Previous Article
Next Article
Table of Contents

4-m Telescope Optics Upgrade Project (1Sep92)
(from CTIO, NOAO Newsletter No. 31, 1 September 1992)

Our program of seeing improvement projects at the 4-m telescope is now
drawing to a close. We are switching our efforts to a comprehensive
plan aimed at significantly improving the telescope's optical
performance. Over a period of approximately two years, we are planning
to use about 15% of our instrumentation manpower and 40% of our
projected instrumentation budget to carry out four projects: repolish
f/8 secondary, image analyzer, improve f/8 secondary alignment
mechanism, active control of primary mirror support. These projects
will concentrate on the f/8 focus, but they also should lead to better
images at the prime and f/30 foci.

The Problems
The motivation for this series of projects lies in the following table,
which shows the effects of different optical problems at the f/8 focus.
The results are based on many, many nights of testing using the
curvature sensing, Shack-Hartmann and Hartmann screen techniques. We
have received a great deal of expert outside help in these tests, and
for this we are extremely grateful to Allain Gilliote, Ray Wilson and
Lothar Noethe (ESO), Claude Roddier (U. of Hawaii), and Fred Forbes
and Nick Roddier (NOAO). The different techniques give good agreement,
so we believe that the results are quite reliable.

The individual contributions to the overall optical performance at the
f/8 focus (expressed in terms of the 80% encircled image diameter d80),
and the planned cures, are:


        Aberration        d80 (arcsec)       Planned Cure

        Spherical            0.51            Repolish Secondary
        Coma                 0.23            Improved Secondary Alignment
        Astigmatism        0.28-0.56         Active Primary Support
        3, 4 psi terms       0.25            Active Primary Support
        High-freq.           0.67            Repolish Secondary
         errors

	Quadratic Sum       0.9-1.0


The astigmatism varies by a factor of two as a function of telescope
position.  The coma is presently fairly small, but we have gone for
considerable lengths of time with it at least twice as large.  For a
2-D Gaussian image, FWHM = 0.67 d80, so the overall image sizes are in
reasonable agreement with the best measured FWHM values of about 0.7
arcsec. For comparison, the total low-frequency errors in the CFHT and
NTT are, respectively, d80 = 0.33 and 0.075 arcsec, while the
corresponding figures for high frequency errors are d80 = 0.30 and
0.125 arcsec. Our telescope's optical performance therefore is, and
always has been, at least two times poorer than today's standards.
This is a fundamental problem which needs to be solved.

The Cures
Since the various aberrations have effects of similar size which add up
more-or-less in quadrature, we need to make a considerable improvement
in each area before we can expect a significant overall gain. The final
column in the preceding table shows the planned cure for each problem
area.

The f/8 secondary will be repolished to have the correct conic constant
to provide best focus (no spherical aberration) at the present position
of the cassegrain instruments. This will also correct serious
high-spatial-frequency errors in its figure which apparently were
caused by "print-through," during polishing, of a number of
light-weighting holes which were bored into the back of the mirror
blank. We are in the process of making a final decision on where and
how to have this refiguring work done, so there is some uncertainty
about how long it will take. Our best estimate is 4-6 months. Since we
need to announce this sort of shutdown way in advance, we have
scheduled it to begin on 1 June 1993, two months before the end of the
semester for which proposals are due this coming 30 September. This
gives us six months to firm up our schedule and still be able to
announce a reopening date in time for the following proposal deadline.

In deciding how to approach the remaining problems, we were heavily
influenced by ESO's New Technology Telescope (NTT) and benefitted
greatly from the help and advice of Ray Wilson and Lothar Noethe of the
NTT team. It had become clear to us during the previously mentioned
series of optical tests that we could never hope to maintain our
telescope in good condition unless we had some sort of image analyzer
permanently mounted and available for routine checking of the
telescope's condition. We are now in the process of grafting a
Shack-Hartmann system onto our existing cassegrain-focus offset guider.
We are using the NTT analysis software, which is the only hard-to-do
part of the system, and which was very generously given to us by ESO.
The Shack-Hartmann lenslet array will be a copy of the ones developed
for the VLT, while the detector will be an adaptation of CTIO's CCD
acquisition-TV cameras.

Once we had decided to have the image analyzer, we naturally started to
inquire whether we could carry out a useful amount of active-optics
correction in spite of having an old-fashioned thick primary mirror.
The answer is that we clearly can.

The first step will be to improve the focus/collimation system of the
secondary mirror so that we can automatically move the mirror to remove
coma resulting from bad collimation. We have elected to simply replace
the existing, cumbersome mechanism for tilting the mirror around its
vertex with one which is accurately encoded and precisely movable under
computer control. This is the easiest thing to implement given the
existing mechanical configuration, and our optical tests show that the
collimation in fact is stable as a function of telescope position; the
collimation errors appear as singular events, perhaps associated with
rotating the telescope's flip top end. However, there will also be
provision for very occasional daytime adjustments of the mirror's
centering by means of manually operated jack screws, to deal with cases
where the required secondary tilt is out of range.

The surprising (to us) result is the ease with which we can apply the
NTT's new technology approach to the primary mirror support system.
This is because the existing axial support system provides most of what
we need. It consists of three hard points and 33 "air bags" (see figure
below). The latter are simple pneumatic pistons which provide axial
force on the mirror in direct proportion to the air pressure fed into
them. The air pressure is presently varied to account for the changing
cos(z) term as the telescope changes its zenith distance z, but the
pressure to all of the bags is varied in unison under control of a pair
of mechanical pressure regulators. The modification which cries out to
be made is to replace these old regulators with 33 individual,
computer-controlled regulators so that each air bag (= force actuator)
can be controlled separately.

		            [figure not included]

Our primary mirror is twice as thick as the NTT mirror, and hence
roughly eight times stiffer. This led Wilson and Noethe to suggest
bending it in its eigenmodes, rather than trying to force it into the
shapes corresponding to the standard optical aberrations as is done
with the NTT. The motivation is that the mirror can be bent the same
amount with less than 1/4 the force, producing a corresponding gain in
dynamic range.  The lowest frequency eigenmodes are in any case the
ones most easily induced by imperfect support.

Noethe and Xiangqun Cui (ESO) used a combination of thin plate theory
and finite element calculations to investigate the feasibility of this
approach. The maximum force needed to correct each aberration must be
added to the nominal force needed to hold up the mirror against
gravity, so for the existing push-only actuators the dynamic range is
from 0 to 2*(nominal force). The following table shows, for each
azimuthal symmetry, the typical coefficient of the existing aberration,
the maximum force needed to correct the aberration, and the fraction of
the system's dynamic range (at the zenith) that would be used up.


                     	              Maximum        Fraction of
    Aberration     Coefficient         Force           Dynamic
                      (nm)              (N)             Range

    r4                3000             4128              0.98
    r2cos2psi          800              156              0.04
    r3cos4psi         1300             1453              0.34
    r4cos3psi          300              973              0.23


Thus with the existing configuration and capabilities of the actuators
we easily can correct the present wavefront errors having 2psi, 3psi and
4psi azimuthal symmetry, up to a zenith distance of about 60 (where the
dynamic range is half the zenith value). The 0 (spherical aberration)
term, however, is beyond our range and will have to be cured by some
other means (repolishing the secondary).

The other result from the work of Noethe and Cui is that the forces
supplied by the individual actuators do not even have to be controlled
to especially high accuracy. Randomly distributed force errors of 1
percent lead to an astigmatism (cos 2psi) coefficient of only 35 nm, which
is less than the 50 nm measurement error in the results from the image
analyzer.  This means that we can control the air pressure to each
actuator with low-cost "smart" pressure regulators. Our default option
is a $200 unit built by Mamac Systems which maintains an output air
pressure to within the required 1 percent of whatever value is set by
an input DC voltage. We plan to connect 33 of these units ($6600 total)
to 33 D-A converters in a small computer. The software in that computer
would simply interpolate the value and orientation of each aberration
from a look-up table which gives pre-measured values as a function of
hour angle and declination, sum up the desired forces at each actuator,
and then load the D-A converters. We estimate that once the image
analyzer is working, it should take only about an extra two to three
months of programming effort to develop this simple, open-loop
active-optics system.

Our general aim is to close down the f/8 focus for roughly six months
starting 1 June 1993, and during that period to have the mirror
repolished, install the new alignment mechanism, and install the image
analyzer. Then during an extended aluminization run during the
following winter (June or July 1994), the active primary support system
would be installed.

Jack Baldwin, Brooke Gregory,
Gabriel P‚rez, Jay Elias

Previous Article
Next Article
Table of Contents