Tilt sensor measurements.

Links to plots:pdf

Links to sections:

10azdec07: azswing after tiedown refPosition updated.
10azdec07: azswings after tiedown nuts adjusted.
07dec07: azswings after painting and adjusting td using 06dec07 data
06dec07: azswings after painting (before td adjust)
07apr07: azswings, adjust tiedowns after more lead removed.
04apr07: azswings, adjust tiedowns,azswings after lead removed.
31mar07: azswings after laser shack removed
27mar07: azswings before start of painting project
01dec06: 4 azswings.
28feb03: azswings and za strips (after dome move)
25feb03: azswings and za strips.
20feb03: azswings and za strips.
19feb03: azswings and za strips.
18feb03: azswings and za strips.
17feb03: za strips.
15feb03: azswings and za strips.
12feb03: Comparing za strips 5feb03 with 12feb03 (before after first shimming).
05feb03: azswings and za strips.
04aug01: azswings and za strips

Introduction (top)

    The offset of the tilt sensors with respect to the floor can be calibrated by rotating the turret floor by 180 degrees and then averaging the two measurements. The tilt sensor values can be used for relative angle changes. For absolute angles of the dome relative to the optics of the telescope, theodolite measurements from ao9 can be used to calibrate the tilt sensor offsets.

Tilt sensor angle with respect to the turret floor (measured using floor spins).  (top)

Measuring the 1,3 az terms with azswings at 10 deg za.  (top)

• 16dec02 2 azwings showing data and fits. The azimuth swings were done during the day with temp=76.6.  The colors black,red show the data minus (c0 +c1*az) for the two swings. The green is the fit residuals. The data has not been corrected back to the theodolites so the constant term is not the true tilt of the dome.
processing: x101/Y02/021216/doitts.pro
• 09sep03 2 azwings showing data and fits. The azimuth swings were done during the evening  with temp=75.  The colors black,red show the data minus (c0 +c1*az) for the two swings. The green is the fit residuals. The data has not been corrected back to the theodolites so the constant term is not the true tilt of the dome.

• 11sep03 2 azwings showing data and fits. The azimuth swings were don
• he new position was computed by:
•  removing the 1 az term from the first two spins.
•  reinstalling the 1az term that was measured on 27mar07.
•  Back in dec06, pointing errors showed a 1 az term of about 9 arc seconds. I could not measure this small amount reliably with the tilt sensors so it modified the 1az term in the pointing model. This left a residual tilt in the platform of the .006 degrees pitch that was measured on 27mar07. Since i wanted to keep the same pointing model, i had to reinsert the .006 degrees in pitch.
• The pitch, roll measured after the tiedown adjustments were within about .003 degrees of the 27mar07 values. This is about as good as we can do with the tilt sensors.

• e during the morning with temp=78.  The colors black,red show the data minus (c0 +c1*az) for the two swings. The green is the fit residuals. The data has not been corrected back to the theodolites so the constant term is not the true tilt of the dome.
processing: x101/030911/doitts.pro
• 01dec06 4 az swings showing data and fits (.ps) (.pdf):  The az swings were done between 6 and 8 am. Some recent pointing tests had shown a 1 az term of about 9 arcseconds in the pointing residuals. The az swings with the tilt sensors were done to verify that the 1 az term was coming from a residual platform tilt. The rm's of the fits were .005 and .006 degrees (this is typically the rms's that we;ve been getting). The amplitudes of the 1 az terms were between .001 and .005 degrees so they are consistant with 0. The 9 arcsecond 1az term is probably a real tilt, but we can't measure it with the tilt sensors. It's probably a better idea to use the pointing residuals to update the pointing model rather than to try and redefine the tiedown reference position.
• 27mar07 2 az swings before painting project started (.ps)  (.pdf). The az swings were done between 11:30 am and 12:00 noon. This is a reference set of data. If a large 1 az term comes up during the painting, we can correct it back to this value (which should be what the model wants).
• 31mar07 2 az swings after laser shack removed (.ps) (.pdf). The azimuth swings were done between 16:00 and 17:00. The laser shack had been removed earlier in the day from corner 4. It weighed about 1600 lbs.
• 04apr07 2 azswings, adjust tiedowns, then 2 more azswings: 10000 lbs of weight had been removed from corner 4 and about 5000 lbs had been removed from corner 12. The az swings were started around 16:00.
  
• 2az swings before the tiedowns were adjusted(.ps) (.pdf ):
• The tiedown positions used were: [14.165 13.905 12.522]. This was the standard position.
  
• 2az swings after the tiedowns were adjusted(.ps) (.pdf ):
• the tiedown positions used were:[13.9707 14.6907  11.9306 ].
  
• The new position was computed by:
•  removing the 1 az term from the first two spins.
•  reinstalling the 1az term that was measured on 27mar07.
•  Back in dec06, pointing errors showed a 1 az term of about 9 arc seconds. I could not measure this small amount reliably with the tilt sensors so it modified the 1az term in the pointing model. This left a residual tilt in the platform of the .006 degrees pitch that was measured on 27mar07. Since i wanted to keep the same pointing model, i had to reinsert the .006 degrees in pitch.
• The pitch, roll measured after the tiedown adjustments were within about .003 degrees of the 27mar07 values. This is about as good as we can do with the tilt sensors.

• 07apr07: 2az swings before the tiedowns were adjusted(.ps) (.pdf ):The rest of the lead was removed from corner 12. The tiedowns were put back at their 27mar07 reference and then two az swings were done. The 1 az term was computed and the tiedowns were then moved so that the 1az term was the same as 27mar07. No az swings were done after the tiedown motion. The tiedown positions used were:
  
• [14.165 13.905 12.522] 27mar07 tiedown ref from 27mar07 used for the az swings.
• [14.9277,15.4312 12.3754] new tiedown reference.
  

• 06dec07: 2az swings after painting (no lead). dome at 12.02 degrees (.ps) (.pdf): Two az swings were done in rate mode on 06dec07 at 23:00 hours with the temp at 71 F. The dome was at 12.0265 and the ch was at 6.9778 deg. The tiedown were at the ref position computed apr07 after the lead was removed: [14.218,14.722,11.66] inches. The only difference between apr07 and dec07 is that the contractors weight was now off the platform.

• 07dec06: 2az swings after updating the td position from the 06dec07 data (.ps) (.pdf): Two az swings were done in progtrack mode on 07dec07 at 19:00 hours with the temp at 73F. The was at za=10 and the ch was at stow. The tiedown positions were updated to remove the residual 1 az term from the 06dec07 data. Alfa was not yet on the turret floor.
  

• 10dec07: 2 az swings after adjusting the tiedown nuts (.ps) (.pdf): Two az swings were done on 10dec07 at 20:00 with he temp at 70F. The dome was at za=10 and the ch was at stow. The  tiedown nuts had been moved on 09dec07. Alfa had also been installed on the floor.
• 10dec07: 1 az swing after readusting the tiedown jack positions. (.ps) (.pdf) : The 2 azswings taken 10dec07 were used to compute the final adjustment to the tiedown reference positions The measured values were:

  	tdReference Used [td12,4,8]	Pitch 1azAmp[Deg]	Pitch 1azPhase[Deg]	Roll 1azAmp[Deg]	Roll 1azPhase[Deg]
  before	[15.114,15.203,15.167]	.0028	24.	.0024	179.
  after	[15.072,15.320,15.093]	.0013	107.3	.0055	171.2

  
  

Azswings and za strips with the tilt sensors (description).  (top)

    These measurements have offsets do to the alignment of the tilt sensor with the uphill direction. This misalignment can also cause a linear error term in roll. Other components (1az, 3az, etc) are accurately measured. The figures are:

Pitch roll versus za:

• Fig 1 Top. The  plot shows the pitch vs za. A linear fit to za has been removed (since the pitch should increase linearly with  za). The constant offset of (typically -9.7xx) is the za where the turret floor is level. The linear term (close to 1) measures any linear pitch error plus any scale factor in the encoder. The encoder measures a linear distance along an arc (and assumes a particular radius) while the tilt sensors measure an angle. There is a .02 deg 3 azimuth term in the pitch. The za strips are spaced every 60 degrees and their phase is close to the minimum and max of the 3az term. The red lines are the azimuth spins at constant za.
• Fig 1 middle. This is the roll versus za. The two black traces are samples of the 3az term (.04 degs amplitude) close to the min and the max. The red lines are the az swings. The large linear ramp is caused by the misalignment of the roll tilt sensor relative to the uphill direction (not exactly 90 degrees).
• Fig 1 bottom. This plot (if available) shows the roll after removing a linear fit to za.  The fit will remove any offset do to the misalignment of the tilt sensor AND any roll vs za that is in the rails..

Fitting pitch, roll to the azimuth swings.

Each azimuth swing was fit to the function f(az)=c0 +c1*az + c2*sin(az-c3)+c4*sin(3az-c5). The linear az term removes any time variation of the platform (do to temperature changes).
• Fig 2. The pitch of each azimuth swing after removing the constant and linear term in az (for each swing).
• Fig 3. The roll vs az for each az swing. The fits are over plotted as dashed lines.
• Fig 4. Pitch fit residuals (data-fit) vs za and azimuth
• Fig 5. Roll fit residuals  (data -fit) vs za and azimuth.

Fig 6,7. The coefficients from each swing plotted versus za.

Comparing za strips 5feb03 with 12feb03 (before after first shimming).  (top)

• Fig 1 top. For each day, a linear fit to pitch was removed from the data and then plotted vs za. The 05feb03 data are the black dots (za strips) and red dots (az spins). The green data are za strips from 12feb03 after the first shimming.
• Fig 1 bottom. This is the difference in the za strips pitch for 05feb03 - 12feb03.
• Fig 2 top is the raw roll data for the two days.
• Fig 2 middle plots the roll data for each day with a linear fit removed.
• Fig 2 bottom is the za strip difference in roll 05feb03-12feb03.

    The roll difference is about .1 degrees about 10 za. It has also changed at low za. The shimming on the ch side should only
affect the roll of the dome up to za of 3 deg. The roll difference has  a linear term of about .15 degrees 0 to 20 za. You would need a change in tilt sensor alignment on the turret floor of about 2 degrees to get this kind of ramp .15*(1-cos(theta))*za.
I need to calibrate the tilt sensor alignment on the floor with the theodolite data to see what the true roll is doing.

Calibrating the tilt sensor data to the theodolite measurements (top)

    The calibration uses tilt sensor that is taken during the ao9 survey. The procedure is:

1. The floor offset of the tilt sensors is removed from the tilt sensor data.
2. The 1 az term is removed from both the tilt sensors and the theodolite. This is needed because there are horizontal offsets of the theodolite data from the center of the reflector that the tilt sensors are not sensitive to. This limits tilt sensor calibration to  ao9 surveys with az swings.
3. A linear fit is done to theodolite - tilt sensor  for pitch and roll

 

1. Remove the 1 az term from the tilt sensor data. It will use the 1az term of the tilt sensors measured during the ao9 survey, or you can pass in the 1az term of the data set.
2. add the linear fits from the above steps.
3. add back on the 1az term of the theodolite.

    The 1az terms used are shown in the table below:
 

1az terms    theod: Acos(az)+Bsin(az)  Tilt: Asin(az-phase)

date	Theod Pitch  cos,sin	theod Roll  cos,sin	tilt Pitch  Amp ,Phase	tilt Roll  Amp,Phase
09aug01	-.02696,   .00821	-.01316,-.03221	.00263,343.6 deg	.00172,344.5 deg
17feb03	-.02722,-.00337	.00107,-.022477	.00186,332.1 deg	.00105,20 deg

    The results of these calibrations can be found under the ao9 survey results.

home_~phil