Previous Article
Next Article
Table of Contents

F/8 Secondary Back in Service at 4-m (1Mar94)
(from CTIO, NOAO Newsletter No. 37, 1 March 1994)

The f/8 secondary mirror went back into service on the 4-m telescope on
January 1st, after an absence of eight months.  It's a good news, bad
news story.  The figure of the secondary appears to be greatly
improved, but the quality of the images at the f/8 focus has not
improved as much as was hoped.

                    First, the Good News....

We deduced from tests on the telescope that the secondary mirror had
two problems: (1) its conic constant was not properly matched to the
primary mirror with the result that there was a large amount of
spherical aberration at the design focus; and (2) there were high order
aberrations coming from surface irregularities with diameters of
roughly 10% of the mirror's diameter.  We decided to correct the
mirror's figure through ion polishing, to be carried out by Kodak, in
Rochester, New York.

We started by obtaining a careful measurement of the mirror's surface
figure, using interferometric tests against a Hindle sphere.  This was
done by Contraves Goerz Corporation, who presently have the NOAO Hindle
sphere in their shop.  Their results confirmed what we had found on the
telescope and provided us with a detailed map of the high-frequency
errors after subtracting off a pure hyperboloid shape (Figure 1a).

                    [Figure 1a not included]

         Contour plot of the high frequency errors on the
         mirror surface before it was ion polished.  The
         units of the high and low values are 0.001 waves,
         for a wavelength of 632.8 nm.  Contour intervals
         are 0.5 wave.

We then calculated the change in the lower order shape needed to remove
the spherical aberration problem at the nominal focus.  This can be
done to adequate accuracy without precise knowledge of the conics of
either secondary or primary.  The correction was specified as a
differential change in the surface both because the ion polishing
process works differentially and because the correction does not then
depend on a detailed knowledge of primary figure, which is in some
doubt at the level of precision required.  We elected to change both
the conic constant and the radius of curvature of the mirror, in order
to minimize the total material removal.  We added this low-order error
curve to the high-order map from Contraves.  After a bit of smoothing
in the middle and extrapolating at the edges, this was the polishing
map that we provided to Kodak.

The mirror was then sent to Kodak, who immediately made an alarming
discovery - there were several series of micro-cracks on and near the
surface of the mirror.  These were scattered around at seven different
locations on the mirror's surface and were 1-10 cm in length.  We are
unsure whether these cracks had always been present but had gone
unnoticed at CTIO, or if they developed during the mirror's travels.
In any case, the fear was that because of the local heating during the
ion polishing process, these cracks might propagate because of stress.
So we had Kodak grind them out into broader shapes in order to stress
relieve them.  This was done under the watchful eye of Gary Poczulp of
the NOAO optical shop.  The ion polishing then proceeded, in just one
iteration taking less than a week, exactly as was originally promised
by Kodak.  We wish to note here that we were very favorably impressed
by Kodak's professional approach and high-quality workmanship.

The final step was to send the mirror back to Contraves for a repeat
measurement of its surface shape.  The "after" result is shown in
Figure 1b, again as a map of just the higher-frequency structure after
subtracting off the lower order shape of the mirror.  Figure 1c shows
encircled energy diagrams calculated from the "before" and "after"
high-frequency maps.  These represent the image degradation just from
the high frequency errors, and show that this went from being a
substantial problem to being completely negligible.

                    [Figure 1b not included]

         Contour plot after ion polishing. Same units as (a).
                                                                                                    [Figure 1c not included]

         Encircled energy diagrams calculated from the
         wavefronts shown in (a) and (b).

The new measurements also showed that the conic constant and radius of
curvature had changed by approximately the right amount.  These
improvements appear to be confirmed by the measurements we have made
since putting the mirror back on the telescope: the spherical
aberration is reduced to less than 10% of its previous value and is now
unimportant.  The higher frequency terms also seem to be considerably
reduced, at least to the extent that out-of-focus images appear to have
a much more uniform surface brightness than they used to, and that when
we use our program for fitting the spot patterns from our Hartmann
screen, the rms high-frequency residuals are significantly smaller
than they used to be.

Success! At least up to this point.

                  But There's Still a Ways to Go...

We are frequently obtaining 1 arcsec FWHM images with the f/8 (Figure
2).  Unfortunately, the remaining low-order aberrations in the system
prevent us from getting the really good images for which we are
aiming.  The images show stable structure on the subarcsecond scale,
frequently breaking up into double images separated by about 0.5 arcsec
(Figure 3).

                    [Figure 2 not included]

         Radial profile of a star image from a 30 sec exposure.
                                                                                                    [Figure 3 not included]

         Contour map of star image from a 0.1 sec exposure.

We think that we interpret this as a combination of astigmatism and
triangular aberration coming from the primary.  At least the magnitudes
of those terms found in our Hartmann screen analysis are unchanged from
the values measured one or two years ago.  The tests made at that time
indicated that the problem was with the primary mirror support system,
and we intend to solve it in late 1994 or early 1995 with an active
support system for the primary mirror.

However, we do not customarily see images like this at any of the other
foci of the 4-m, so we are checking very hard to see if there could be
an additional problem with the support of the f/8 secondary.  Some
major changes were made to the secondary mirror cell while the mirror
was being repolished, so something may have gone wrong.  All we can say
for sure at this time is that we're trying to understand what is going
on.  We are carrying out numerous tests and measurements while using
the mirror for routine observing (and generally receiving positive
reports about the image quality, we should add).

            ...And Finally, Some Deeply Felt Apologies

We believe that the secondary mirror is much better than it used to be,
and that this is worth the time that was spent without an f/8 focus.
However, there was a highly unfortunate delay in getting the mirror
back onto the telescope which prevented us from being able to support
several scheduled observing runs.  We did our best to either shift the
affected astronomers to one of the other 4-m foci (if they could make
profitable use of the time), or to reschedule their runs at a later
time.  We wish to humbly apologize to our colleagues who were affected
in this manner, and even more so to the small number of astronomers who
just plain lost their observing time.  The problem arose mainly because
of huge overruns in the time needed to measure the mirror's figure, and
to a lesser extent because of the cracks.  The schedule included what
we thought was a generous contingency factor for this type of problem,
but obviously we were wrong.

Jack Baldwin, Brooke Gregory

Previous Article
Next Article
Table of Contents