This will count as one of your downloads.
You will have access to both the presentation and article (if available).
Photogrammetry has been used as an alternative measurement technique for the 32-m primary since 20151, and has gradually replaced our use of holography and laser trackers for this task2 during recent years. Once the object has been targeted, photogrammetry maps may be obtained in around one hour. The technique does not require the installation of special equipment on the antenna, and has the advantage of allowing surface maps to be taken at any chosen elevation. The main drawbacks for the LMT application are environmental, since the antenna operates without an enclosure; strong winds may prevent use of the site tower crane for image taking, while the formation of condensation and frost on the reflector surface will "switch off" the reflective targets.
In this paper we discuss comparative measurements taken as the first outer segments were installed, and the use of photogrammetry to carry out the alignment of the fully installed 50-meter surface. At the time of writing this activity is still in progress, however full-surface alignment to the order of just over 100 microns was achieved quite quickly, with multiple elevation maps allowing the development of a usable 50-m active surface model for compensation of gravitational distortions.
Prior to assembly of the new segments, a review of the original surface support system was carried out. Based on previous experience with the alignment and stability of the inner rings, it was decided to upgrade a large number of the early subpanel support and adjustment components. A key modification was the inclusion of lateral adjustment for subpanel support points, helping to minimize bending moments on the panels both during and after integration. Improvements in the ease of surface setting and greater surface stability were immediately observed following component overhaul. Form setting of individual segments was carried out at the LMT facilities in Puebla and again at the telescope site, using the iterative process developed previously that makes use of laser tracker surface measurements.
While the original implementation of the subpanel support system enabled the setting of individual segments to a mean surface error of around 30 micrometers RMS, this mean value was improved to around 20 μm for the entire set of 96 segments for the outer two rings, with the best segments coming in at around 18 μm RMS surface error. We believe this brings us close to the limit of achievable surface accuracy for the LMT design of laminated composite subpanels supported on simple mechanical differential adjusters.
We present an overview of primary surface improvements since 2011, and the main aspects of the LMT aperture expansion relating to the assembly and alignment of the surface segments for the outer two rings of the 50-meter primary.
Media Lario used their in-house coordinate measuring machine to adjust the surface during assembly, with the reflector panels facing upwards. As part of the Final Acceptance Review measurements of the surface were undertaken by LMT staff at the Media Lario factory, using both a laser tracker and photogrammetry. Measurements were also made of the electroforming mold for the central panel. The reflector was mounted on a rotating stand allowing surface measurements to be performed according to the respective gravitational load cases. Measurements at the Media Lario factory provided a useful reference for repeat data taken at the LMT site, since the reflector was shipped as a fully assembled unit, designed to require no further adjustment after leaving the factory.
In this paper we present the surface measurements conducted during the review, and comparisons of the observed gravitational load deformations with those predicted by FEA. Although the latter were often at the level of measurement uncertainty, we were able to verify specific cases, as well as performing a sanity check on the manufacturer's design analysis. The measurements confirmed final surface error values leading to reflector acceptance by the project. An RMS surface error of the order of 25 microns over the entire reflector was recorded at 60 degrees elevation using photogrammetry data after adjusting to the best-fit parabola, showing compliance with the LMT specification. Acceptance review measurements also provided a baseline for surface measurements at site prior to installation.
In the first tests, a laser tracker was employed to measure groups of targets on the M3 platform, the M4 mirror, and the receiver cabin floor. The baseline distances were then compared continuously for several hours. In this test, the M4, which is supported directly from the M3 platform, was found to be more stable than the receiver cabin floor. In most cases, the errors were consistent with thermal variations in the structure. The most dramatic change was observed near sunset, with position drift rates of about 300 μm/hr. Later at night, the M4 position stabilized, but the receiver cabin still sometimes showed position variations of over 100 μm/hr. These results put a bound on the maximum allowable time between checking the pointing and focus of the telescope.
The second tests measured the stiffness of the receiver cabin floor by measuring the underside of the platform from the floor below while weights were placed at different locations in the testing area of the floor above. As expected, the largest deflections were measured when the load was placed at the center of the floor grating between the mid-span of the smallest floor structure I-beams, with a stiffness of 14 N/μm. The stiffness was about 10% higher (just under 16 N/μm) directly at the smaller I-beams near their mid-span. A more dramatic difference was measured for loads near a main structural cross beam. In that case, targets that connected to the beam itself were found to have a stiffness of nearly 34 N/μm, more than twice the mid-span stiffness. However, in that location, the stiffness for loads in the middle of the floor grating increased only to 17 N/μm, because the flexibility is dominated by the floor grating itself. Comparison of the unloaded condition of the structure after each test showed slow drifts of the relative positions of the platforms, consistent with the thermal drift hypothesis supported by the first tests.
This paper presents the tests and analysis, together with the detailed results of the receiver room motion and floor stiffness.
Accurate photogrammetry requires a robust strategy for the incorporation of multiple camera stations, a task complicated by the size of the antenna, obstructions of the surface by the sub-reflector and tetrapod legs, and the practicability of using the site tower crane as a moving camera platform. Image scaling is also a major consideration, since photogrammetry lacks any inherent distance reference. Therefore appropriate scale bars must be fabricated and located within the camera field of view. Additional considerations relate to the size and placement of reflective targets, and the optimization of camera settings. In this paper we present some initial comparisons of laser tracker, holography and photogrammetry measurements taken in 2015, showing clearly the status of alignment for distinct zones of the currently operating 32.5 m primary collecting area.
To address these concerns, the project conducted a series of field tests, within the constraint of having minimum impact on night time observations. The supplier sent two coupon samples of a reflector panel prepared identically to their proposed M2 surface. Temperature sensors were mounted on the samples and they were temporarily secured to the existing M2 mirror at different distances from the center. The goal was to obtain direct monitoring of the surface temperature under site thermal conditions and the concentration effects from the primary reflector. With the sensors installed, the telescope was then commanded to track the Sun with an elevation offset. Initially, elevation offsets from as far as 40 degrees to as close as 6 degrees were tested. The 6 degree separation test quickly passed the target maximum temperature and the telescope was returned to a safer separation. Based on these initial results, a second set of tests was performed using elevation separations from 30 degrees to 8 degrees.
To account for the variability of site conditions, the temperature data were analyzed using multiple metrics. These metrics included maximum temperature, final time average temperature, and an curve fit for heating/ cooling. The results indicate that a solar separation angle of 20 degrees should be suitable for full performance operation of the LMT/GTM. This separation not only is sufficient to avoid high temperatures at the mirror, but also provides time to respond to any emergency conditions that could occur (e.g., switching to a generator after a power failure) for observations that are ahead of the motion of the Sun. Additionally, even approaches of 10 to 15 degrees of angular separation on the sky may be achievable for longer wavelength observations, though these would likely be limited to positions that are behind the position of the Sun along its motion.
The Large Millimeter Telescope Alfonso Serrano (LMT) is a 50-meter (currently 32m) diameter single-dish telescope optimized for astronomical observations at millimeter wavelengths in the range 0.85 mm < λ < 4 mm. During initial operation, the LMT makes use of the central 1.7 meters of a 2.5m hyperbolic secondary reflector constructed of cast and machined aluminum. Following the first light campaign in 2011, a program of iterative surface sanding was carried out to reduce the surface error of the central area to a level compatible with that presently achieved for the primary reflector. Metrology during the sanding process was conducted using a Leica laser tracker. A total of 22 sanding iterations were interspersed with tracker measurements at differing spatial resolutions, allowing the RMS surface error to be reduced from 63 to 35 microns. Maps for the final iterations were repeated for distinct scan patterns to check for systematic variance. Since the work was carried out in early 2013, repeat measurements of the dismounted secondary have confirmed the stability of this reflector.
In this paper we present details of the surface improvement program with emphasis on the metrology techniques used throughout the process. We discuss issues such as data sampling, measurement geometry, and mirror orientation. We also consider the steps taken to ensure tight control of the sanding task itself, since this process was carried out entirely by hand. Finally we present some comparative metrology results obtained using our laser tracker and photogrammetry equipment.
The primary reflector of the Large Millimeter Telescope (LMT) Alfonso Serrano is presently composed of 84 surface panels arranged in three concentric rings, providing a 32.5 meter collecting area. Each panel comprises 8 precision composite subpanels having electro-formed nickel skins bonded to an aluminum honeycomb core. Differential thread adjusters beneath each subpanel allow for the manual removal of tip/tilt and piston errors, in addition to facilitating some fine tuning of the surface shape. An assembled panel provides a surface area of approximately 8-12 square meters.
Preparation of surface panels in 2012 and 2013 for Early Science observations made use of a Leica laser tracker. Measurement and adjustment of panels was carried out off the antenna, achieving a mean panel RMS surface error of 29.5μm for the 67 panels processed to date, with a spread of 23-37μm. A panel stability check consisting of surface walk-on tests and repeat metrology resulted in an increase in the mean surface error to 31.0μm. Following installation, in situ tracker measurements of 19 panels showed a final mean error of 45.3μm. Panels are adjusted by hand using an iterative process. In-house data processing uses fiducial marks scribed onto the subpanel molds and replicated during manufacture, to achieve accurate registration of the surface point cloud during data fitting. The number of iterations varies, depending mainly on the behavior of the differential adjusters. A well-behaved panel may be set within around 7 hours. In this paper we describe the iterative panel surface adjustment process used to date. We focus on metrology technique and data processing using the laser tracker, and present comparisons with trial photogrammetry measurements.
View contact details
No SPIE Account? Create one
