Intro:

• 170920: Eye passed over ao 11:30
  
• wind speeds, rain, etc
  
• tilt sensor in dome during hurricane
  
• tiedown oscillations during hurricane.
  
• 171017: az swings with tilt sensors to see if things have changed.
  
• 171130: pointing model installed right after maria.
  
• 180227: telescope gain variation pre/post maria
  
• 180501: compare sband narrow gain curves pre/post maria.
  
• 200519: steps to realign optics after maria

Tilt sensor data during hurricane maria

    The tilt sensors are mounted on the rotary floor.

•  The pitch tilt sensor is aligned along the azimuth arm direction when the floor is at 75 degrees
• It has a range of +/- 15 degrees.
• The roll tilt sensor is perpendicular to the pitch sensor. It has a range of +/- 2 degrees.
• The rotatory floor is level at a zenith angle of 10 degrees.
• During hurricane maria, the floor was at 75  degrees. The gregorian was at 8.4792 degrees
• The azimuth arm was bolted to the platform during the hurricane.
• The gregorian dome was pinned to the azimuth arm.
• Tilt sensor data is sampled at 5 Hz.
• Data was taken from 19:38 19sep17 till 18:33 20sep17

The first plots shows the tilt sensor data for the entire time (.pdf).

• Black is the pitch tilt sensor
• red is the roll tilt sensor. it clipped.
• the eye passed over ao around 11:30 on 20sep17 (20.48 daynumber).

The 2nd plot has the tilt values and wind speed,direction for 20sep17 (.ps) (.pdf)

• Top: pitch tilt sensor for hour of day
• middle: Wind speed on platform (3 sec avg) vs hour of day
• bottom: wind direction
• The eye passed over ao around 11:30 (11.5 on x axis).
  

The last plot has the spectrum of the pitch tilt sensor by hour for 20sep17 (.ps) (.pdf)

• The spectra were computed for each hour of the day (till 18:00)
• The fftlen = 5*3600 samples
  
• the spectrum was computed as:
•  spc=abs(fft(tilt)
• keep fftlen/2 points (since real data)
• When plotting smooth by 11 channels to give 11/3600=3 millihz resolution.
• The maximum bandwidth is 2.5 Hz.
• The spectra are plotted by hour with offsets for display.
• 0-1 hours at the bottom (black)
• 17-18 hours at the top grayish blue.
• Some of the strongest frequency  components have been flagged with vertical dotted lines:
• .553 Hz
• .641 Hz (strongest)
• .805 Hz (crosshead?)
• 1.037 Hz
• 1.357 Hz
• this drifted to lower freq as the wind got stronger. It
  
• Any frequencies above 2.5 Hz would alias back down.
  
• You can see the oscillation amplitudes increased until 11:00 when the eye passed. They pick up again around 12:00

tilt summary

• oscillation frequencies measured:

• freq Hz	Notes
  .553
  .641	strongest
  .805	crosshead(?)
  1.037
  1.357	drifted to lower freq

  
  

processing:x101/Y17/170920/domaria.pro

Tiedown tensions during hurricane maria.

    Tiedown cables connect a cement block on the ground to each corner of the platform. There are 2 cables from each corner down to the corresponding cement block. The cables are attached to the block with a 150 ton jack that is computer controllable. The two cables at each corner connect to the jack with a pin/loadcell that measures the tension in each cable.

•  Prior to the hurricane the td jacks are moved all the way up (24 ")  to release as much tension as possible.
  
• The load cell output level is in the millivolt range. An amplifier is used to kick the signal up to a/d levels (volts).
• We've always had trouble with  the stability of these amplifiers.
• There is a calibration resistor the lets you calibrate the amplifier. It has a gain and dc offset adjustment.
• Historically we've found that the gain tends to be stable, but the dc level drifts.
• The a/d used  to measure the signal is set to go 0 to +volts. When the dc level drifts below 0 volts, part  of the signal is cutoff.
• During the hurricane 1 amplifier at each corner had a large dc offset and was unusable.
  
• td12_2, td4_1,td8_2  had non zero values.
• for each td corner  i used the the above tensions , doubling them to get the total kips/corner.
• the tiedown tensions are read once a second.
  

The first plots shows the td tensions for 20sep17 (.ps) (.pdf)

•  The data is sampled at 1 Hz.
• the plots show the total tension on each corner.
• I've removed the median value (since the dc offset differed)
• The bottom portion of the plots are clipped.
  
• the a/d clipped at 0 volts. this was caused by the dc offset < 0.
• Top: td 12
• middle: td4
• bottom: td8
• after 14.5 the value went offscale. I think the cable from loadcell to cabinet had problems
• Td 12 and td4 had similar amplitudes. Td8 were a bit higher. This may have been a amp gain problem.
• You can see the hurricane eye passing around 11.5
• The peak amplitude for 12,4 was around 20kips.
• Since it was an oscillation the tension was probably changing by +/-20 kips

The second set of plots  has the  spectra of the td tension by hour of day (.ps) (.pdf)

• The spectra was computed as spc=abs(fft(kipshr))
  
• I computed the spectra for each td corner separately
• The spectra were smoothed by 11 channels giving 11/3600= 3millihz resolution
• td12 and td 4 spectra were added together
• td8 spectra started to get pretty noisy after 6am so i didn't include it
• 1 Hz sampling gives a maximum bandwidth of .5 Hz.
  
• Any oscillations higher than this will alias back down.
• the spectrum for each hour is plotted with an offset for display
• 0-1 hours at the bottom (black)
• The vertical dotted lines flag some of the stronger frequencies
• .320 Hz
  
• (strongest during the hurricane)
• not present prior to wind picking up
  
• .358
  
• strongest prior to wind
• This is close to the computed oscillation freq of the  telescope.
  
• .450
• present prior to wind.
• The wind died 11-12 when the eye passed overhead.

Td Summary

• The td tensions
• picked up around 6am
• peaked around 10:00
  
• died around 11-12 when the eye passed overhead
• picked up again until dying away aroud 19 hours.
• td12 and td 4 and similar amplitudes with a max of 15-20 kips.
• td8 values were larger. Not sure whether this was real, or a gain calibration of the loadcell amp.
• Measured oscillation frequencies:
• Anything about .5 Hz will alias down

• freq (Hz)	Notes
  .320	strongest oscillation. Only present with high winds. Not present prior to hurricane
  .358	present prior to winds
  .450	present prior to winds. This is not an alias harmonic of .358*2. That would show up at .284Hz.

    
  

processing: x101/Y17/170920/mariatd.pro

td,tilt sensor summary

• The td kips measurements had problems with calibration.
• td12,td4 showed a maximum oscillation of 15-20 kips per td block.
  
• the measured oscillation   frequencies:

• freq	loc	notes
  .320	td	Strongest during winds. Not present with low winds
  .358	td	present without high winds
  .450	td	preset without high winds
  .553	ts	present without high winds, broad during high winds
  .641	ts	present without winds. Strongest with high winds
  .805	ts	only present with high winds (although this is the crosshead frequency for the receivers on the floor).
  1.037	ts
  1.357	ts	broad during high winds.

•  
• .641Hz tilt sensor strongest value is the 2nd harmonic of the td .32Hz 
  
• I don't think the td value is aliased.
• I wonder if the rocking along the azimuth arm is doubling the vertical oscillation freq.
• Could be if the oscillations in individual td cables were out of phase.
  

home_~phil