Goodman Hardware
The SOAR Goodman Spectrograph features one 4096 x 4096 pixel Fairchild CCD. The CCD is read out through 1 amplifier using a Spectral Instruments controller. This CCD has only minor flaws which has little impact on its scientific performance. Figure 2 shows an internal quartz lamp flat for the 300 l/mm setup. Please note the fringing pattern at the red end of the image. There is also an arc-shaped artifact at the blue end of the image due to an internal reflection off the secondary filter wheel.
Figure 2: An internal quartz lamp flat of the 300l/mm setup.
Figure 3 shows the QE of the Goodman CCD.
Figure 3: The QE curve of the Goodman CCD.
The most difficult assembly in the spectrograph is the camera articulation stage. Motion is accomplished by a wormdriven annular stage directly encoded with a resolution of 0.6 μ-radians. The central bearing alone is not stiff enough to meet our flexure requirements for an assembly cantilevered to one side, so the camera platform also rides on a concentric 400mm curved bearing rail from THK. The platform that holds the camera optics and dewar is attached at two points to the central stage and at two points to the bearings on the curved rail. The coupling between the bearing assembly and the camera platform is through tuned flexures that both relieve the overconstraint between the central bearing and the rail, and act as a restoring spring for two piezo-electric actuators that can move the whole platform up and down to compensate for instrument flexures. These flexures are pre-loaded with 100kg of tension, which is more than twice the total weight of the camera assemblies, to insure that the bearings on the curved rail remain on the same contact surface (the underside of the rail) during rotation of the instrument. Figure 5 shows a close-up of the flexure and piezo-electric actuator assembly. Flexure compensation on the orthogonal axis uses the articulation motion at very low speed.
The camera is balanced by a counterweight, so that the torque requirements to move it are not excessive, however, the entire load exceeds the manufacturer recommended inertia for the annular stage by a factor of 10. These ratings assume motion at the full rated speed of 80 deg./s, which is much faster than we require. By limiting the speed to less than 1/10 of this value, the energy dissipated at full stop is well below the maximum rated values. Slewing the stage through its full range only requires around 30 seconds, and is dominated by setting motion near the target encoder position.
The camera optics tube rides on lead-screwdriven crossed roller bearing stages. The camera stage is a custom low profile design that had to be incorporated into the articulation assembly. The camera focus stage incorporates external temperature sensors, constructed from temperature-to-voltage converters that feed built-in analog-to-digital converters in the Silvermax motors driving the stage. The optics mounts do not include passive thermal compensation, so measurements are required to correct for focus changes with temperature.
Even with no shutter in the optical path, it was difficult to meet the space constraints imposed by the camera articulation, and the limited pupil space. The clear aperture at the front of the camera is 4” and it is 2.8” at the last optic, which doubles as a dewar window. No commercial shutter of 3” or larger size would fit in the available envelope, and we preferred a shutter in the collimated beam to eliminate field dependent gradients in the exposure time, so we made our own compact design that fits on the front of the camera optics tube. After several iterations, we settled on a design that adds only ¼” to the width of the camera optics (except for a strategically positioned motor), and adds only 1” in length to the front of the camera. It consists of a friction driven curved stainless plate 0.010” thick that rides in a curved teflon track to cover the 4” entrance to the camera optics. The stepper motor can open or close the shutter in under 200 msec.
We acquired VPH gratings of 300, 600, and 1200 l/mm that are adequate for first light, but not optimum; these gratings are still in use. The 300 and 1200 l/mm gratings scatter more light than we would like, and the 600 l/mm has cylindrical curvature that will degrade the resolution in best seeing. Because of these shortcomings, and frustration with the timescales for iteration with vendors, we decided to launch an effort to construct our own VPH gratings. We constructed a holographic exposure facility that is currently capable of making 4” size VPH gratings. With some effort, we have produced gratings of quality equal to or exceeding those we purchased and are now learning how to make larger sizes and multiplexed gratings. More importantly, we have created software design and testing tools for VPH gratings and spectrographs. These have already been useful in designing gratings for a TMT instrument in collaboration with Epps. Our first priority is to construct a 2400 l/mm grating for the Goodman spectrograph, and then to replace the other gratings with optimized versions. VPH grating production will be the main project of the Goodman Lab once the Goodman Spectrograph is commissioned.
The Grating Rotation and Translation Stages
The grating changer passes through the annular stage responsible for the motion of the articulated camera. It can position any of three gratings at the 75 mm pupil, or lower them out of the way for imaging mode. This translation is subordinate to the grating rotation, so that the grating can be inserted and removed quickly from the pupil without resetting the angle. The rotation is driven by a Newport rotary stage at the bottom and a matching bearing at the top. This stage has been retrofitted with a Silvermax motor because the servo controller and electronics provided by Newport were unusable, and would not have conformed to our RS485 network approach to motor management. The stage is directly encoded with a resolution of 0.9 μ-radians, and the Silvermax motor uses feedback from this encoder for fine position control. Gratings are mounted in frames that are held by ball detents in the translation mechanism.
The 4-inch diameter filters are placed in the collimated beam where they cause a pupil shift instead of a more irritating refocus, but this made them large, to accommodate the 75 mm pupil, and difficult to place. The wheels are suspended from a plate mounted to a cantilevered extension to the truss. The wheels are tilted enough to place any reflection ghosts the filters generate outside of the imaging field. Filters are mounted in rings that are held in the wheels using spring loaded ball detents. This allows exchange of filters without tools or fasteners that get lost or dropped in the instrument. Likewise, the wheels are held on their bearings by a hub that can be removed by hand. The wheels have teeth around their perimeter and are driven by smaller gears engaged by a spring loaded mechanism.
The Goodman Spectrograph collimator has a set position at this time and cannot be moved.
Three major requirements drove the slit changer design, which must house both long-slit assemblies and custom slitmasks. We wanted to maximize the number of slit frames the changer could hold, maintain simplicity in the exchange mechanism, and implement rapid alternation between through-the-slit imaging and unobstructed imaging, without losing slit registration. This last requirement is vital for rapid slitmask alignment, and spectrographs that neglect this feature are chronically wasteful of observing time. The first requirement argued for a circular carousel to hold the slits, since linear mechanisms always waste 50% of the space to allow for translation. To simplify the exchange, we used magnets to hold the slits in the carousel, so that a single mechanical actuation, coupled with translation, can remove and replace the slits in their stations. Finally, we implemented a three position slit translation mechanism. At full extension, it engages a slit in the carousel where a gripper opens or closes to drop or pick-up a slit from its position. At midrange, it is in a detent and holds a slit frame in the focal plane. Fully retracted, it can hold the same slit frame without obscuring the field available for spectroscopy. This will allow rapid and reliable replacement of the same slit in the focal plane.
Meeting our design goals required matching the size of the slit frames to the smaller (3’x 5’) spectroscopic field rather than the entire imaging field, which means that the slit masks do not completely cover the field stop located just behind them. To block out the perimeter required the addition of a mask, which is driven by a stepper motor and flips into place whenever a slit frame is being held. The mask translation and carousel rotation, like most other motions in the spectrograph, are driven by smartmotors from Silvermax corporation, which contain motor, encoder, and microcontroller in one unit. The smartmotor responsible for slit translation (via leadscrew), also generates the signals to open and close the mechanical gripper and the perimeter mask, allowing autonomous coordinated motion with no overhead from the control computer or bandwidth from the RS485 communications network that links the smartmotors together. Another nice feature of the smartmotors is their built-in torque limit, which can shut them off in the event of a mechanical clash, and eliminates the need for mechanical limit switches. The slit translation and rotation stages incorporate Hall-effect sensors that allow homing of the stages at start-up, but afterwards the encoders on the motors keep up with their positions.
