S. C. West, R. H. Nagel, D. Harvey, A. Brar, B. Phillips, J. Ray, T. J. Trebisky, R. Cromwell, and N. J. Woolf (Steward Observatory)
C. Corbally and R. Boyle (Vatican Observatory Research Group)
D. Blanco (National Optical Astronomy Observatories)
L. Otten (U. Seattle)
Proceedings of SPIE conference on Optical Telescopes of Today and Tomorrow, 2871, 1996
1.0 Abstract
2.0 Facility
3.0 Telescope Optics
4.0 Primary Mirror Cell
5.0 Secondary Mirror Positioner
6.0 Mount and Telescope Control System
7.0 Optical Alignment
8.0 Wavefront Sensing
9.0 Imaging
10.0 Acknowledgments
11.0 References
1.0 Abstract
The Vatican Advanced Technology Telescope incorporates a fast (f/1.0) borosilicate honeycomb primary mirror and an f/0.9 secondary in an aplanatic Gregorian optical configuration. We provide a brief technical and performance overview by describing the optical layout, the primary and secondary mirror systems, and the telescope drive and control system. Results from a high resolution wavefront sensor and a current wide-field image taken at the f/9 focus demonstrates the overall fine performance of the telescope.
KEYWORDS: telescopes, mirror supports, thermal control, wide-field imaging, wavefront sensing, servos.
2.0 Facility
The Vatican Advanced Technology Telescope (VATT) is a cooperative effort between the University of Arizona Steward Observatory (UA) and the Vatican Observatory Research Group (VORG). Located atop Mt. Graham at 3.2-km (10,500-ft) altitude, it serves as a demonstration of technology planned for larger telescope projects by the UA and other collaborators. A photograph of the facility taken in December 1994 is shown in Figure 1.
The VATT's construction and subsequent operation were funded by generous contributors to the Vatican Observatory Foundation, a U.S. incorporated tax-exempt foundation. Its two components, the Alice P. Lennon Telescope and the Thomas J. Bannan Astrophysics Facility, were named after principal benefactors. The University of Arizona Steward Observatory contributed the primary mirror which is one quarter of the project's cost. M3 Engineering (Tucson, AZ) designed the facility.
3.0 Telescope Optics
The Vatican Advanced Technology Telescope (VATT) is an aplanatic Gregorian (AG). The optical prescription is shown in Figure 2. Both the primary and secondary mirrors are ellipsoidal. The maximum field of view (15 arcminutes diameter) is limited by field curvature and astigmatism. The Gregorian configuration was chosen mainly because a concave secondary mirror was considerably less expensive to test than a convex mirror. The 560-kg 1.83-m diameter honeycomb borosilicate primary mirror was manufactured at the Steward Observatory Mirror Laboratory. It was spun cast to achieve f/1.0 curvature of the faceplate and was the first mirror to be polished using the stressed-lap technique.1-4
The VATT secondary mirror is an f/0.9 ellipsoid on a 0.38-m diameter single-arch Zerodur substrate that weighs 30.6-kg. It is has a simple central support via a flanged central hole.5, 6 The mirror was generated and polished by the Space Optics Research Lab (Chelmsford, MA).7
The final interferogram for each mirror is shown in Figure 3.
The relatively stiff 1.83-m diameter mirror has relaxed support requirements compared to larger mirrors. The primary mirror cell is shown in Figure 4
The positions of the 36 pneumatic axial actuators and 16 elevation-oriented passively counterweighted lateral actuators were determined from finite element models.8,9 The axial actuators are placed at the honeycomb rib intersections and the lateral actuators are attached to invar bars glued to the edges of the mirror front and back plates. The inner 12 actuators have peak forces of 112.2-N, the middle 12 actuators have 120.7-N peak forces, and the outer 12 have 225.1-N. The corresponding mirror surface deformation determined from the finite-element models is 5.4-nm rms (30-nm p-p) at zenith and 5.8-nm rms (40-nm p-p) at horizon pointing. The actual implementation however contains 2 unique axial force actuators (inner 24 vs. outer 12) with area ratios of 2 connected to a common pressure head that is further divided into 3 radial sectors in order to control mirror tilt. Each sector contains a stiff hardpoint that provides a positioning reference that incorporates both axial and lateral high-force break-aways and an axial force measuring load cell for servo control of the pressure head.
The main functions of the primary mirror thermal control are to eliminate figure distortion by maintaining temperature uniformity of the mirror and to force the mirror to track ambient temperature to avoid local seeing effects. These requirements are achieved through forced air convection of the inside of the honeycomb structures.10, 11 The specifications for the thermal control system are:12
Figure 4 also illustrates the air flow for a single VATT ventilator unit. A fan (Patriot PD48B2) pulls air out of the return plenum and pressurizes the input plenum. A heat exchanger (Lytron 6210-G1) isothermalizes the air and brings it to ambient temperature. The air is then forced into the honeycomb cells through backplate perforations via 82 air nozzles at an air pressure of ~30-Pa across each nozzle. Eight ventilators are evenly distributed about the cell perimeter, and provide 7-liters/sec of air volume to each honeycomb cell.
A schematic of the thermal system is shown in Figure 5. The off-board chiller (Neslab HX150) supplies a water-glycol mixture to the on-board heat exchangers for removal of heat from the cell weldment, mirror, and ventilation fans. It is located in a thermally managed room 12-m below the telescope. The chiller is capable of removing 0.5-kW of heat even at low temperature extremes of - 15 C ° which is adequate to slew the cell and mirror by 1 C °/hr. The on-board plumbing uses multiple ``T'' connections to balance the flow rate through the ventilators. A Neslab CP55 pump provides 30-L/min (8-gpm) liquid circulation. The chiller is programmed by a VxWorks computer instrumented with thermocouples reading dome ambient (target) and mirror temperature (feedback).
5.0 Secondary Mirror Positioner
The secondary mirror must be precisely positioned relative to the primary mirror. To achieve our goal of 0.1-arcsecond rms images, active collimation is essential because: 1) the predicted gravity deflection of the telescope structure (30-) is larger than the collimation decenter requirement, and 2) a 0.1 C ° temperature change causes the 2-m optical support structure to change length by the entire focus error budget (see section 7.0).
Figure 6 shows the 6-axis positioner for the secondary in the ``mirror-up'' orientation. The tripod has a hole in the middle for connection to the central hub support in the mirror. The tripod is positioned with 3 decenter actuators connected perpendicular to each tripod arm and 3 tip-piston actuators connected to the tripod ends. Six lvdts measure lateral and piston displacements. Each drive consists of a stepping motor geared through a harmonic drive and micrometer screw resulting in a decenter resolution of < 0.1 µm/step, a despace resolution of 0.2 µm/step, and a tip-tilt resolution of 0.2 arcsecond/step.
The details of the actuators are also shown in Figure 6. Each arm of the tripod has both a decenter and tip actuator. The figure shows the decenter actuator for the tripod arm sticking out of the page and the tip-piston actuator for the tripod arm to the left. The tripod is pulled up against the three tip actuators by three tension spring preloads. The tip and piston of the mirror is controlled by the extension of the micrometer screw actuator. A PZT is placed in series with the screw to provide high frequency tip-tilt correction of the atmosphere (not yet implemented). Preliminary tests show that the positioner has its lowest tip-tilt resonance near 30-Hz.
The decenter actuator acts through two pivots in a titanium flexure in order to convert axial into lateral motion. The flexure is clamped tangent to a tripod arm. A compression spring preload keeps the flexure loaded against the actuator screw.
Because of the tight collimation requirements, this positioner must be accurately calibrated. A granite box-parallel was attached to the tripod and instrumented with a kinematic set of contact lvdts in order to find the influence of each actuator on the rigid body motion of the tripod. The result is a 6 x 6 matrix that transforms the desired cartesian motion of the secondary mirror into feedback voltage increments for the positioner.
6.0 Mount and Telescope Control System
The optical support structure and alt-azimuth mount were built by L & F Industries and Paragon Engineering (Figure 2) and allow a 1.8-m telescope to fit within a 7-m diameter dome.13 The 12,000-kg of moving mass is supported on a 12-pad hydrostatic azimuth bearing. Excluding the pier, the lowest resonant frequency is near 19-Hz.
The altitude and azimuth motors are surplused 1.2-m diameter CAT scan direct-drive brushed DC torque motors each capable of producing 1360-Nm of torque (Sierracin Magnedyne). The 80-kHz PWM Copley amplifiers are filtered to be RFI-free. The resulting motion is extremely smooth and free of cogging and periodic errors inherent in gear-driven systems. Both axes use friction-drive incremental encoders (Heidenhain ROD800) to derive both the position and velocity feedback. Combined with 25x interpolators, final resolutions on the sky are ~60-pulses/arcsecond for both axes. The de-rotator uses a precision 1114:1 dynamically-preloaded spur-gear reducer. An incremental encoder provides control loop feedback with 0.11-arcsecond positional resolution. Tracking normally requires 20-Nm of torque and occasionally up to 600-Nm in stiff winds.
Rather than using tachometers, the velocity is obtained digitally in hardware using an LM628-based VME servo card which samples the encoders and updates the command voltages to the DC torque motor amplifiers at a 3-kHz rate. Position-loop control is closed with software in the VxWorks machine updating the 32-bit registers in the LM628 at a 100-Hz rate. A Sony MagneSwitch provides a reference start-up position for each axis. The maximum slew velocity is 8-deg/sec given by the maximum pulse rate of the encoders, but is limited to 2-deg/sec for safety reasons.
The VATT telescope control system (TCS) hardware consists of a single VME Bus incorporating a Motorola VME 147 CPU with Ethernet hardware for communications to a SUN Ultra host computer. Axes control uses a 3-channel Green Spring IP-LM628 servo control card which reads the incremental encoders and closes the three velocity loops. A GMS digital I/O board controls and senses various digital signals (mirror cover, dome position and shutter, fans, subsystem power, elevation axis air brake, and counterweight motors). The dome rotation servo system is handled with a Burr Brown MPV 904 DAC VME-board.
Secondary mirror position feedback is acquired with a 16-bit Acromag 9330 ADC from 6 lvdt sensors. This board also provides access to counterweight and servo amplifier currents used to auto-balance the telescope. An Oregon Micro Systems VME8 stepper motor control board is used to drive the secondary mirror actuators.
Serial communications to low bandwidth intelligent telescope subsystems such as the primary mirror thermal control and the offset guider motion control is provided by a 6-port VME Force serial I/O board. A separate serial port communicates with an external GPS clock that provides accurate date and time for system initialization and produces a precision 1-kHz time base for the TCS software.
Guiding and object acquisition capabilities are handled through an Eltec 8-bit VME video frame grabber that digitizes video from the Photometrics ATC-5 CCD camera attached to the offset guider.
The user interface is provided by a Motif-style windowed GUI displayed on a SUN Microsystems Sparc Ultra main console which also acts as a host computer for the control system's real-time OS, control software, and databases via standard TCP/IP ethernet communications.
Since the TCS is run from a single computer system, Wind River's VxWorks multi-tasking/multi-threaded Operating System was chosen because of its compact and robust real time capabilities. In addition Wind River's Wind-X X-Windows driver software is used to generate the GUI on the host SUN computer. A view of the engineering VATT TCS Run-time GUI is shown in Figure 7
The Servo Control Task is interrupt driven and is responsible for overall control of the mount and the field de-rotator. Every 1/100th-second the control loop takes the user specified right ascension and declination and adds in any commanded bias rates (e.g. tracking comets and asteroids) to construct a new commanded position. This new RA and Dec is transformed into altitude, azimuth and de-rotator angles using the system computed LST. The altitude is then corrected for the effects of atmospheric refraction. The new altitude and azimuth are then corrected for telescope flexure and mount misalignments and converted into commanded positions in encoder counts. These encoder-based positions are then fed to the three axial software Servo position loops diagrammed in Figure 8
As previously mentioned, an LM628 Servo board functions as a digital tachometer. This board runs a standard PID loop in hardware and takes velocity commands from the Servo Control task. The software position loop is a modified PID algorithm incorporating a square root approach for the Type 0 loop and a detachable integrator loop for the Type 1 and Type 2 loops. Gains, switch points, and limits are user definable to accommodate systems of various bandwidths and resonances.
Pointing data runs recently performed using the VATT have enabled accurate modeling of telescope flexure and misalignments The analytic modeling software produced by Patrick Wallace of the Rutherford Appleton Laboratory in the UK called Tpoint was used to map the misalignments and inherent flexures of the system. Study of the mount using this software indicates that despite the nearly 0.26-degree zenithal alignment error, the telescope is capable of pointing to better than 4-arcseconds over it's entire range of motion. In addition, tests performed using unguided CCD imaging confirm tracking accuracy of better than 1-arcsecond over a 7-minute interval. Figure 9 shows the final pointing accuracy after removal of mount misalignment and flexure as modeled by Tpoint.
Currently the VATT TCS is undergoing a major facelift of it's Engineering GUI including the addition of robust autoguiding software which will automatically scan various on-line star catalogs for optimal selection of guide stars, a more compact and ``astronomer friendly'' (rather than ``engineer'' handy) Window and menu system better suited to astronomical observing, integrated network access for smart instrumentation and remote observing, and a more equitable distribution of system control tasking using client software running on the SUN host.
7.0 Optical Alignment
Exacerbated by the increased sensitivity of coma on secondary misalignment for the Gregorian design compared to an equivalent Cassegrain, the very fast optical system leads to a tight collimation and focus error budget. The performance target is for 0.1-arcsecond rms images over the entire 15-arcminute field. This translates into a required positioning accuracy of 2.5- despace, 17- decenter, and 10-arcseconds of tip. Although the on-axis zero-coma condition can be used to relax the decenter and tip constraints, it is not a good idea for a fast telescope with a wide field of view.14-15 As a consequence, we chose the optical alignment procedure to independently null tip and decenter with respect to the optical axis of the primary mirror. The procedure was simple and utilized the fact that the Gregorian prime focus is between the primary and secondary mirrors. A snout was attached to the center hub of the secondary mirror that contained a CCD spaced from the vertex precisely at the prime focus position. The mirror and CCD were attached to a high precision rotary stage and the exact CCD pixel that went through the secondary optical axis was located. Then the CCD was positioned precisely at the prime focus point through direct imaging. Due to extremely large field-dependent coma at the f/1.0 prime focus, the center of the field was readily identified. The snout was removed and the positioner was programmed to rotate the secondary vertex about the prime focus until the field coma was removed at the f/9 focus. This quick and successful procedure insures that the primary and secondary optical axes are coincident.
8.0 Wavefront Sensing
Given the relatively small size and high stiffness of the VATT's optical support structure and primary mirror, a low-resolution wavefront sensor for nightly routine collimation should be unnecessary. Instead we built a high resolution wavefront sensor to be used for annual or semi-annual calibration of collimation look-up tables for the secondary positioner vs. elevation and diagnosis of imaging problems should they arise. The device is shown in Figure 10 and is based on the interferometric Korhonen-Hartmann technique developed at the Nordic Optical Telescope (NOT).16-18
This type of wavefront sensor differs from the traditional Shack-Hartmann design in that it directly measures phase (rather than slopes). Light coming out of the Hartmann apertures is converged toward a common focus. Each adjacent quadruplet of apertures forms a 2-dimensional interference pattern that consists mainly of a small m=0 spot. Spacing irregularities between interference spots are used to calculate the phase difference between apertures. At a certain distance either inside or outside of this common focus, the m=0 interference fringe of one quadruplet of apertures overlaps the m= fringe of the adjacent interference pattern. The detector is located in this plane. In order to minimize contamination from overlapping m=1 fringes on the m=0 fringe position, the aperture size and geometry is chosen so that the m=1 interference fringes are suppressed by the sinc2 envelope of the aperture diffraction.
The NOT instrument incorporates a beam splitter at the telescope focus in order to feed a reference light source into the device. This introduces spherical aberration, and the transfer optics were designed to remove it. We chose instead to place a two-position turret at the telescope focus that selects from a diode laser source or a pinhole. The pinhole is placed into the center of a 45-degree mirror so that an auxiliary camera may be used for direct viewing. As a consequence, the optical design is simpler and off the shelf lenses were used. The analysis software was purchased from Opteon, Inc. (Piikki, Finland).
Typical reference and stellar interferograms are shown in Figure 11 along with a geometric spot diagram obtained shortly after the secondary mirror was installed. Despite virtually no optimization of the telescope or environment, we realized 0.4-arcsecond 80% encircled energy diameter images from the telescope optics once the atmosphere was removed. Due to other priorities, the wavefront sensor has not been used during the past year, so accurate collimation look-up tables have not yet been produced.
9.0 Imaging
Figure 12 shows a wide-field image taken with the VATT and field-flattened 2k x 2K CCD.
10.0 Acknowledgments
We acknowledge the excellent work of the Steward Observatory technical division, the Steward Observatory Mirror Laboratory, the Mount Graham International Observatory crew led by John Ratje, and the efforts of Buddy Powell, George Coyne, and Peter Strittmatter.