Foreword A concentrated effort to investigate the source of the "not as good as expected" image quality at the periphery of the MegaPrime field started in early June 2004. With MegaCam now operational in a nice "cruising" mode, with most of the MegaPrime features implemented, and with the MegaPrime optics aligned as well as possible, time has come for this concentrated effort. A small task force has been created at CFHT, with Greg Barrick, Tom Benedict and Derrick Salmon as the main players, Derrick leading the group. Many individuals, in house and in the CFHT community, will participate in the process. This web page will share with all those interested in the MPIQI the progress of the investigation. This project has been given top priority at CFHT, following the clear recommendation of the Scientific Advisory Council. Christian Veillet - Executive Director -
June 2, 2004
|
Because of heavy WIRCam staff loads and
severe limitation on observing time due to exceptionally poor weather
this winter, we have not attempted any modifications to the wide-field
corrector or its alignment over the past six months. Nonetheless,
considerable effort has been made towards understanding the origin of
off-axis image quality degradation and the reasons for improved image
quality with lens L3 inverted. Despite these
efforts, only limited insight has been gained.
Lens L3 has
been positioned on short
standoffs in its inverted
orientation for all observing runs starting in December 2004.
A plot of image quality versus field radius in this
configuration is provided in Figure 1.
Imaging performance with lens L3 inverted provides a strong constraint for the development of optical models that account for image quality in both the L3-inverted and initial design configurations. To date we have been unable to develop a model that accounts consistently for both these configurations, although we have seen results which are tantalizingly close.
In early January we characterized primary
mirror wavefront errors using Shack-Hartmann data taken at the
telescope in
November, 2004 and included these, with correct rotational orientation,
in
MegaPrime optical models. The primary mirror
wavefront is shown in Figure 2 together with synthetic and real
through-focus images. Figure 3 is a plot of image
quality
versus field radius using both synthetic star images derived from the
optical model and real image data. If
these were the only data available we would likely have assumed that
image quality deterioration at large field radii had been accounted
for, and not taken the investigation further.
Figure 2 --
Primary mirror wavefront measured in November 2004, together with
through-focus images synthesized from the wavefront and
observed on the sky.
Many of the features seen in real defocused images are
reproduced in the
synthetic images, although differences remain - an indication that the
wavefront
map approximates the true primary mirror, but is not a perfect match.
Figure
3 - Observed (blue) and modeled
(magenta) image quality versus field radius. Both
data sets are for lens L3 in the design (non-inverted)
configuration.
The modeled
points include
the effects of primary mirror wavefront errors.
However,we
have data taken with lens L3 inverted. A plot of sythetic and
observed image quality in this configuration is provided in Figure
4.
Agreement is not good, and therefore neither is the optical
model.
Figure
4 - Observed and modeled image quality versus ield radius with
lens L3 inverted and with primary
mirror wavefront errors included. Plots such as this suggest that
MegaPrime suffers from an error
in at least one of the
fundamental optical parameters of the system.
If
we were to (rashly) assume that the observed image quality with lens L3
inverted
somehow returns the corrector to its design imaging performance, then
the strong tails seen at large field radius can be partially
accounted for by a de-space error between the primary mirror and the
corrector. Similarly, the splitting of the image quality
distribution can be
accounted for by
including a focal plane tilt and/or a decentering of the wide field
corrector, as demonstrated in Figures 5 and 6.
Figure
5 – Image quality versus field radius. Synthetic images
have been calculated for the
wide field corrector at its designed spacing from the primary mirror
(left) and
de-spaced 5.5 mm
toward the primary mirror (right).
Figure
6 – Image quality versus field radius showing splitting of the
image quality
into several
arms at large field radii corresponding to each of the four curves of
the CCD
array. Splitting in the
synthetic data is the result of a ± 75 micron focal plane tilt
included in the optical model. This
model was optimized
assuming an index gradient in lens L1, with lens L3 inverted, but is
typical
of
results obtained on other models with focal plane tilts.
We hope to test the
effects of changes to corrector alignment with lens L3 inverted at the
end of semester 2005A
or the start of 2005B – a time which should permit the MegaPrime
optical configuration to remain undisturbed for the largest possible
fraction of a semester.
To explain the improved
image quality with lens L3 inverted seems to require an error in at
least
one of the
fundamental parameter of the MegaPrime optical train. Early
in our modeling effort we examined the changes in optical parameters
needed to account for this unexpected improvement.
The results are summarized in Table 1 which lists the
effects of changes of lens surface radii, conic constants and
refraction indices for lenses L1, L2 and L3 and the conic constant of
the primary mirror, all with lens L3 inverted.
Each entry in the table includes, first, the re-optimized optical model merit function (smaller values are better), followed by the design value of the parameter being optimized, the re-optimized value and the percentage change after optimization. Entries with a small merit function combined with a small percentage change suggest a parameter worthy of further investigation.
|
R1 |
R2 |
CC1 |
CC2 |
Index |
|
|
|
|
|
|
Lens 1 |
0.1897 |
0.2485 |
0.2313 |
0.2291 |
0.2439 |
|
-877.03 |
-1065.02 |
0.0000 |
0.0000 |
1.516728 |
|
-875.04 |
-1068.77 |
+0.0035 |
-0.0072 |
1.523234 |
|
+0.22% |
-0.35% |
------------- |
---------- |
+0.42%
|
Lens 2 |
0.1839 |
0.2000 |
0.1129 |
0.1101 |
0.2282 |
|
-1999.05 |
-651.357 |
0.0000 |
0.0000 |
1.516775 |
|
-1943.22 |
-654.672 |
0.2616 |
-0.0096 |
1.509802 |
|
+2.79% |
-0.51% |
------------ |
----------- |
-0.46%
|
Lens 3 |
0.4634 |
0.3531 |
0.1259 |
0.4836 |
0.3538 |
|
-7967.88 |
Flat |
0.0000 |
0.0000 |
1.5167775 |
|
-5813.929 |
+6544.41 |
19.94635 |
No affect
|
1.879223 |
|
+27.0% |
------------- |
------------ |
------------ |
+23.9% |
P.M. |
|
|
0.4823 |
|
|
Figure 7 - MegaPrime image quality starting December, 2004.
A more detailed discussion of the current understanding of MegaPrime image quality issues will be provided here shortly.
Ideal and measured long exposure imaging observed performance is shown graphically in Figure 9 where image fwhm for one particular exposure and ideal performance are plotted against field radius. The breadth of the fwhm distribution results from the fact that image quality varies significantly as a function of azimuth in the field. A mosaic of individual star images at selected points in the MegaCam field as the camera passes through focus is provided in Figure 10.
In addition to the long-exposure image quality
problems noted above, Chris Pritchet has recently demonstrated that
plots of fwhm versus field radius for successive, otherwise identical,
30 second exposures show considerable variations in image size over the
field. Examples of the effect can be seen in Figure 11.
Although opto-mechanical instabilities were
initially suspected to be the root cause of these unexpected changes,
our current belief is that these variations are a result of an
interplay between dynamic atmospheric turbulence, static,
non-axially-symetric MegaPrime optical aberrations and the software
used to calculate image fwhm data.
A report on these effects and on efforts to
explain them is available here.
Image Quality Evaluation and Test Plan:
There may remain some simple steps we can take to try to solve the image quality problem. Barring a quick solution, which we do not expect at this point, we plan first to identify the optical element(s) which contribute significantly to image quality degradation, to characterize the specific causes of the problem, and then implement corrective action.
Several classical optical evaluation tools are at our disposal. These include inspection of focused and defocused images, Hartmann and knife-edge tests, cross-polorized stress birefringent tests, interferometry and evaluation of the mechanical systems involved in the mounting of lenses in corrector. The order and manner in which these tools will be applied will depend largely on what we find as investigations proceed. We also have an array of manufacturing and post-acceptance test data which will be reviewed in detail.
We have asked Chris Pritchet, at University of
Victoria, John Tonry, at the Institute for Astronomy, Guy Ratier at
ESA, and Murray Fletcher at HIA to act as external participants and
advisors in the IQ improvement program. Both Murray
and Guy have many years of experience in the testing and evaluation of
large optics and have kindly agreed to participate in the upcoming
efforts. Guy and Murray were critically involved in
the evaluation of our primary mirror while it was being fabricated many
years ago and are thus already familiar with CFHT optics on an intimate
basis. Chris and John are long-term users of high quality imaging
data and will provide critical scientific evaluation and advise.
John Pazdor at HIA has also provided valuable input.
September 20, 2005 |
Residual NW-SE detector tilt
removed by tipping MegaCam 0.00037 radians. IQ now
acceptable. Effects of primary mirror-to-corrector spacing
changes on iq
resulting from changes in ambient temperatures remain to be seen. |
July - August, 2005 |
Height of primary mirror in its
cell monitored to see if its position correlates with image-to-image
changes of iq distribution. No correlation detected. |
July 26, 2005 |
Primary mirror to corrector
spacing increased by 3.4 mm, with L3 remaining in its inverted
orientation. |
December 2004 - June 2005 |
Extensive optical modelling to
determine root cause of off-axis image degradation. Refractive
index gradient in L1 a potential explanation. Primary mirror
figure errors ruled out. |
November 4 - 23, 2004 |
|
|
L3 returned to its inverted orientation with an attendant 28 mm displacement towards L2, and limits on focus stage changed to accommodate change in focus. |
|
L3 returned to correct design
orientation. |
November 1 - 3, 2004 |
|
|
L3 and its cell accidently
inverted in the corrector provided much improved image quality but with
a 4.5 mm lower camera focus which precluded some filters from coming
into focus. L3 confirmed to be mounted correctly in its cell. |
October 29 - 31, 2004 |
|
|
Wavefront for the naked primary
mirror evaluated on the sky using Shack Hartmann sensors from Imagine
Optic (HASO) and Adaptive Optics Associates. (Wavescope) Visit by Pierre Kern from Observatoire de Grenoble to loan and operate HASO wavefront sensor. |
October 22 - 28, 2004 |
|
|
Internal geometry of wide field
corrector - lens axial thicknesses and seperations - measured.
Spherometer measurements of 8 lens surface radii made using
loaned spherometer from Mosaic Optics. |
September, 2004 |
|
|
Wide field corrector optics
re-assembled. |
August 24 - 27, 2004 |
|
|
L2, L3, L4 and the ISU plate
were tested individually for stress-induced birefringence at 0° C
and 20°C. L1 was tested previously at both temperatures. L2 and L3 show strong signs of highly localized stress around their circumferances associated with the band of glue beads securing each lens to its cell. These effects appear at both 0° C and 20° C. At 20° C L3 alone shows signs of glass-to-glue bead seperation over a portion of the surface of roughly a quarter of the glue beads - noteably on the West and SouthWest sides. These effects are not evidence once the lens and cell are returned to 0° C. Lenses L1 and L2 show strong signs of radially increasing bulk birefringence which is not likely to contribute to imaging problems. |
July 30 - August 6, 2004 |
|
|
At 0°C L1 shows radially
symmetric birefringence of the bulk glass which likely results from the
slumping process used during glass blank fabrication. No
localized mount defects are evident in these tests. L1
birefringence does not appear to contibute to image quality problems. L2 and L3 tested
together show what appears to be significant stress-induced
berefringence localized near their edges most likely as a result of
lens-to-cell mounting problems. L4 and the ISU plate do not show
stress-induced berefringent affects. |
July 26 - July 29, 2004 | |
|
Tests for stress-induced birefringence show significant localized stress the for assembled corrector at 0C. Further test will follow with L1 separated from remainder of corrector in order to identify the optical element(s) responsible. Teleconference with SAGEM to discuss findings and possible corrective actions have started. |
July 5 - July 11, 2004 |
|
|
Hartmann spot centroiding tool
developed by Chris Pritchet |
June 28 - July 4, 2004 |
|
|
Mosaic sub-array extraction
software tool provided |
|
"O" ring removed from L1 cell |
June
21 - 27, 2004 |
|
|
Discussions of IQ problems held
with SAGEM's Roland Geyl and Eric Ruch. Preliminary discussions with Imagine Optic's Xavier Levecq for Shack Hartmann testing of CFHT primary mirror and MegaPrime wide field corrector tentatively scheduled for September, 2004 |
|
SPIE
meeting in Glasgow, Scotland |
|
Review of
IQ data pre/post May 7, 2004 suggests efforts on May 7 did not effect
image quality. (details
here) |
June
14 - 24, 2004 |
|
|
Departure
for SPIE |
|
Planning
and system data review meetings. 'O' ring on lens L1 scheduled to
be removed on Tuesday, June 29. Next MegaPrime run (July 7-26)
will not have 'O' ring in place |
June 7 - 13,
2004 |
|
|
Review of
wide field corrector mechanical drawings suggests that a large 'O' ring
air seal on the cell of lens L1 may contribute to cell mechanical
distortion and possibly distortion of L1 itself. |
|
Chris
Pritchet has provided CFHT with his software to plot image fwhm vs.
field radius |
|
Preliminary
Hartmann test data and defocused images taken at the telescope.
Observations to support explanation of image quality variations with
time (Pricket effect) also completed. |
May
31 - June 6, 2004 |
|
|
Creation of
this web page - 1st meeting of the MPIQI working group |
May
24 - 30, 2004 |
|
|
Defocused
images taken for IQ evaluation. Initial inspection shows
significant pupil distortion near field edges. (details
here) |
May
17 - 23, 2004 |
|
|
CFHT SAC
meeting in Victoria, B.C., Canada |
May 10 - 16,
2004 |
|
|
CFHT User's
meeting in Campbell River, B.C., Canada |
Frame No. Comments
753619x.fits focus frame - excellent I.Q., z filter, 30 sec.
1st f = 0.579, 100 μm steps
753620o.fits
I.Q. 0.49
753679x.fits
focus frame - excellent I.Q., і filter, 30 sec.
1st f = 1.249 100 μm steps
753682o.fits I.Q. 0.50
753683o.fits I.Q. 0.52
753685o.fits I.Q. 0.53
753446o.fits focus series
753447o.fits 100 μm steps
753448o.fits
753449o.fits
753450o.fits
753451o.fits
753452o.fits
753453o.fits
753454o.fits
749707x.fits focus frame, no filter, 200 μm steps
749708x.fits focus frame, no filter, 100 μm steps
749709o.fits no filter 80 sec.
749710o.fits no filter 180 sec.
749711o.fits no filter 180 sec.
49712x.fits focus frame R
749715o.fits focus = 6.8
749716o.fits focus = 6.8
749717o.fits focus = 4.8
749718o.fits focus = 4.8
749719o.fits focus = 1.8
749720o.fits focus = 1.8
749721o.fits focus = -1.2
749722o.fits focus = -1.2
749723o.fits focus = -3.2
749724o.fits focus = -3.2
742565o.fits April 24/25/04
744310o.fits May 8/9/04
744662o.fits May 11/12/04
742274o.fits series shows
742275o.fits Pritchet I.Q.
742276o.fits variations
742277o.fits
Frame No.
June_07_04_31 Hartmann amp 00 +11.648
June_07_04_32 +6.00
June_07_04_33 -6.00
June_07_04_34 amp 71 - 6.00
June_07_04_35 +6.00
June_07_04_36 +11.655
June_07_04_37 amp 17 +11.655
June_07_04_38 +6.000
June_07_04_39 -6.000
June_07_04_42 amp 54 - 6.000
June_07_04_43 +6.000
June_07_04_44 +11.634
June_07_04_45 amp 03 +11.634
June_07_04_46 + 6.000
June_07_04_47 - 6.000
June_07_04_48 amp 68 -6.000
June_07_04_49 +6.000
June_07_04_50 +11.634
June_07_04_51 amp 14 +11.633
June_07_04_52 + 6.000
June_07_04_53 - 6.000
June_07_04_54 amp 57 - 6.000
June_07_04_55 + 6.000
June_07_04_56 +11.620
June_07_04_59 amp 26 +11.609
June_07_04_60 + 6.000
June_07_04_61
- 6.000
Acad
Files - Correcteur
Assemble. dwg
L1cell.dwg