Several frequency sweeps of the azimuth and the altitude drives for the 2.5m telescope were taken. The data appears to be reasonable. An attempt at measuring azimuth friction appears also to give reasonable results. Attempts were also made to measure wind disturbance. No useful information has come from the wind measurements so far. We were unable to make measurements on the rotator drive.
Before making any measurements, it was necessary to clean the surfaces of the azimuth drive disks, since an in situ grinding of the rotator disk had taken place the previous week. It was also necessary to finish the installation of the counter weights since the telescope needed to be balanced. With instructions and help from Jon Davis, Angela Prosapio was able to wriggle into the rather cramped area of the azimuth disk and remove the motor oil preservative and any grinding dust embedded in it and replace it with a thin layer of WD 40 (A very thin oil). In order to clean the area directly under the drive capstans, it was necessary to rotate the structure 30 degrees or so. With instructions by remote control from Charlie Hull in Seattle, the wind screen was rotated along with the telescope. The drives were cleaned up by Friday afternoon. It also appeared that we could get a little more azimuthal motion between the telescope and the wind screen if the wind screen floor plates were moved back a bit from the telescope. This did in fact give us about a third more motion.
We started with this list more or less in order of priority:
The instruments used were an HP 3563A system analyzer to produce the drive waveform and record the response in either the time or frequency domain, a Wilcox accelerometer with a magnetic base for acceleration, (I didn't realize until I saw the data sheet that this device has a velocity mode as well) a Kepco bipolar power amplifier to drive the servo motors in voltage mode, a Tektronix current probe for measuring motor current, and a gain of 20 instrumentation amplifier for the tachometer signal. There was a Kollmorgen tachometer installed on the drive shaft of one azimuth motor. The tachometer was specified as generating five volts/rad/sec. The reason for driving the motors in voltage mode is the thought that perhaps the telescope would have less tendency to wander during the sweeps because of the damping effect of driving with a low impedance. Torque is the input parameter of interest however, hence the use of the current probe. A laptop PC was used to extract the data from the HP analyzer onto disk.
The biggest worry for me was the possibility of the structure slowly wandering during the sweep. This turned out in practice not to be the case. It was apparent after first driving the azimuth that the constraints at right angles to the azimuth preload springs were loose, allowing the azimuth motors to rock back and forth. It turned out that the majority of the bolts holding the constraints in place were loose. After going around and tightening four sets of bolts the motor swaying ceased. The structure was very stable and a sweep could be run for an arbitrary length of time. I chose a sweep resolution setting of 133 points/decade which took about 10 minutes to run to completion. At the maximum output voltage of the power supply (43 volts pk) the excursion of the telescope would hit the wind screen at a frequency lower than about 0.2 Hz. This frequency is above the corner frequency of the structure. By 0.2 Hz the phase is already -90 degrees. I choose a three decade span from 0.2 to 200Hz for the frequency sweeps. We set up a pointer made from a gas welding rod an attached it to the end of the structure and taped a ruler to the floor to give us an idea of the telescope position during the sweeps. By Friday evening, we had two or three sweeps, including a one hour run just to check on the degree of fine structure of the plot. Examining the data later that evening, it looked odd. There seemed insufficient attenuation of the magnitude plot at higher frequencies.
On Saturday morning, I checked the setup, and the tach amplifier output had a large offset (200mV). I had no ground reference with very high impedance inputs, and the tachometer has essentially infinite impedance to ground. The CMOS input leakage of the amplifier would over a period of minutes charge up the windings and associated wiring of the tach, resulting in many volts of common mode voltage. One 100K resistor from each input leg of the amplifier to ground cured the problem. There was now about 2mV of offset at the tach amplifier output. We then ran the azimuth sweeps again and the results now looked much more believable. We also did sweeps with only one motor (the one on which the tach was not mounted) to see if there were any magnetic coupling effects between the motor and the tach (there appear to be none).
Saturday during the day the weather was marginal, and Jon Davis was uncomfortable with the idea of uncovering the building. We did azimuth sweeps at 0 and about 25 degrees within the building. We also did an accelerometer run at 0 degrees. It is evident the accelerometer in velocity mode has 0.5Hz corner frequency. Since we had done most of the sweeps we could inside the building, I tried to measure the friction. The azimuth was driven with a very low frequency sine (39 mHz) at a voltage just enough to break the static friction. I first tried 10 mHz, but when break away occurred, the telescope motion was too large and it hit the wind screen. After 2 or three attempts at 39 mHz we got a record free of windscreen collisions. The record is in the time domain and contains a tach voltage record. On Saturday evening the weather cleared and we attempted to move the building, but I was intimidated by the Klaxon and decided to wait until Sunday. I plotted the magnitude data for a few sweeps Saturday night, and they looked much better than those from Friday.
On Sunday morning Charlie Hull showed us how to move the building. In the course of the morning we did azimuth sweeps at 90, 60, and 45 degrees of altitude. I noticed during the course of moving the telescope around what appeared to be a pronounced torsional resonance of the secondary weight disk. To check it we mounted the accelerometer to the disc tangential to the edge. The torsional resonance of about 9Hz clearly showed up when the telescope was driven in azimuth at 90, 60 and 45 degrees.
At this point I decided it was time to move the tach from the azimuth to the altitude axis. Jon Davis crawled into the nether azimuthal regions and removed the tach from the altitude. While trying to remove the bearing preload plate from one altitude drive, the drive disk slipped. Jon checked the slip torque with a torque wrench and it read 20 lb ft. If we assume a drive disk diameter of 4" and a coefficient of friction of 0.1 the implication is that that altitude drive has about 600# of preload. Jon was able to get the tach mounted, although he noted the apparent strength of the permanent magnet was much less than when he had first mounted it a year ago. Now any regrets about missed azimuth measurements were by definition passed. The first altitude measurement was a sweep with velocimeter mounted on the secondary. There was no torsional resonance evident when the altitude axis was driven. We had time to do tach and accelerometer sweeps of the altitude at 0 degrees.
Sharon Lackey arrived Sunday night at explained that she would be unable to complete much of her planned work, so that we could have access to the telescope for much of Monday. This then was gravy day. The weather was very fine, but unfortunately for us almost windless. Our first sweep was an altitude run at 90 degrees. We spent the rest of the day trying to get some wind disturbance data with marginal success. Runs were taken in the time domain with a 5Hz dithering signal sent to the servos to break the static friction. For many of the runs the voltage to the motors was 150 mV which is clearly not enough to break static friction. This small amount of drive was visible both in the accelerometer mounted at the end of the structure and the tach. The implication the dither is coupled through the compliance of the structure. By late Monday we had decided to call a halt and took down the equipment, replaced the rotator floor and put the building back. Angela and I packed the instruments back into their boxes in preparation for shipping them back to Fermilab.
My thanks to the APO staff for their help and assistance and particularly to Jon Davis for his long hours of service on our behalf.
Last modified 02/23/99