pointing error when tiedowns lose tension
27may04
Background:
The platform is run in an unbalanced configuration.
As the dome moves up and down in za, the entire platform tilts. The elevation
rails have been shimmed so that this weight imbalance causes the horns
to tilt onto the paraxial surface. This should be repeatable as long as
the weight in the dome and the temperature remains constant.
If the main cables are the same length, then temperature
changes should make the entire platform move up and down without any extra
tilting. We use the tiedown cables to compensate for the motion of
the platform due to temperature. The distomats measure the average height
of the platform every 2 minutes. If the average platform height is more
than .04 inches from the nominal position, the tiedowns will pull (or release)
equally on all 3 corners to bring the average height back to the nominal
position.
Losing tension in a tiedown cable.
The tiedown cables can lose tension as the dome passes
in front of them. This occurs more often at high zas and high temperatures.
The program controlling the tiedowns does not use the td tensions to monitor
this condition. It continues to control the average height
of the platform by moving all 3 tiedowns the same amount. With tension
lost in one tiedown this operation will tilt the platform and cause a pointing
error.
On 27may04 the source 3C48 (J0137+331) was used
during a calibration run with sbw. The source was tracked from 9:00 AST
to 10:30 AST. The temperature was about 84 F. The plots show the
pointing error caused by the tension loss in tiedown 4 (.ps)
(.pdf)
(az=122.87 degrees).
-
Fig 1 top: This is the tension in the tiedowns during the run. The
tiedowns are color coded: green is td 12, Blue is td 4, and purple is td
8. There are two cables at each tiedown block. The tensions in the 2 cables
at each td should be the same. Differences are caused by calibration errors
of the load cells in each cable. The gradual motion of the tensions is
caused by the dome and azimuth arm moving while tracking the source. At
10.2 hours, td 4 (the blue lines) stop changing. These tiedown
cables have gone slack (the 5 and 7 kip readings are calibration
errors).
-
Fig 1 second: This plots the pointing error versus hour. The black
is the azimuth error and the red is the zenith angle error. When
td 4 loses tension (10.2 hours), the azimuth and zenith angle errors start
increasing.
-
Fig 1 third: The azimuth angle during the track. Td 4 lost tension
at an azimuth of 156 degrees. The azimuth of td 4 is 122.87 degrees.
-
Fig 1 bottom: The zenith angle during the track. Td 4 lost tension
at a zenith angle of 16.4 degrees.
-
Fig 2 top: The platform temperature during the track.
-
Fig 2 bottom: The plot shows the (Average platform height
- nominal platform height) in inches. At 10.2 hours, td 4 went slack
and the platform began to drop. At 10.6 hours we lost the reading from
one of the distomats. When this happens the control program switches to
use the last recorded platform height and temperature. Every two minutes
it reads the temperature and computes the delta temp from the last good
reading. The delta temp is used to compute how far to move to get back
to the nominal position (using .2 platform inches/deg F and 1.73 td inches/platform
inch). This causes a jump in the tiedowns and since the platform was not
at the nominal position when the last reading was taken.
Losing tension in a tiedown causes the average height
of the platform to be too low. A more serious problem is that the tiedown
control algorithm will cause the platform to become tilted. The tilt is
a pointing error that can be much more serious than the gain loss from
being out of focus. For 3C48 the pointing error was 90 asecs (one beam
at 3500 Mhz).
This problem is occurring more frequently as we
continue to add weight to the dome. Adding 4000 lbs to the dome should
cause the platform to tip by .4 inches at za=20 degrees. This will move
the tiedowns by .7 inches.
We need to find a reliable way to monitor the loss
of tension in a cable so that the control algorithm does not make things
worse.
Some other pages:
processing: x101/model/misc/tdLoseTension_may04.pro
home_~phil