From pmuirhea@umich.edu Thu Sep 23 09:11:51 2004 Date: Thu, 29 Jul 2004 16:18:40 -0400 From: Philip Muirhead To: John Monnier Subject: Summary of actuator experiments Dr. Monnier,   I know I sound like a student and not a professional, but forgive me - I haven't had Astro 429 yet:   Testing the step sizes of the X and Y components, we set up a basic Michelson interferometer with a laser at 633 nm.  The actuator controlled one of the differential mirrors which when moved changed the interference pattern and thus the intensity hitting a detector.  This intensity was converted to an amount on a 0-4095 scale by a 12-bit data acquisition card. The noise from vibrations and relaxations in the set up was reduced by floating the table on pistons designed to absorb building vibrations, but relaxations in the mirror on the actuator itself were evident every time the actuator was moved.  These could change the intensity by a fraction of the full intensity of the laser (500 to 800 on the 12-bit scale of the DAQ card, when the beams were perfectly aligned).  That should be kept in mind when reviewing these numbers.   The tests went basically like this: the axis inline with the beam would be moved 1 step every half second to cover a small number of wavelengths as read by the DAQ card, each wavelength being half the laser's, or 316.5 nm.  For Z, 12 steps were used, assuming each step is around 100 microns (~4 wavelengths), and for X 160 steps were used assuming 1 step is around 4 nm (~2 wavelengths).  A series of these sets was taken moving both forward and backwards (5 each way for X, 9 each way for Z).   These sets were then fit with a sin curve with all the appropriate parameters - wavelength, amplitude, phase and baseline - where these parameters were optimized for a minimum chi-squared using Microsoft Excel's Solver function.  The step size was then found from the fit wavelength. Here are the results:   Z moving forward: +- stand_dev = 104.5790 +- 17.2121 nm   Z moving backwards: +- stand_dev = 73.6434 +- 6.7508 nm   X moving forward: +- stand_dev = 4.393190 +- 0.430377 nm   X moving backwards: +- stand_dev = 0.913341 +- 0.913341 nm    Keep in mind these stand_devs are not estimates of the error due to the minute motions in the set-up components, but just the deviation in the results from the individual trials.   Testing the reproducibility of location, the set-up consisted of laser light focused into a single-mode optical fiber held onto the actuator platform by a weighty fiber positioner (~half a kilogram, with the maximum weight specified by Luminos as 1 kilogram).  The fiber led into a detector where the intensity was once again recorded using a 12-bit DAQ card. The Z-axis was placed inline with the optical axis, with X and Y as coordinates in the optical plane.  The fiber was manually moved to a position of high intensity, then a square was mapped out moving the fiber across X in steps of 100 (~400 nm) and incrementing Y 100 steps before moving across X again.  In the end, a 5000 x 5000 step range was mapped out ( 20 microns x 20 microns, delta .4 microns). The resulting image of the map was fit to a two-dimensional Gaussian to clear out the noise, which was significant near the location of maximum intensity, or the center of the laser light's focus.   The peak of the 2D Gaussian was recorded, and then the actuator was moved to a distant location and brought back to the same Z-position for another mapping.  Four maps were taken with the drastic move of the actuators in between (for one particular Z-location).  Two Z positions were used.  Here are the standard deviations in the positions of max intensity:   Homing all actuators in between mapping:   First Z position: stand_dev in X = 8.9 steps = 0.0356 microns stand_dev in Y = 81.8 steps = 0.3272 microns   Second Z position: stand_dev in X = 11.7 steps = 0.0468 microns stand_dev in Y = 133.9 steps = 0.5356 microns   Moving all actuators to absolute position {1,1,1} (to avoid home switch):   First Z position: stand_dev in X = 10.4 steps = 0.0416 microns stand_dev in Y = 37.6 steps = 0.1504 microns   Second Z position: stand_dev in X = 12.6 steps = 0.0504 microns stand_dev in Y = 9.8 steps = 0.0392 microns   Moving all actuators to absolute position {128000,128000,128000} (~max range):   First Z position: stand_dev in X = 10.4 steps = 0.0416 microns stand_dev in Y = 86.0 steps = 0.344 microns   Second Z position: stand_dev in X = 12.2 steps = 0.0488 microns stand_dev in Y = 94.6 steps = 0.3784 microns   This is assuming that the maximum intensity position relative to the lab table does not change despite the noise.  I am tempted to think that the higher variations in the Y may have something to do with the weight of the fiber-holder/manual-positioner sitting on the platform.  To test whether triggering the home switch had any extra effect on the reproducibility, the actuators were brought right up to the home switch without triggering it at {1,1,1}.  There is no major change in the X variation, but it is much smaller in the Y.  Perhaps the weight pushes a little harder on the home switch of the Y axis.  Although, the Y variation is just as high when brought to 128000 and back.     Whenever it's convenient, let me know what else I should include and any changes you think I should make . -Phil