AST Monthly Progress Report September, 1999 Most of this narrative has to do with the project schedule as it pertains to jobs currently underway. A separate narrative discusses jobs that involve contracts for future work. During September, 1999, we had three employees (Andre Hedrick, Bill Pilkinton, and Mike Williamson) working at TSU on the telescope program in addition to M. Krebs and M. Wells, who provide machining work at their private shops. We have hired Mike Williamson, a Vanderbilt graduate in computer science, on a consulting contract to do computer programming and instrumentation work. See symbol Mwi in the attached schedule. A list of the jobs he is working on is attached. The next four sections discuss highlights of the schedule for September. I. OVERHEAD/ENCLOSURES Task 30: Assessment of spare parts: We have established criteria for deciding what spare parts to keep on hand and presented them to Dr. Gull in August (see attached list of criteria). On the basis of these criteria, we made an inventory of spares in hand and noted these in our parts list. We have a further list of parts to buy before the telescope becomes operational, most notably more timing cards (see 159 below) and some power supplies. Enclosures: These tasks are either done or underway by Boyd, our maintenance contractor in Arizona. Busby visited Fairborn Observatory in late June and found progress on the motions of the enclosure. Eaton will probably go out in October or November and assess what needs to be done to finish preparing the enclosures for the telescope. II. MECHANICAL INTEGRATION Tasks 75&95: Assembly of telescope: The telescope is close to being reassembled. We must still put on the top end [quadrapod] (2.5 hours work with adjustments), hook up three of the lateral supports for the primary mirror (2 hours), reattach a puck we knocked off disassembling the telescope (2 hrs over 2 days), engage all the mirror supports (2-3 hours with adjustments for the hard points), add the surrogate secondary mirror (20 min), balance the telescope tube (1/2 hour), add the altitude drive tractor (1.5 hrs), finish adjusting/modifying the clamps for the drive tractors (6 hours under tasks 75 and 77), add the boxes for the control electronics (2 hours with adjustments). This work should take about three weeks to complete at two mornings a week. Task 83: Insulation: We have ordered new insulation, having decided the stuff we have in hand won't work right. We already have templates for cutting the insulation to shape once it arrives. Since applying the insulation requires the telescope to be taken apart, we will get a manufacturing technician (MT in the schedule) to put it on as part of Task 131. Task 84: Construct wiring harnesses: This task is about 70% complete with final wiring waiting on having the control-electronics boxes on the telescope. Tasks 100-104: Modify oil bearings: We used plexiglass pads to assess the working of the oil bearings and the oil-return system. These tests indicate we should reduce the clearance between the seals and the supported pad to reduce seepage out of the primary oil-return system. We will do this as part of Task 131 when we take the telescope apart to ship it to Arizona. We have also put the secondary oil-return system in place, and it seems to be working properly. As the acceptance test of these systems, we will run the oil bearing for eight straight hours, or as long as we can stay at the armory on a single day, and observe its behavior. Task 109: Boxes for electronics; These were designed by Eaton, built by Hamilton Machine, skinned by Bill Pilkinton, and detailed by Eaton. They must still be insulated by Pilkinton and set up in the lab to receive the telescope- control electronics before being fitted into the telescope. This task is essentially finished. III. CONTROL SYSTEM Work in September 1999 concentrated on the control system. Figure AST-13-A01 (attached) gives a functional breakdown of the control system, showing its division between the two computers we will use. Work on the programs for the executive computer (that runs the spectrograph and does scheduling) is part of tasks 151, 152, 1nd 184, so our schedule naturally concentrates on the telescope- control computer at this time. Figure AST-13-A02 shows the programs running on that telescope-control computer, the one to the right in AST-13-A01. For your information, the basic control of telescope motions is through a PID controller running on a Baldor NextMove motion-control board. Tasks 156&157: Basic timing, coordinate-conversion, and trajectory calculations: These programs have been essentially finished since 1998. We have spent the equivalent of four days on them this month and now have a set of programs that take commands (star position, stow, go to a specified altitude/azimuth) and calculate slews and tracking rates for the telescope. The commands are communicated to the control program through the named pipes we will use in the actual control system, from a front-end program with a command-line prompt for entering commands. A plot of the calculated position of the telescope every five seconds in a simulated observing session is attached. Task 158: Functions for communicating between control processes: We are communicating between processes and computers in the control system with standard unix (System V) IPC, which will make it possible to integrate any linux/unix computers we would use in this instrument. With reference to AST-13-A02, we have run prototypes of programs for all three of the IPC techniques referred to in that diagram. For the past six months we have had a program that would run stepper motors through a daemon on one computer through an internet socket from a second computer. We have communicated between front-end and stepper-control processes for both serial devices (seriald in AST-13-A02) and the telescope-control daemon (telescoped) through named pipes. Finally, we have this month written a protype of the control for the shared memory we will use to communicate data among various processes as shown in AST-13-A02. Task 159: Writing device drivers for motion-control: There are two device drivers for communicating with the computer boards at the heart of the motion-controller for the telescope, a timing board from TrueTime and a NextMove motion-control board. The timing board synchronizes its internal timebase to timing signals from the GPS satellites and delivers accurate times and interrupts to the motion-control program and NextMove board. Andre Hedrick is writing these drivers under a consulting contract. He is close to finishing the driver for the timing board, having mainly quality-control work left to do. The driver can program the board, read information out of it, and catch and identify its interrupts. This driver still requires writing certain functions needed to extract information efficiently for our application program, but Hedrick's existing code gives the basis for these functions. Furthermore, Hedrick is working with TrueTime and SuSE to get SuSE to maintain the linux driver for this GPS card. The driver for the NextMove motion-control card has been working for months now, but it requires some further programming, specifically work on the interrupts and writing functions to extract positional information more efficiently. Even so, the driver is in a state in which it could be used in the control system as it is, but the further modifications will be very useful. Task 160: Implement the servo loop: This involves integrating the timing and motion-control cards into a system that sends calculated velocities to the axes. We have had for the past year, apparently without your acknowledgement, \a prototype program of this sort that calculates a parabolic motion profile and commands motors driving the reflected inertia of the two axes of the telescope to follow it. We use it on a two-axis simulator in our lab. This program allows inputting values of the various parameters of the PID control algorithm, which gives us a way to determine what values to use for these quantities. This simulator has allowed us to experiment with implementation of the limit switches, and it will allow us to experiment with implementing the home-position sensors (part of task 169). Task 160 is only 50% complete just now because the device driver for the timing card has not yet been complete enough to set up the whole motion-control algorithm. Tasks 161,162,&166: Write and test serial daemons: These programs communicate with devices that control stepper motors and I/O units over serial lines. We have a working communications program that Mike Williamson is improving and has made much more reliable. He can now communicate with the steppers and I/O units over the same serial line, commanding moves of the steppers (linear actuators) and reading their positions and statuses, setting and reading DIO units, such as the positions of the absoulte encoders for the altitude and azimuth axes, and reading the the analogue I/O device that will measure the oil pressure and the voltages of the various power supplies. Task 171: Set up Apogee camera in lab: We bought this camera at the suggestion of Larry Ramsey at the "PDR" last September and will use it for our tracking tests (Tasks 175 et seq.). Hedrick was supposed to set this camera up this month, but we decided his time would be better spent on other jobs. So we have deferred the task and assigned it to Williamson. There is a linux device driver/control program from the Univ. of Iowa for this camera, and we have a copy we bought on speculation. IV. SMALL SYSTEMS This work is mostly done and is being wound up into testing the control system. We have assigned most of it to Mike Williamson, who has begun testing these systems under computer control. Jobs that remain are Task 228, a contingency for modifying the absolute-encoder mounts (which requires a few small mechanical modifications), Task 239, testing of the secondary cell for stiffness (which requires Williamson and Eaton to set the cell up with indicators and move it around). QUESTIONS OUTSIDE THE SCHEDULE 1. Criteria for shipping the telescope. In June we presented Drs. Withbroe and Gull a rational set of criteria for deciding when the telescope is ready to ship to Arizona (see the appended list). These are implicit in the schedule we also presented in June. Specifically, we expect to get the oil bearings working reliably (Tasks 100 and 104), to get the drive tractors adjusted and working (Tasks 74 and 75), to get the control components for the alt/azimuth motions integrated into the telescope so as to perform basic movements (slewing motions, tracking motions, finding home positions, working of the limits) under computer control (tasks 168&169) sufficiently well to work all night without problems (essentially for the 7-hr period the National Guard Armory is open). We have consistently presented a logical breakdown of the various tasks involved in perfecting the TSU telescope and its control system between those that can be done in an assembly building (Armory) and those that must be done at a site with clear access to the sky (Fairborn Observatory). Dr. Gull has proposed in an E-mail of 21-Aug-99 a draft alternative set of criteria that are much more restrictive. Specifically, under his point 3, we would practically have to finish and perfect the telescope control system before shipping the telescope to Arizona. This approach may be perfect for an instrument on a satellite, but it certainly is not necessary for a ground-based telescope, as your consultant Larry Ramsey agreed at our meeting in May. Our proposed criteria divide the development between mechanical work, which has to be done where the parts can transported into a shop (although Boyd has a pretty good one at Fairborn Observatory), and those that involve primarily programming and fine- tuning small systems, which can be done more easily in our labs and transferred to the observatory over our existing internet connection (at least for the computer programs) or by UPS. We have designed the telescope and its control system specifically so the various parts can be developed as modules and perfected through the process of continuous improvement. Doing this process of continuous improvement in a covered building in Nashville is considerably impractical if not impracticable. The door through which we would observe is too low to see hardly any of the sky. The GPS antenna would not see the sky. The Armory does not have the right electrical connections for running this telescope except for rudimentary tests, whereas Fairborn Observatory does. The Armory is open from about 7:30AM to 3:00PM and is locked up at night when we would have to make the tests Dr. Gull proposes to do there. We would not have the ethernet/internet service at the Armory needed to run the tracking tests Gull wants, while we will have this connection in Arizona. Furthermore, our agreement with the National Guard is based on the schedule you agreed to, which gets the telescope out of their building by the first of the year. Dr. Gull's proposed criteria seem to us to be too restrictive for the reasons we have just detailed. We respectfully request that our original criteria for shipment be accepted. 2. Setting accuracy. In his E-mail of 21-Aug-99, Ted Gull suggested adding a small tracking telescope to the 2-m scope to use in acquiring stars. Such a telescope would have 12-arcsec pixels with the 512x512 CCD we have for guiding tests, and would be useless for guiding. It would also be useless for acquisition of all but the brightest stars, since there would be so many of the things in a two-degree field that we couldn't possibly identify them automatically. For such a telescope to be useful for guiding, it would have to have a 10 arc-min field of view, or smaller. As for finding stars, Dr. Gull questions that we can adjust our telescope mount, or define the zero points of its axes, or calibrate the misalignments with a mount model. We estimate the vertical axis of the telescope will be set to better than an arcminute. The tilt axis will be perpendicular to it to within 6 arcsec. The optical axis will be perpendicular to the tilt axis by about 1.5 arcmin simply by mechanical adjustment alone, with better results possible eventually through observations of stars. The zero points will be known to about 2 arcsec each. Runout of the azimuth slewing ring is equivalent to 25 arcsec. All these misalignments can be easily calibrated with a mount model and some further minimized mechanically. One notes that such mechanically shoddy instruments as the McMath-Pierce and coude feed at Kitt Peak find stars quite well with such calibrations, and we ourselves must use mount models for our other telescopes. Even if we did not have such a mount model, the obvious step for finding at least bright (navigation) stars is to execute a spiral search around the expected position, as we do with our other telescopes. Dr. Gull simply ignored this suggestion when Eaton pointed it out to him in August. 3. Focusing and tracking. Asserting that we will not be able both to track and focus the telescope automatically, Gull continues to quote Don York on his experience with the Sloan telescope, apparently ignoring the difference between our classical Cassegrain telescope and the wide-field, imaging Sloan. In our case, we will focus by acquiring a bright navigation star, if need be, (our strategy with the existing APT's) and measure the size of its image with the acquisition camera. If need be, the telescope will go through a focus routine and reset the focus. Development of this technique is part of Task 177, scheduled for 2Q2000. *********************************************************************************** Attachments: A. JOBS FOR MIKE WILLIAMSON 1. Test setup for pressure sensor 2. Set up motor controllers 3. Run tests on absolute encoders 4. Programming od control daemon for serial communications 5. Set up select command 6. Tests of various motors in the lab 7. Test running RS422/485 from same controller 8. Test oil-pump controller in lab with other motors 9. Test of CM4050 to see if outputs use pull-up resistors 10. Run Apogee camera in lab B. SPARE-PARTS PHILOSOPHY 1. Have spares at least one deep for all critical components at Fairborn Observatory. Critical components are pieces that are likely to break and cannot be reordered quickly or replaced easily. Examples include electronic components, motors, and some bearings. 2. Have spares of the electronic components of the telescope in a simulator in the lab at TSU in Nashville. C. CRITERIA FOR SHIPPING TELESCOPE TO WASHINGTON CAMP I. The telescope must be MECHANICALLY SOUND 1. Mechanical parts fit together 2. Oil bearings work 3. Tracotor units track on drive circles 4. Mirror-support system works mechanically II. The ELECTRICAL SYSTEM must be READY FOR DETAILED TESTS 1. Tractors move telescope under computer control (but telescope does not necessarily track stars) 2. Limit switches and emergency-stop switches work; zero-point detectors are in place 3. Power distribution is in place 4. Wiring harnesses for motion-control system are in place 5. Wiring harnesses for stepper motors fit (possibly in place) AST CONTRACTS Mechanical Technician Pilkinton Mechanical Technician Bolin Optical Consultant Warren (pending) Instrumentalist/Programmer Hedrick Programmer Williamson Machinist Wells Machinist Krebs Data Reduction Hall (in preparation) CCD Controllers San Diego State