Cerro Tololo Inter-American Observatory, Victor M. Blanco 4-m Telescope: an upgrade to the telescope control system Timothy M. C.Abbott a,1, German Schumacher a, Michael Warner a, Eduardo Mondaca a, Ricardo E. Schmidt a, Rolando Cantarutti a, Manuel Martinez a, Omar Estay a, Francisco Delgado a, Alistair Walker a, and for the Dark Energy Survey Collaboration2 a National Optical Astronomy Observatories, 950 N. Cherry Ave., Tucson, AZ, USA 85719 ABSTRACT The CTIO V. M. Blanco 4-m telescope is to be the host facility for the Dark Energy Survey (DES), a large area optical survey intended to measure the dark energy equation of state parameter, w. The survey is expected to use ~30% of the telescope time over 5 years and use a new 520 megapixel CCD prime focus imaging system: the Dark Energy Camera (DECam). The Blanco telescope will also be the southern hemisphere platform for NEWFIRM, a large area infrared imager currently being commissioned at the Mayall Telescope at KPNO. As part of its normal cycle of continuing upgrades and in preparation for the arrival of these new instruments, the Blanco telescope control system (TCS) will be upgraded to provide a modern platform for observations and maximize the efficiency of survey operations. The upgraded TCS will be based on that used at the SOAR telescope and will be a prototype of the TCS to be used by LSST. It will be optimized for programmed and queued survey observations, will provide extended real-time telemetry of site and facility characteristics, and will incorporate a distributed observer interface allowing for on- and off-site observations and real time quality control. Hardware modifications will include the use of absolute tape encoders and a modern servo control and power driver systems. Keywords: Telescope, Telescope Control System
1. INTRODUCTION The V. M. Blanco 4-m telescope at Cerro Tololo Inter-American Observatory has been one of the pre-eminent facilities for U.S. astronomy for over three decades. As such, it has undergone a series of upgrades in every respect, from instrumentation to basic infrastructure. In recent years, the operating model has been modified to reflect the changing needs of its observer community. A reduced number of instruments are now supported and the focus is on efficiency of operations while maintaining a consistently high quality of service. The telescope has seen a wide range of instruments over the years, including a series of prime focus cameras of steadily increasing areal coverage and, more recently, infrared imaging cameras. This evolution is expected to continue and will see a significant jump in the next few years with the delivery of NEWFIRM and DECam and the development of survey-oriented observing modes.
1
[email protected], phone & fax: +56 51 205 200, www.ctio.noao.edu. Funding for the DES Projects has been provided by the U.S. Department of Energy, the U.S. National Science Foundation, the Ministry of Science and Education of Spain, the Science and Technology Facilities Council of the United Kingdom, the National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign, the Kavli Institute of Cosmological Physics at the University of Chicago, Financiadora de Estudos e Projetos, Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro , Conselho Nacional de Desenvolvimento Científico e Tecnológico and the Ministério da Ciência e Tecnologia and the Collaborating Institutions in the Dark Energy Survey. The Collaborating Institutions are Argonne National Laboratories, the University of Cambridge, Centro de Investigaciones Energeticas, Medioambientales y Tecnologicas-Madrid, the University of Chicago, University College London, DES-Brazil, Fermilab, the University of Edinburgh, the University of Illinois at Urbana-Champaign, the Institut de Ciencies de l'Espai (IEEC/CSIC), the Institut de Fisica d'Altes Energies, the Lawrence Berkeley National Laboratory, the University of Michigan, the National Optical Astronomy Observatory, the Ohio State University, the University of Pennsylvania, the University of Portsmouth and the University of Sussex.
2
Ground-based and Airborne Telescopes II, edited by Larry M. Stepp, Roberto Gilmozzi, Proc. of SPIE Vol. 7012, 70123K, (2008) · 0277-786X/08/$18 · doi: 10.1117/12.789614
Proc. of SPIE Vol. 7012 70123K-1 2008 SPIE Digital Library -- Subscriber Archive Copy
NEWFIRM[3] is a 4kx4k infrared imager already in use on the Blanco’s sister telescope, the Mayall 4-m, on Kitt Peak. It will be brought to Cerro Tololo in 2009. DECam[2] is a large area CCD mosaic currently being built by the Dark Energy Survey Collaboration specifically for use on the Blanco, first light is expected in 2011. Both of these instruments are survey instruments and to operate most efficiently, make considerable demands on the functionality of the telescope. In particular DECam will require capabilities not currently available – with a combined total of 520 megapixels and a readout time of less than 20 seconds, this facility has much in common with instruments now developed for modern, 8-m class telescopes and makes similar assumptions of the host telescope. Indeed, the Blanco telescope is considered a component of the instrument itself within the collaboration. In response to the needs and expectations of these new instruments, CTIO has embarked upon a series of upgrades under the rubric the CTIO Facilities Improvement Project. This project includes improvements to the telescope itself [1] and construction of a specialized instrument maintenance facility. In this paper, we are concerned with the upgrade to the telescope control system (TCS). The TCS is the software and hardware system by which the observer controls the telescope. It incorporates the telescope drive, dome control and subsystem communications and control. It is expected to interface tightly with the instruments in use. The Blanco TCS was last upgraded significantly around 20 years ago and an upgrade is past due. Such an upgrade, while informed by DECam and NEWFIRM’s needs, must maintain the classical observing mode, permitting use by inexperienced visiting observers. It must continue to support the existing instrument set and must be installed without interrupting ongoing observations. CTIO has recent experience in the development of a modern TCS at the SOAR telescope on Cerro Pachon and is deeply involved in the production of the LSST TCS. It is natural, therefore, to seek to design the new Blanco TCS in accordance with the same principles used for these modern telescopes. In effect, we will retrofit the previous generation Blanco with a current-generation TCS based on SOAR and in the future will consolidate LSST into the support of all three. The Blanco TCS development is also, therefore, a prototype for LSST.
2. PROJECT SCOPE AND GOALS The upgrade is comprehensive and will replace all components of the telescope drive except the drive motors and their drive trains. This includes encoders, motor controllers and power amplifiers. Telescope, dome and instrument control will be decoupled, modularized and distributed among multiple computers with local autonomy. Many obsolete and obsolescent components will be replaced (some are 1970’s originals!). Where possible, the new components will be taken off-the-shelf and have an anticipated market lifetime of at least 5 years, ensuring future operational support and availability of spares. We will seek to improve delivered image quality with improved tracking algorithms. We will use a state-of-the-art mount model with optimized accelerations and slew trajectories. Functionality will be extended to incorporate and record telescope and environment telemetry, connectivity and flexibility will be upgraded via modern human and machine interfaces. The project has adopted a small number of explicit specifications for the upgraded TCS. DES provides a slew requirement of 2 degrees in 17 seconds, to match the camera’s readout time. The project has adopted a goal of 3 degrees in 20 seconds, track-to-track. Tracking will have a drift of less than 0.5”/min everywhere within the DES survey area (airmass < 1.5) and a jitter of 0.1” r.m.s. DECam will generate guiding error signals at the rate of 1Hz. The operational deadline for the upgraded TCS is September 2010, although we have programmed the work to be complete by January of that year.
3. HARDWARE CHOICES AND THE SERVO LOOP The existing Blanco TCS is built around a VME chassis and PMAC controller incorporating components that are obsolete and difficult to replace. For the upgraded TCS, we will capitalize on our experience with PMAC by continuing with a current generation controller, but will move to a PCX/PCI bus and Linux boxes to address longevity and sparing needs. Communications will be via Ethernet wherever feasible. We have selected Delta-Tau power drivers to replace the existing, custom-built units. The new power drivers will permit modification of the drive algorithms in software, whereas the existing drivers are hard-wired.
Proc. of SPIE Vol. 7012 70123K-2
Each telescope axis is driven by two custom-built, counter-torqued motors, these will not be replaced. Currently, the motor controller selects motor drive rates (Figure 1). Feedback derived from tachometers is applied through a hardwired rate loop. The motor controller receives feedback only from the encoders.
Figure 1: existing (left) and upgraded (right) servo loop designs. In the upgraded TCS, the rate loop and motor preload logic will be integrated into the motor controller. It will receive feedback from both tachometers and encoders and calculate the torque to command of the power driver. The power driver will not receive external feedback. This arrangement permits more flexible control of the drive, allowing realtime calculation of optimum slew trajectories to minimize acceleration energies at critical telescope resonances. A telescope lumped-mass model and sample slew trajectories have been developed to ensure that the upgraded system will meet our requirements. Slew trajectories are calculated to minimize any spectral components in the acceleration that could excite the telescope structure during a slew. A step acceleration is convolved with a Chebyshev window, producing a low pass frequency cutoff. Figure 2 shows the telescope trajectory obtained and Figure 3 shows the corresponding acceleration power spectral density with a sharp cutoff above 1 Hz. Existing interlocks are hard-wired into the power drivers, upgraded interlocks will be distributed to both motor controller and power drivers and mediated through software and firmware. An additional benefit of the upgrade will be the introduction of an independent rate limiting system to prevent runaways. The existing TCS does not incorporate such a safety feature. The existing system uses dual absolute and relative encoders. The absolute encoders are too coarse to use for tracking. The relative encoders, while they have the fine resolution necessary for tracking, are friction driven and cannot operate at the high speeds required for a large slew without risk of slipping. Mode switching between the two encoder sets for a slew costs valuable time and prevents the existing system from achieving the required specifications. We will therefore replace the existing encoders with a single tape encoder on each axis, each with two read heads and with sufficient resolution and reliability for all anticipated telescope velocities. (For the purposes of providing a fiducial, the existing absolute encoders will be retained, but not used once the new encoders’ zero points are set). Our chosen vendor for the encoders is Heidenhain with whose encoders we have considerable experience from their deployment at SOAR. The Blanco telescope was not designed to use tape encoders and we must therefore retrofit appropriate mounting surfaces. In the DEC axis, this requires a completely new assembly, but the RA axis provides an excellent site in the form of the lower oil bearing which has a region above the oil pads that is appropriately uniform and clean. While this site is longer than necessary and hence demands greater expense for the encoder tape, it is still less expensive than attempting to build a new encoder track nearer the RA axis. Studies from the SOAR telescope show these tape encoders to have a correctable periodic error[6][3]. However, the long RA tape renders this error at too high frequency and too low amplitude for the mount servo to respond and it can be disregarded. A shorter DEC encoder tape will not generate problems from this periodic error because DEC tracking drive rates are considerably smaller.
Proc. of SPIE Vol. 7012 70123K-3
1 Ssec trajectory (4sec -BOdb Chebyshev Window) 3.5
3
2.5
2
S
U.S
U
-U.S U
2
4
6
9
IU
12
14
16
19
2U
Ti me(sec)
C C C C C
C
PS(deg/sec2sqrHz))
C
C
Figure 2: Telescope slew trajectory simulation for a 3degree step.
Figure 3, Trajectory acceleration power spectral density showing a sharp cutoff above 1 Hz.
Proc. of SPIE Vol. 7012 70123K-4
The mounting surface for the RA encoder has been examined with digital micrometers during operation to measure its motion due to energizing the oil bearing, the shifting weight of the telescope between pointings and temperature variations (Figure 4). These were found to be well within the known tolerances of the tape encoders. The RA encoder tape has been mounted. Design of the DEC encoder mounting surface, to be placed between the telescope barrel and the RA yoke, is complete and fabrication is under way. Blanco Data Jan2l-25 2009
I
-0.05
•
Mitutoyo#l Mitutoyo#3
-0.1
-0.15 E E
-0.2
I: * Si
0 -0.25 C
0
*
.•.. +•._ --
V **
4—
— — ——
a?
'4
-0.3
St •_•
-0.35
-0.4 -0.45
-0.5
-90
-60
-40
20
-20
40
60
90
Hour Angle [deg]
Figure 4, Radial displacement measurements on RA axis encoder mounting surface, which is also one of the telescope’s oil bearing surfaces.
To accommodate the requirement that the upgrade proceed while the telescope continues in normal operation, we will need to be able to quickly and reliably switch between the old and new systems. To that end, a patch panel has been developed and installed which allows signals and control lines to be routed to either system as required. Some telescope and instrument subsystems are controlled by the so-called Smart Motor Controllers (see Figure 5), an inhouse designed and built motor controller system directed via an RS-485 serial connection. These will be retained for the time being and connected to the TCS via an Ethernet/serial converter. Eventually, these controllers will be upgraded to full Ethernet connectivity as well.
4. SOFTWARE The existing TCS is written in ‘C’ and runs under VxWorks. The upgraded TCS will be written with the graphical programming language LabView v8[7] wherever appropriate and feasible, running under Linux with the Real Time Application Interface (RTAI) extensions. C/C++ will be used when LabView does not suffice. A schematic diagram of the upgraded TCS design is shown in Figure 5. Processing will be distributed among several computers each with welldefined operational responsibilities – in particular, dome, guider and comparison lamps that are currently controlled from the same CPU that runs the pointing model will be passed to a separate utility computer.
Proc. of SPIE Vol. 7012 70123K-5
Figure 5: Upgraded TCS schematic design.
The SOAR TCS employs a setpoint/status control principle (Figure 6) in which a control application computes a set point which it sends to a device process. The device process then attempts to achieve that set point and reports its status back to the control application. An example control application is the TCS application which computes appropriate set points, e.g. telescope orientation coordinates, and sends them to the process running in the motor controller. This device process then drives the telescope to position, closing time-critical loops locally, and reports back to the TCS application when it succeeds.
Figure 6: The setpoint/status control principle.
Proc. of SPIE Vol. 7012 70123K-6
The TCS computation engine will be provided by the Rutherford Appleton Laboratories TCS kernel TCSpk [5]. Already in use at SOAR and many other telescopes, this sophisticated package incorporates an advanced and very well tested telescope model appropriate for our needs. The TCS architecture is based on Supervisory Control and Data Acquisition (SCADA) principles in multi-tier hierarchical control model with local autonomy via a master/slave strategy. Control is exerted via a publish/subscribe message passing mechanism using RTI’s Data Distribution Service (DDS, see Figure 7 and [4]).
Distributed Applications COMMON SERVICES
Transport Figure 7, Upgraded TCS software infrastructure. Message types are signals, events, commands, status and requests. Signals are rapidly generated and usually contain time-critical data; in most cases, it is more important to get to the next issue than retry a dropped one. Events are asynchronously generated, priority messages which must be delivered reliably. Commands are sequential instruments which must be received in order – they return immediately and trigger actions each in a separate thread thus permitting multiple commands to be given while actions proceed, allowing the latter to be stopped or monitored. Status messages contain persistent data about states or goals, timeliness will differ from one application to the next. Requests are twoway, request/reply transactions for a specific service or data. All telemetry data will be accumulated and published in a MySQL facility database, running on a dedicated computer, which may be accessed through subscription by external systems. Other telescope subsystems, such as the weather station and sky camera, will be adapted to populate this database and provide additional environment context data for the DECam and NEWFIRM (or any other) observing environments. A number of services will be provided. A connection service will allow the distributed applications to locate and connect to each other. An event service will support high-performance publish/subscribe communications. A command service will provide client/server communications for application control. A logging or telemetry capture service will collect, record, distribute and analyze system messages. A persistent store service will hold system configuration information, calibration information, performance data, etc. An error handling service will monitor the whole system for improper behavior, support recovery operations and report errors to applications and users. LabView GUIs will be produced, based on the SOAR model, for both regular operations and for engineering. This will also allow for remote control. Version control will be managed via the open source SubVersion package.
Proc. of SPIE Vol. 7012 70123K-7
Instruments will communicate with the Blanco TCS via Ethernet and TCP/IP accessing the relevant services and using a library of commands/messages appropriate for all aspects of controlling the telescope for astronomical observations. These instruments may employ the proprietary DDS messaging software for tighter integration, but need not provided that the appropriate messaging protocol is followed.
5. REFERENCES [1] [2] [3]
[4] [5] [6] [7]
Abbott, T. M. C, et al., “Cerro Tololo Inter-American Observatory, Victor M. Blanco 4-m Telescope, and the Dark Energy Survey”, Proc. SPIE 6267, 119-126 (2006). Depoy, D., et al., “The Dark Energy Camera (DECam)”, these proceedings (2008) Probst, R. G., Gaughan, N., Liang, M., Hileman, E. A., Penegor, J., Daly, P. N., Chisholm, G. H., Hunten, M. and Merrill, M. K., 2004, "Project status of NEWFIRM: the widefield infrared camera for NOAO 4-m telescopes", Proc. SPIE Vol. 5492, 1716-1724 (2004). Schumacher, G. and Delgado, F., “The LSST middleware messaging system”, these proceedings (2008). Wallace, P. T., “A rigorous algorithm for telescope pointing”, Pro. SPIE 4848, 125 (2002). Warner, M., Krabbendam, V. and Schumacher, G., “Adaptive periodic error correction for Heidenhain tape encoders”, these proceedings (2008). www.nci.com/labview
Proc. of SPIE Vol. 7012 70123K-8
