<title>OVLA prototype telescope control system</title>

The OVLA prototype telescope control system O. Lardière a, L. Arnold a, J.P. Berger a, C. Cazalé a, J. Dejonghe b, A. Labeyrie b, D. Mourard c a

b

Observatoire de Haute-Provence, F-04870 Saint-Michel-l’Observatoire, France Laboratoire d’Astrophysique Observationnelle, Collège de France, F-04870 Saint-Michel-l'Observatoire, France c Observatoire de la Côte d’Azur, Département Fresnel, F-06460 Saint-Vallier de Thiey, France

ABSTRACT A prototype telescope for the Optical Very Large Array (OVLA) project is under construction at the Observatoire de HauteProvence (OHP), France. The OVLA will be a long-baseline optical interferometer of 27 mobile 1.5m-telescopes. In 2000, the functioning of the OVLA prototype telescope will be tested outdoors alongsite the two other telescopes of the GI2T (Grand Interferomètre à 2 Télescopes of the Observatoire de la Côte-d'Azur, France) to form a 3-telescope interferometer (GI3T). Firstly, we briefly present the design of this telescope hightlighting its unusual characteristics, which include a spherical mount and a thin active primary mirror. We had to study a specific control system for driving mount and for the active mirror cell. Hardware and software design of these two systems are also presented, as well as some test results. Lastly, we propose a complete electronic architecture for the fully equipped OVLA prototype telescope. The telescope system is partitioned into elementary distinct subsystems each controlled by a small embedded calculator linked to each other by an addressable serial bus. With this kind of architecture, the telescope is fully autonomous. Thus the future installation of the OVLA prototype telescope at the GI2T site should be easier, as well as the installation of a large interferometer such as OVLA where 27 telescopes are expected. Keywords: interferometers, telescope design, tracking, active mirror.

1. INTRODUCTION The Optical Very Large Array (OVLA) will be a ground-based aperture synthesis interferometer for high resolution imaging in visible wavelengths (Labeyrie et al. 1991, Labeyrie 1997). With 27 telescopes of 1.5 meter forming a "dotted ring" aperture of 600m, the interferometer is expected to provide snapshot images using an iterative phasing algorithm (Pedretti et al. 1997). The telescopes will have to move accurately during the observation, for full flexibility of array configuration as well as for keeping a zero optical path difference between the 27 beams, this constraint will imply compact and light telescope mounts. A prototype telescope, being built at the Observatoire de Haute Provence (OHP), has unusual characteristics: a spherical mount and a thin active primary mirror made of ordinary glass (Dejonghe et al. 1998). The spherical mount also has other significant advantages: low thermal inertia, no vibrations, low wind buffeting, including that few reflections are needed for the coudé beam (Labeyrie et al. 1988, Plathner 1988). The complete version of the OVLA interferometer will contain at least 27 telescopes, a large number of subsystems and a huge data flow between the different subsystems are expected. To ensure a proper simultaneous running of such a telescope array, it is necessary to have fully autonomous telescopes controlled through a hierarchical communication network. The control system of the prototype telescope under construction is designed in this way. This prototype telescope should be finalised at the OHP in 1999, and will join the Grand Interféromètre à 2 Télescopes (GI2T) (Mourard 1994a) to provide a third element in 2000. In this paper, we describe the OVLA prototype telescope design and propose a general electronic architecture for each telescope of the interferometer. A presentation of the spherical mount and the active primary mirror control system will be accompanied by some preliminary test results and performances.

2. THE OVLA TELESCOPE DESIGN

Figure 1: The OVLA telescope.

The OVLA prototype telescope shown in figure 1, is a 1.52m-gregorian telescope sending an afocal coudé beam towards the central lab with a magnification of 20. For the purposes of compactness and lightness, an 1 inch thin meniscus-shaped primary mirror (M1) with a focal ratio of f/1.7 is housed in a 2.8m spherical mount. The primary mirror is supported by 32 actuators to compensate its flexures (Arnold 1997, Dejonghe et al. 1998). A CCD camera equipped with two interchangeable lenses is placed near the primary focus to get extrafocal and intrafocal images needed for the wavefront analysis. A parabolic secondary mirror (M2) is installed on a movable support to correct the focus and the coma due to misalignment errors. A plane tertiary mirror (M3) can have a variable inclination to reflect the afocal beam exactly towards the central hub of the interferometer through a 100° meridian slit, for all possible pointing directions. With this design, only stars having zenithal distances lower than 50° are visible. To track the sidereal motion, the spherical mount is supported by six barrelshaped wheels, three of which are motorized (see figure 2). The motorized wheels are distributed under the sphere at 120° from each. On each wheel, only the cental axis is motorized so as to impose only one component of the local velocity vector. Also the rotation of the six perpendical barrels is free. In these conditions, three motorized wheels can drive the sphere for all possible angular motions. Indeed, we will see below that it is possible to track stars by a well suited velocity combination for the three motors. In addition, a dual axis tiltsensor is used as a pointing encoder. If the slit goes in front of a wheel, a solid shutter can move along the slit to obturate a small part of the slit and then ensure a continuous stiff contact of the wheel on the sphere during a long tracking time.

Figure 2: One of the six barrelshaped wheels which carry the sphere.

The fully equipped telescope will be carried by an hexapod translator currently studied at the OHP and at the Laboratoire de Robotique et d'Intelligence Artificielle, Paris. Each of the interferometer telescopes will walk to keep a zero optical path difference without long delaylines. A coarse position of each telescope will be known firstly by a GPS receiver and an absolute tridimensional laser metrology system will also be used to estimate the telescope positions to an accuracy of about 1 µm.

3. THE DRIVING SYSTEM 3.1. GEOMETRY OF THE MOUNT - CALCULATIONS OF MOTOR VELOCITIES Figure 3 shows the geometry of the spherical mount with its characteristic vectors. All motions of the telescope are r interpreted in terms of the instantaneous rotation vector W (Mourard 1988). We utilize the alt-azimuthal frame of the r r r r observatory site (0, i , j , k ) defined in figure 3. During the observation, the telescope optical axis unit vector U must

r

track the star. The whole sky has a rotational movement around the unit vector D directed towards the celestial pole. Ω is r the rotational velocity of the sky. Moreover, the coudé beam carried by the unit vector V must remain perfectly motionless and directed towards the central lab during the observation. This stability is ensured by a rotation of the telescope around its r optical axis U , and by a rotation of the tertiary mirror located at the center of the spherical mount.

Figure 3: Geometry of the spherical mount.

r

We can thus write that the rotation of the telescope is the sum of a rotation around D and a rotation around the optical axis r r r r W = Ω ⋅ kU + D (1) U:

(

)

where k is the angular velocity around the optical axis (in sidereal speed units) needed to stabilize the coudé beam. In addition, we need the vectorial equation ensuring the stability of the coudé beam. Consider the point F of the slit where r r the beam exits the sphere. The beam is stable if the point F remains in the plane (0, U , V ) formed by the optical axis and the coudé beam itself (the tertiary mirror keeping the beam horizontal). The linear velocity of point F must be parallel with r r plane (0, U , V ), implying :

r r r r (W ∧ V ) ⋅ (U ∧ V ) = 0

(2)

r The equations (1) and (2) form a system of two vectorial equations with two unknown factors k and W . Before solving this system, let us consider the motions which will be added to the instantaneous rotation vector when pointing or guiding corrections are needed. First, we can use classical equatorial corrections (right ascension and declination), but another correction axis system seems also useful for the spherical mount pointing and autoguiding. Indeed, contrary to the equatorial corrections, this correction system that we have named (γ, β), is related to the spherical mount geometry. The correction γ is a rotation of the telescope around the axis of rotation of the tertiary mirror. This axis, called r m , is orthogonal to the coudé beam and to the optical axis of the telescope. The β correction is a rotation of the telescope r around the coudé beam V . Figure 4 shows these two correction systems. Note that a slow additive rotation f around the

Figure 4: Two systems of corrections are available for pointing or autoguiding r optical axis U can be used to correct the coudé beam direction without de-pointing the star. Including the (α, δ) and the (γ,β) correction systems, equation (1) gives:

r r r r r r & γ& r W = (1 + α& )⋅ Ω ⋅  D + δ ⋅ D ∧ U + (k + f )⋅ U  + β& ⋅ V + m δ cos sin γ  

(

with :

α& δ δ&

)

(3)

: right ascension correction speed (in sidereal speed units) : stellar declination

f

: normalized declination correction speed (in sidereal speed units) : additive field rotation speed (in sidereal speed units)

β&

: β correction speed (in rad/s)

γ : angle between the optical axis and the coudé beam. γ& : normalized γ correction speed (in rad/s) r r r m = V ∧U For the first interferometric tests, we expect to install the OVLA prototype telescope without hexapod translator at constant r azimuth (West) from the central GI2T lab. According to equations (2) et (3), we can find k and the vector W :  k+ f    r d ⋅ sin l W = (1 + α& )⋅ Ω ⋅  −  cos δ  0  

    cos A l           γ& d − ⋅ cos l  ⋅  B  +  0  + k+ f  cos δ sin γ    C   − sin l  d ⋅ cos l k+ f   cos δ   d ⋅ sin l cos δ

0

− C 0     ⋅  0  + β& ⋅  1   A  0    

d d     C ⋅  sin l + ⋅ B cos l  − A ⋅  cos l + ⋅ B sin l  cos δ cos δ     k= C 2 + A2

r where A, B and C are the components of the vector U directed towards the star of hour angle H and declination δ, for a site located at the latitude l :  A   sin l. cos δ . cos H − cos l . sin δ      − cos δ . sin H   B =   C   cos l. cos δ . cos H + sin l. sin δ     

According to the positions (M1, M2 and M3) and the active directions (u1, u2, u3) of the three motorized barreled wheels, the relation between the instantaneous rotation vector and the three motor rotation velocities V0, V1 and V2 (in rpm) is:   1 0  V0     1 3 = ⋅ − V n  1  2 2 V   1 3  2 − −  2 2

 1  r 1 ⋅ W  1 

Where n is the gear ratio. Figure 5 shows the evolution of rotation velocities V0 and V1 of the M0 and M1 motors (respectively) during the course of one night for various declinations. The velocity curve V2 of the motor M2 can be deduced from the V1 curve by a symmetry around the H=0 axis.

Figure 5: Change in motor velocities for different declinations with respect to time (hour angle). Curves have been plotted only for the possible pointing directions (i.e. z