramme after initial assembly

6.5

Mechanical alignment

Using accurate CMM measurements of the tooling balls on the three mirrors, their positions were adjusted so that they were within 25μm of their required positions (fig. 15 and fig. 16). It was expected that this process would reduce the pupil wavefront to a few waves, thus allowing interferometric alignment to commence from a position close to the global minimum.

Fig. 15.

Mechanical alignment

Fig. 16.

Interferogramme after mechanical alignment

6.6

Optical alignment

Extensive modelling of an alignment technique using feedback between an interferometer and optical software showed the system should converge to the expected nominal wave front within only a few cycles.

However, in practice, it quickly became apparent that the system was not converging as predicted. Consequently, alignment continued using visual analysis of the interferogramme only. Fig. 17 shows the best alignment achieved. Analysis of this wavefront showed mirror deformations that could not be compensated for by further adjustments.

Fig. 17.

Aligned camera interferogramme

6.7

Locking

The mirror mounts were locked and adjusters removed in a carefully derived sequence so as to minimise further misalignment. A setting jig was used to determine the position and alignment of the image plane within the structure. With this knowledge, the CCDs were aligned within the FPA (fig. 18) which was subsequently shimmed into focus.

Fig. 18.

Focal Plane Assembly (FPA)

7.1

Vibration

Final vibration tests were conducted using the flight camera and spacecraft structural model (fig. 20). The instrument behaved as predicted through launch representative vibration levels. Post-vibration functional tests verified that the operations of all subsystems were within specification. Fig. 21 shows an example of vibration levels measured at the spacecraft and instrument interfaces.

Fig. 20.

Acceptance vibration tests

Fig. 21.

Acceptance vibration accelerometer data

7.2

Thermal vacuum and balance

The multi-layer insulation blankets covering the outer surfaces can be observed in fig. 22. The structure and door mechanism were qualified through survival temperature cycles between -20 to +50°C. At extremities of predicted operational temperature range (-5 to +36°C), functionality of the FPA and door actuator were demonstrated. Thermal balance tests validated the thermal design in the expected worst hot and cold case environments. These tests provided data to refine the thermal mathematical models and verify temperature gradients. Based on this, the models will then be used to provide in-flight support.

Fig. 22.

Preparation for thermal balance with MLI