Feasibility of Using Holographic Measurements to Set GBT Steel
Ronald J. Maddalena
June 14, 1994
In GBT memo 68, R. Norrod, S. White, and I describe what were then our plans
on how to make holographic measurements on the Green Bank Telescope (GBT). I
recently learned that we may have to augment these plans because of additional
requirements that might be placed on the holographic measurements. This memo
documents the new requirements and the feasibility and costs of meeting the
requirements. I also will document how the new requirements translate into
requirements on, or an acceleration in, the development of other GBT systems.
II. New Requirements on Holography
Originally, holography was to be performed as a research effort aimed at trying to improve the surface of the telescope and to calibrate the gravitational deformations of the structure. The first holographic measurements were to be made well after the telescope was in routine operation, with subsequent measurements made over the lifetime of the telescope to further increase the accuracy of the dish.
RSI was supposed to set the complete surface to something like 1.5 mm rms and panel-to-panel adjustments to a much higher accuracy (few hundred micrometers). Thus, the originally-planned holographic measurements could provide confirmation of the panel-to-panel setting and, by using the actuators, could be used to remove large-scale errors in the surface. The originally-planned holographic experiments were going to be challenging in that they were to be done early in the life of the telescope when many of the necessary characteristics of the telescope might still not be well known. If the initial experiments did not work out or gave bad results, any setting of the surface could be easily undone by returning the actuators to their initial position. The results of the holographic experiments were never planned to permanently alter any component of the telescope.
Recently, I was told that RSI could not determine how to weld the actuator mounting brackets for parts of the surface to better than 1/8" (over 3 mm), with an unknown maximum error greater than this. Without our input, RSI will not be able to provide a telescope with a working frequency above about 5 GHz. Since the travel of the actuators is not great, and we are expecting that compensating for gravitational deformations will require a significant part of the actuator range of travel, actuator brackets must be welded to better than about 1/8" of their optimum value. I was asked to study whether NRAO could perform holographic measurements to confirm that RSI has welded the structure to this accuracy.
If the holographic measurements find a section of the surface that deviates by more than 1/8" then RSI has agreed to adjust the mount to our specifications (for example, by breaking the welds, adjusting the position of the brackets, and rewelding the mounts). Thus, our holographic measurements provide important information on how RSI has welded or should weld the structure. Since the welding must be done while RSI is still on site and two or so months before the telescope is accepted by and turned over to NRAO, we most likely will need to make these measurements sometime around November 1995 (assuming a completion date of January 1996) and we cannot take more than a few weeks to perform them.
III. Overview of Consequences to the GBT Project
The changes in requirements grossly moves forward by almost a year when we were expecting to perform holographic measurements. We have to reinvestigate the experimental parameters given in GBT memo 68 since the reason for the measurements have changed and, among other things, the maximum errors of the initial surface are at least twice what we anticipated. Most importantly, we can no longer consider holography a research project since we will make the measurements under tight time constraints to make possibly permanent changes to the surface of the telescope. Note that the November 1995 holographic experiments are in addition to the measurements described in memo 68 since we still need to make follow-up measurements to further improve the surface after the telescope is in routine operation.
If we decide to use holographic measurements to decide how to weld or adjust the actuator brackets, we cannot tolerate any failure in making the measurements. Many development areas of the GBT project (hardware and software) must be accelerated by months to have in time what will be needed for the measurements. To guarantee success in the limited time RSI will make available, I believe we must test months before November 1995 all of the applicable GBT software and hardware on an existing NRAO telescope -- I will assume we will use the 140-ft telescope since probably costs would be higher or the tests might not be sufficiently similar to the GBT measurements if we were to use a different telescope. The tests will allow programmers and engineers an opportunity to fix problems in their systems before the critical measurements in November 1995. I believe we will need two test holographic measurements to ensure success, the first by June 1, 1995 (i.e., about six to seven months before RSI turns the telescope over to NRAO) and the second something like two months after the first. Thus, the actual deadline for many software and hardware components is now June 1995.
Before we can make holographic measurements we must first learn about certain characteristics of the telescope (e.g., pointing, focus, slew behavior, settling times). For other telescopes, these characteristics are usually well understood before anyone attempts holographic measurements. Thus, project leaders should be warned that these critical measurements (e.g., the coefficients for the traditional pointing model) must be made as soon as the telescope is equipped with receivers and associated equipment, can move, and can take astronomical data. The results of the measurements must be close to being right. For example, if the pointing coefficients are significantly wrong, the holographic measurements could mistakenly produce an astigmatic surface.
To make these critical measurements of the behavior of the telescope, and the holographic measurements themselves, the GBT MUST be a working astronomical telescope equipped with the subset of equipment needed to make accurate continuum measurements of astronomical sources. The GBT must have a complete surface and be capable of moving at its top slew speed and of taking astronomical data. Therefore, NRAO must have access to the telescope before November 1995 to put the necessary hardware (receivers, backends, I.F. cables, etc.) in place.
If the initial measurements of the characteristics of the GBT cannot be made, are flawed, or suggest that a successful holographic measurement is impossible (e.g., if the pointing parameters cannot be determined to sufficient accuracy), then, at best, we will not be able to supply RSI with the information they will need to weld or adjust the actuator brackets or, at worst, we will give them faulty information and the dish might be permanently disfigured beyond the correctable range of the actuators. Even with all these preparations, holography might still not be a workable system on the GBT in the allotted time. And, even if we think it has worked, we have to have some way to corroborate the holographic results. I strongly suggest that an alternative method of measuring the GBT surface be ready in case holography was to fail and to confirm the results of the measurements.
The rest of this report gives my initial guess at what will be needed before
the pre-June 1995 test measurements and the November 1995 measurements. Since
preparation time is short and decisions need to be made quickly, I have had to
generate the list of requirements hastily. Warning: the list is FAR from
complete and one should expect that further requirements will surface. I give
only the broad details of what will be needed by these dates since the fine
details would make this preliminary report excessively long.
IV. Parameters for Additional Holographic Measurements
The new requirements require holographic measurements accurate enough to measure the placement of individual panels of the GBT to an accuracy of at most a few mm with initial surface errors expected to be at least 3 mm and maybe much more. Since much of the material presented in GBT memo 68 also applies to the newly-requested holographic measurement, I advise that readers of this report should first familiarize themselves with the contents of memo 68.
As in memo 68, we must ask whether we want to make the measurements from prime or Gregorian focus. Prime focus has the advantage that we will measure the primary surface directly while, from Gregorian, we will measure the combined deviations of the secondary and primary. In memo 68 we decided on doing the observations from the Gregorian focus since the measurements were intended to tweak an already reasonably-good surface and what really counts at high frequencies are the combined errors from both primary and secondary surfaces. However, since the November 1995 holographic measurements will be used to virtually weld the primary in place, we probably should make sure we have a well-shaped primary. Thus, I think we should make the measurements from the prime focus, which will require adapting the holographic receiver we were going to use at the Gregorian focus for use at the prime focus. This will entail a new feed and probably some repackaging of the receiver components.
The large deviation we expect for the original surface is a significant fraction of the wavelength we were intending to use for holography (25 mm). Most holographic measurements of other telescopes have been made where the ratio of observing wavelength to initial surface errors is much higher than in our case. Any area of the surface that deviates by more than 12 mm (half the observing wavelength) from the average will produce phase wrap-around in the holographic map. Sometimes we can correct for phase wrap-around but we will have difficulties if, for example, the areas with large deviations were many, small, or randomly distributed across the dish. Since 12 mm is only a few times the expected minimum error of the initial surface we should rethink the wavelength at which we want to make the measurements. For example, we may want to observe at 75 mm, the wavelength of the transponders aboard C-Band geostationary telecommunication satellites.
The need to measure the setting of individual panels translates roughly to a map consisting of at a minimum 10,000 points (about five points per panel so we can fit for panel tilts and offset) and double or triple this number would be well worth our efforts. If I assume 20,000 points, the holography maps will be about 150 by 150 pixels in size where pixels in the image plane must be a little less than a beam width apart. If we observe at 25 mm the images will be 2.5° by 2.5° in size and at 75 mm will be 8° by 8°.
Note that the geostationary satellites that transmit at 25 mm are uncommon and, for previous holographic work on other Green Bank telescopes, we have had no problems finding a satellite that has no neighbor within many degrees that has a transponder at the same observing wavelength. However, satellites transmitting at 75 mm are very common and I anticipate problems finding a satellite that is sufficiently isolated (spatially or in frequency) if we choose to do the observations at this wavelength.
If we use geostationary satellites with transponders at wavelengths of either
25 or 75 mm, the most significant source of rms noise in the holographic map
will not be source strength but probably the pointing of the telescope. If we
observe at 25 mm, to achieve 1 mm rms accuracy in the holographic map will
require that relative pointing errors (as described in GBT memo 68, section 12)
be under 15" across the 2.5° by 2.5° map. If we observe at 75 mm, the
acceptable errors are 45" over an 8° by 8° area. [I suggest someone check
these numbers since much of the planning hinges on them.]
V. Changes in Hardware Requirements
As stated above, we probably should perform the measurements at prime focus. If we observe at a wavelength of 12 mm, we will need to build a new feed and maybe repackage the receiver we were intending to use at Gregorian focus for prime focus. We can continue to use the reference receiver that we were anticipating using at 12 mm.
If we decide to observe at a wavelength of 75 mm then not only would we have to build a new feed and do some repackaging of the Gregorian receiver we would also have to build the reference receiver for 75 mm observing. We would also need to reinvestigate where we would want to mount the reference receiver because the reference receiver will need to have an unobstructed view of the sky that is much larger than we originally anticipated. No matter what wavelength we use, we must measure beforehand the phase stability of the receivers and the phase and amplitude characteristics of the feeds.
Roger Norrod can generate accurate estimates of the costs of the changes. I
crudely estimate that if we observe at 12 mm the extra costs should be small
[one month full-time employee (FTE), a feed, and maybe some hardware]. The costs
for 75 mm observing would be much higher because of the extra receiver. We must
also anticipate the extra costs in adapting and testing the equipment on the
140-ft. I see no extra costs involving the already-built holographic backend but
we have the additional costs of providing by November 1995 most of the
instruments for continuum observations of astronomical sources (e.g., continuum
backends, receivers, L.O. and I.F. systems). We also must anticipate that NRAO
should mount its equipment on the GBT during the months before November 1995.
VI. Changes in Software Requirements
Luckily I see no major changes from what was planned in memo 68 in either the data collecting or analysis software. However, in some cases we will need the software about a year before we were anticipating needing it.
By June 1995, the monitor and control group must prove on the 140-ft that they can slew a telescope in the non-rectilinear coordinate system dictated by the holographic measurements. They need not worry about spectral-line or pulsar observing but they need to have ready all of the abilities associated with continuum observations. Since we can anticipate not knowing exactly how we should conduct the experiments on both the 140-ft and GBT, we will need a good, flexible, powerful, and quickly adaptable user-interface to the monitor and control software.
I have looked over the holographic software currently in AIPS and find it uses many simplifying assumptions that we would have to eliminate for our experiment. We will need near real-time processing of data since we must catch problems as soon as possible. Beyond holographic analysis, we will need many standard tools for reducing continuum data so that we can find, for example, pointing coefficients, telescope gain, and radial and lateral focus curves (see GBT memos 58 and 68). All this software must be usable by June 1995.
Again, I am not the one to gauge how the acceleration of software development
will increase the cost to the GBT project. Is staffing sufficient for producing
what we will need by June 1995? Can priorities be rearranged if staffing is
VII. Changes in Requirements on Other GBT Systems
As stated above, we will need to make measurements of various telescope characteristics. The success of these measurements is critical in providing accurate-enough holographic measurements.
We must measure the pointing of the telescope and produce accurate enough coefficients for what has been called the traditional pointing model (see GBT memos 103, 105, 110, and 112). We must compensate for atmospheric refraction (memo 112). Similarly, we must know how to move radially and laterally the receiver to follow the motion of the prime focus as the telescope moves through the holographic map and as the temperature of the structure changes. As much as possible, we should test the software and equipment for making these measurements on the 140-ft during or before the June 1995 tests. The groups in charge of the measurements may have anticipated a latter date for completing their systems so we should anticipate extra costs in accelerating the development of the systems to meet the June 1995 deadline.
Some aspects of the GBT cannot be tested before November 1995 when we will have our first chance using the telescope. For example, we have the choice of slewing the telescope during the holographic measurements either in azimuth or elevation. We need to pick the slew direction and speed for which the telescope behaves the best. We need to know the telescope's dynamic response to motion and winds so that we can anticipate settling times. Thus, not only would it be important to have beforehand some estimates of the telescopes dynamical behavior we must also confirm these estimates by measuring the telescope's dynamical behavior before the November 1995 holographic measurements. Again, NRAO may incur extra costs to prepare earlier than expected some GBT systems.
VIII. Changes in Personnel Requirements
Originally, the holographic measurements were meant as a research project that was to be started after the GBT was in operation. By that time, we anticipated having in Green Bank a few staff members available paid out of operation's budget whose job would entail working the bugs out of the GBT system. We anticipated these already-available 'friends-of-the-GBT' would perform the originally-planned holographic measurements described in GBT memo 68.
Any holographic experiment requires the joint expertise of a few people who can discuss many small but important details concerning the experiment. For example, for the holographic measurements I made of the 140-ft telescope in 1987 through 1988 I had two individuals who acted as consultants and workers who were part-time members of scientific services. I also had lots of assistance from members of the engineering and programming staff in planning the experiment and in doing some work. Without these people, the experiment would not have been successful.
I anticipate that to plan for and execute the 1995 holographic measurements we will need two FTE working 25% to 50% of the time preparing for the holographic experiment. The individuals must have already performed holographic measurements of telescopes (i.e., we cannot afford the luxury of on-the-job training with so critical an experiment). We need at least two people so that they can exchange ideas. Both should be staffed in Green Bank since they will need to interact heavily with electronics staff and the various programming groups (AIPS++ and monitor and control). Both must start work as soon as possible because of the amount of necessary planning and the short time scale we are unfortunate to have. Eventually, these individuals will be so well versed in the behavior of the GBT that they will be well suited to become 'friends-of-the-GBT' once the telescope is in routine operation.
IX. Alternatives to Holographic Measurements
We have a large choice of alternative methods we could use to measure the surface of the telescope. Besides holographic measurements, we could use traditional mechanical techniques (e.g., theodolite and tape measure), newer electrical distance measuring devices, the laser-ranging system that is under development, or photogrametery. We require that any method must measure the placement of actuator brackets to an rms error of at most a few mm.
Holographic methods differ from these other methods in some very important ways. One cannot use holographic experiments to measure part of the dish and the telescope must be capable of astronomical measurements. Thus, we cannot measure the placement of actuator brackets until all brackets are welded in place and all surface panels are installed.
If we choose holography and if we find any areas of the surface that are in error by more than 3 mm, then we must come up with a way to adjust these areas. The welding process makes the 2-inch bracket adjustment screws unusable unless the welds are broken. If we don't want to go through the trouble of breaking the welds, we can use the panel-to-panel adjustment mechanisms, with a usable range ħof 3 mm and accessible from the top of the surface through holes. We possibly can reset the nuts at the bottom of the bolt that makes up the panel-to-panel mechanisms. The latter adjustment, however, can only be performed from underneath the surface but can produce about 6 mm of adjustment in only the upward direction. Thus, if we don't want to break welds, we are limited in the range or direction of adjustments we can make and, in some cases, the adjustments will be difficult to make.
The alternative measuring methods I listed can be made without having to take
a single astronomical observation and with the telescope at any convenient
orientation. The telescope need not move to make the measurements and,
therefore, actuator brackets don't have to be welded in place. Thus, we probably
can still use the 2-inch bracket adjustment screws to correct for any badly-set
areas of the surface. The alternative methods, in comparison to holographic
methods, probably provide a cheaper, faster, less labor intensive way of setting
the surface to our requirements.
I think it is possible but risky for us to use holographic measurements as a way to help RSI determine how to mount and adjust the actuator brackets of the GBT. To reduce the risk we must anticipate having to perform months beforehand holographic measurements on the NRAO 140-ft telescope. Even then, the risks might be high enough that we might want to use another measuring technique to corroborate the holographic measurements of the GBT.
Costs to NRAO and maybe RSI might be higher than we have been anticipating because many GBT projects have to be accelerated. There will be some extra, non-trivial hardware and personnel costs, as described above.
Holography hinges on being able to outfit the GBT for astronomical observations at an early date and being able to move the telescope to make a large suite of astronomical and holographic measurements. If we use holographic measurements, we are forced to weld the actuator brackets beforehand and, thereby, eliminate the cheapest and most versatile way to adjust the brackets if any are found to be in error.
Other methods of measuring the surface, unlike holography, do not require moving the telescope or making any astronomical observation. The actuator brackets need not be welded and we possibly could use the already-available, versatile bracket adjustment screws to fix quickly and easily any errors.