ACAM suffers from scattered light when observing near the moon.
The aim of this project is to identify the path of the moonlight
through the telescope and ACAM (by experiment, and by
zemax modelling), and baffle accordingly.
Team:
Don Abrams,
Tibor Agocs,
Chris Benn (instrument specialist),
Kevin Dee,
Lilian Dominguez (deputy instrument specialist),
Annemieke Janssen,
Neil O'Mahoney, Servando Rodriguez.
Resources required:
It's expected that the staff effort required for this project during 2012
will be a few weeks of effort from each of the astronomy group, and
the telescopes and instrumentation group.
External effort will be required from Kevin.
Current status:
In November 2011, the scattered moonlight was significantly
reduced (by up to a factor of 3, depending on radius from the moon)
by inserting a 9-annulus 'Chinese lantern' baffle
inside the below-Nasmyth turret.
It has not yet been possible to reproduce the symptoms of the scattered light
in a Zemax model.
The results of all tests are recorded on, or linked from, this page.
Documentation for observers and support astronomers will be
provided on the
public ACAM web pages and on the
intranet ACAM pages.
Sections below:
Scattered light is currently a serious problem when imaging through
broad-band or narrow-band filters in wheels 1/2, if the telescope
is pointing within ~ 28 deg of the moon.
Imaging through focal-plane filters appears not to be affected.
We advise
observers to take a test image before observing, to check whether
scattered light is present.
The scattered moonlight typically manifests itself as a bright
fan-shaped
feature covering most of one half of the imaging area, and with
intensity a factor 2 - 3 times higher than the intensity in the rest
of the imaging area:
Fig. 1 - Typical appearance of scattered moonlight on an ACAM imaging
exposure.
The intensity of the illumination in the unaffected area
is consistent with that expected from moonlit sky.
The scattered light is in the bottom/top half of the CCD when the
elevation of the telescope is higher/lower than that of the moon, and
is on the right/left side of the CCD when the azimuth of the telescope
is less than / greater than that of the moon.
The intensity of the scattered light varies in an interesting
way with angular
distance from the moon (Fig. 2).
Fig. 2 -
Intensity of scattered-light feature (e.g. as seen in bottom part
of Fig. 1) over and above the smooth moonlit sky background
(e.g. as seen in the top part of Fig. 1).
Even at an angular distance of 40 deg,
where the moonlit sky background is about
500 counts, the scattered light is a significant problem.
See Appendix A1 for details.
At a radius of 90 deg, there is no scattered light (Appendix A7).
ACAM flat-field exposures (twilight sky, or dome)
typically show a radial gradient, with
the intensity at the centre of the flat field being 5 - 10% higher
than at the edge. This is probably due to sky (or dome) light being scattered
into the light path by the same mechanism that scatters moonlight
into ACAM. (Imagine summing the illumination distribution of Fig. 1
over all possible position angles - the result would be a
radial gradient.)
Based on the results of the tests carried out so far
(see Appendices A0, A1, A2, A3...), the moonlight is probably scattering
not off the primary mirror, but off the inside surface of the Nasmyth
fog baffle (80% Lambertian, 20% direct reflection, say).
The variation of intensity with angular distance from the moon (Fig. 2,
above), shows two distinctive minima (almost no scattered light)
which match well the expected
shadowing of the fog baffle by the secondary mirror assembly
(see Appendix B).
This confirms that moonlight impinging on the inside
surface of the baffle is indeed the source of the problem.
At large angular distances from the moon (probably when the WHT is completely
shielded by moonlight), no scattered moonlight is detected on the ACAM
images, and the level of the sky background is as expected (Appendix A7).
The scattered-light effect
has been reproduced during the day in a darkened dome,
by bringing the telescope to low elevation, and shining a torch from
the Nasmyth walkway (angle <~ 40 deg from telescope axis)
onto the far inside surface of the fog baffle.
The tests of 24 June (Appendix A5) and 8 July 2011 (Appendix A6)
showed that the scattered light is not affected by the absence
of the WHT secondary mirror, or by covering it up.
The tests of 8 July 2011 (Appendix A6) confirmed that when the
main aperture of the fog baffle is blocked, the scattered light disappears.
They also showed that significant scattered light is
not passing through the side apertures in the Nasmyth turret.
Attempts to reproduce the scattered light
within Zemax have been partially successful, see the
draft report by Annemieke and Tibor.
The modelling predicts significant scattered light from the fog baffle,
but the form of the scattered light (most of the energy in an arc across
the field of view) is wrong.
The intensity is hard to constrain because small errors in the reflectivity
of successive scattering surfaces could lead to differences of orders
of magnitude in the intensity of scattered light at the CCD.
The moonlight cannot impinge directly on the
ACAM flat mirror in the A&G - this path is blocked by the fog baffle
and the baffle of
the secondary mirror (Kevin's email of 21/6/11).
When the f/ll focal plane is stopped down with circular apertures
of decreasing radius R, the illuminated area on the CCD shrinks accordingly
(Appendix A0), with no scattered light detected at radius > R.
This is consistent with the scattered light already being in the
science path by the time it arrives at the f/11 focal plane.
(And *not* with all the scattered light entering through a restricted
area at the centre of the focal plane - thanks to Annemieke for
correcting this mis-interpretation.)
The scattered light passes through the aperture of the filter wheel,
not around it (Appendix A0). However, when the filter is moved to
a 'clear' position, the gradient of the scattered light reverses
(for each of the f/11 focal-plane masks tested),
suggesting that much of the scattered light is at the edge of the pupil,
corresponding to a large angle of incidence in the f/aa focal plane.
Moonlight can be blocked at several points:
before it reaches the telescope;
by the dome; in the fog baffle; in the A&G box; or in ACAM.
Using the dome / shutter to block the moonlight
This is currently an effective strategy if the observer finds
that moonlight would otherwise compromise the science observations.
Both upper and lower shutters can be used, although the range of the latter
is limited.
It's not a long term solution.
The feasibility of a moveable curtain inside the dome aperture
is worth exploring, but maintenance of such a curtain may be problematic
and expensive.
Preventing moonlight scattering off the inside surface of the fog baffle
This isn't straightforward, because (1) the moonlight impinges on the
surface at a shallow angle, (2) the 'thickness' of any baffling rings
(outside minus inside radii) can only be a few cm at the top of the baffle
before starting to vignette the science beam,
(3) moonlight could be reflected upwards by any extra baffling, towards
the secondary mirror.
Annemieke suggested a shorter fog baffle, to allow 'thicker' baffling rings.
The fog baffle was probably made as long as it can be without
vignetting the science beam. How short
can it be while still preventing moonlight reaching the central hole
in the primary mirror?
A crude attempt at providing a rougher inner surface to the fog baffle
(dark-coloured tubing glued around the inside of a paper cylinder
inserted into the fog baffle) only made the scattering worse.
Kevin 7/11 suggested
inserting an annular baffle
at the base of the fog baffle, in order to block direct lines
of sight from the ACAM fold flat to the inside surface of the fog baffle.
The annular baffle should have outer diameter 700 mm and inner diameter
530 mm (starts vignetting at ~ 6 arcmin fov) or 580 mm
(starts vignetting at ~ 8 arcmin fov).
Baffle the A&G
As of November 2009, extra baffling in place included
a border baffle round the 45-deg flat, black card on the disk in the
A&G opposite the ACAM port, a cylindrical baffle at the ACAM port
(reaching 20 mm towards ACAM and 30 mm towards the ACAM flat),
and black cloth round ACAM.
These did not help. The black cloth has now been removed.
In October 2010, Tibor installed a circular-aperture baffle
between the A&G-box flange and the slit slide.
This had no effect on the scattered light.
Baffle ACAM
If the moonlight is entering at a sharp angle through the central 20 mm
of the focal plane (see Appendix A0),
and passing through a 50-mm or 60-mm aperture/filter in the
wheel, it's plausible that it's bouncing off the inside wall of
that section of ACAM holding lens 1 (this section has a diameter
only slightly greater than that of L1, which isn't ideal
for avoiding scattered light).
If so, it's likely (Fig. 6)
that this can be blocked with a top-hat shaped baffle
i.e. annulus plus flared cylinder, with the annulus (inner diameter ~ 65 mm)
close to the filter wheel, and the concentric ~ 65-mm cylinder
extending from that position back along the light path towards lens 1.
It remains puzzling that, if the scattered light is entering through
the central 2 arcmin of the focal plane, the problem isn't seen
when using the 106-mm narrow-band filters in the focal plane.
Needs checking.
Another puzzle is that the pattern of scattered light on the CCD
is similar for R and Hα filters (as one would expect) but
completely different (looks rotated by 180 deg) when no filter is
in the beam (tests of 17/1/11).
Below is a summary of the outstanding actions I'm aware of.
[ ] = who and when (if known).
High priority:
- Build and test baffle for Nasmyth turret.
Medium priority:
- Report on Zemax predictions - how closely does the predicted
scattered light match that observed, in form and intensity [Neil, Annemieke?]
- Test effect of masks of successively smaller diameter in filter
wheel (Annemieke's email of 22/6/11).
- Check that scattered light problem is not present when
observing with filters in the focal-plane slide (8-22/7 or
10-13/8/11).
- Measure more accurately the
intensity of the scattered moonlight on ISIS (slit-view)
and on LIRIS.
- Baffle Nasmyth-flat base-plate [Neil?, 11/7/11]
- Fit black card disk at bottom of fog baffle, where it attaches
to Nasmyth turret [Neil?]
- Explore feasibility of curtain at dome aperture for blocking
moonlight [Diego?]
- Cost a replacement feed prism for Cass autoguider with high-quality
anti-reflection coating, and black out reflective
brass structures (Neil's email of 13/6/11).
- Extend intensity vs angle measurements (Fig. 2) to angles > 40
deg (90 deg done 17/7/11).
- Review scattered-light solutions made for other instruments.
Low priority:
- Investigate whether scattered light affected aux-port images.
Actions completed
- Build and test multi-element cardboard baffle for ACAM [8/11]
- Build and test multi-plate cardboard baffles for the Nasmyth
turret and for the fog baffle.
- Measure scattered moonlight when AF2 is at the top end
instead of the secondary mirror
[Ovidiu, 8/7/11]
- Cover side opening on fog baffle, test effect [Neil?, 11/7/11]
- Test whether scattered light affected by closing mirror petals
(App 1)
- Reproduce scattered light during daytime, using torch.
- Check change in appearance of scattered light as A&G rotated (App. 1)
- Constrain light path of scattered light using masks in focal plane
(App 1)
- Test whether scattered light blocked by inserting mask in filter wheel
(App. 1)
- Provide zemax files describing all possible scattering
surfaces in (1) the telescope and (2) the A&G box.
Minutes of meetings
- 17 June 2011
These tests were mostly aimed at constraining the path of the
scattered light.
What does the moonlight initially scatter off?
An early finding was that scattered light appears
on the CCD only if moonlight is falling on,
or close (<~ 1 m) to, the primary mirror.
This was tested by Ovidiu, Lilian and Marie 17/1/11,
by stepping the WHT dome position
until scattered light could be seen on the CCD.
The form of the scattered light doesn't change when the mirror
petals are closed (tested by Ovidiu 20/2/11, runs 1562921/2).
The intensity of the scattered-light feature is also unchanged
(tested 12/4/11, runs 1574233/4), suggesting that
the scattered light does not enter via the primary mirror.
The intensity of the smooth sky background does
increase (typically by a factor of two) when the mirror petals are closed,
presumably because the brightly-illuminated petals are scattering
moonlight up to the secondary mirror.
When moonlight falls on the telescope top end (but not on the primary),
there is no scattered light on the CCD (Marie, 19/1/11).
What happens inside the A&G box?
The pattern of scattered light rotates with Cassegrain mount PA
(runs 1547465-468).
When scattered light is seen on ACAM, it probably also affects
the ISIS slit-view camera.
This was tested by Ovidiu 20/2/11,
runs 1562900 (ACAM, scattered light visible), 901 (ISIS slit-view,
strong gradient in illumination, probably scattered moonlight).
What's the path of the scattered light through the focal plane?
When black-card masks with central circular apertures are inserted
in the ACAM focal-plane slide,
the intensity of the scattered light on the CCD shows no
dependence
on aperture diameter, for diameters between 20 and 100 mm.
This implies that all the scattered
light enters through the central 20-mm diameter of the focal plane.
Black card masks with central holes
of diameter 100, 80, 60, 40 and 20 mm were inserted in the focal
plane on the night 17/1/11 (by Lilian, Ovidiu and Marie).
The appearance of the scattered light (no filters in the light path)
was ~ as shown in the figure above
(but at different position angle) and the intensity
was the same for masks with holes of diameter 100, 80 and 60 mm.
The telescope was then moved and the test repeated for
a 40-mm mask. The intensity was about twice as high as before
(presumably because the telescope was moved).
When the 20-mm mask was inserted, the intensity did not change.
Runs 1546749-757, 763-769
When a mask blocking the whole 100 mm is inserted, the scattered
light vanished.
This last result was confirmed by Marie on 19/1/11 - she inserted first
a mask with a 20-mm hole (scattered light seen), then a mask with no hole
(no scattered light seen), then the 20-mm mask again
(scattered light seen). Runs r1547873-890.
What's the path of the scattered light near wheels 1 and 2?
The scattered light passes through the aperture of the filter wheel,
as opposed to scattering round the wheel
(tested by inserting a black mask in the wheel.)
The form of the scattered light depends on the filter in the beam.
In the above tests 17/1/11, when Sloan r or Hα filters were moved
into the light path (as opposed to no filter),
the distribution of the scattered light
on the CCD (as shown above) rotated ~ 180 deg.
On the night 12/4/11, we measured the variation with angular
distance from the moon of (1) the uniform sky background (e.g.
as seen at the top of Fig. 1), assumed to be moonlit sky,
and (2) the peak brightness of the scattered-light feature
(e.g. as seen at the bottom of Fig. 1).
The moon was 5 nights from full (i.e. 1 night from 'grey').
The WHT was pointed at the same azimuth as the moon.
At each of 38 different elevations relative to the moon,
a 1-sec image of the sky was taken through the Sloan r filter
(runs 1574251 - 288), using
4*4 binning and fast CCD readout.
When pointing the WHT well away from the moon,
the intensity of the sky background
is SKY = 500 counts per binned pixel.
This is broadly consistent with SIGNAL's prediction for
the surface brightness of the moonlit sky with this binning
and readout speed: 1300 counts in full moon, 300 counts on a grey night.
The brightness of the moonlit sky rises to 1.4 SKY at a radius of
15 deg from the moon, 2.2 SKY at a radius of 10 deg,
and 5.4 SKY at a radius of 4 deg.
The scattered-light intensity (minus the moonlit sky background)
is shown as a function of elevation
in Fig. 2, at the start of this document.
It is 0 at small angles, rising to ~ 3.5 SKY for angles 10 - 15 deg,
and drops to zero again between ~ 21 and 27 deg.
From 27 to 42 deg away from the moon, it is ~ 0.4 SKY.
Tibor et al investigated the source of the scattered light on
night of 12/4/11, using a camera looking back along the light path.
See the
draft report by Annemieke and Tibor.
Conclusions: the main conclusion is that the source of the scattered
moonlight is reflection off the inside surface of the Nasmyth-level
fog baffle.
As a result of Tibor's findings from the previous night, the inside of the
fog baffle was lined with a paper cylinder to which had been taped ~ 10
rings of dark-coloured half-tubing, in an attempt to suppress the scattering:
Fig. A3.1 - First attempt at baffling the fog baffle.
On-sky (that night), the scattered light was still present (and perhaps
worse), see Annemieke's email of 14/4/11.
Conclusion: baffling structure probably
too reflective.
Neil and Annemieke lined the ACAM barrel from the junction with the
filter box, back towards lens 1,
with 'black velvet' paper. No difference in scattered light (torchlight
on fog baffle) was observed.
Conclusion: the paper is probably not significantly more absorbent than
the inner (and slightly corrugated) surface of ACAM (Neil's email of
21 June 2011).
During the AF2 run, i.e. with no secondary mirror present,
Lilian, Ian and Marie
configured ACAM to be susceptible to scattered moonlight
in the usual way.
The intensity of the scattered light is similar to that seen when M2
is present.
(The intensity of the scattered light almost doubles when
the mirror petals are closed).
Conclusion: scattering of light via fog baffle and then M2 is not a
serious problem.
This is a brief summary of the
(word-format) report.
The telescope and ACAM were configured in the usual way to reproduce
the scattered-moonlight problem. Various apertures were then covered
up to see what effect this had:
Test Items covered up Scattered light
1 Secondary mirror Present
2 Turret side apertures Present
3 Secondary + turret side apertures Present
4 Secondary + turret side/top Not present
5 Turret top (circular aperture) Not present
Conclusion: the scattered light passes through the fog baffle,
and not via either the secondary mirror or the side apertures in the
Nasmyth turret.
On 17/7/2011, Ian took some short exposures (e.g. r1651626)
with ACAM/Sloan-r of the
sky ~ 90 deg from the moon (3 days after full).
There's no evidence of scattered light on these exposures, confirming
that at large enough angular distance (or when the WHT is completely
shielded from moonlight), the intensity of the scattered light
does drop to zero (by contrast, see Fig. 2,
which shows significant scattered light at an angular distance of 40 deg).
The intensity of the sky background is 60 counts per unbinned pixel
(slow readout mode), exactly as predicted by SIGNAL for bright of moon,
confirming that it does not include a significant scattered-light
component.
Conclusion:
The observed sky background observed at large angular distances
from the moon is conistent with the expected surface brightness
of the moonlit sky, i.e. includes no scattered-light component.
Kevin designed
two annular baffles
to go inside the fog baffle to obstruct
direct lines of sight from any part of the inside surface
of the fog baffle to the ACAM fold flat.
The ring baffles have outside diameter 635 mm.
The inside diameters are 530 mm (baffle 1) and 580 mm (baffle 2)
and Kevin predicts that these leave unvignetted Cass fields of view of
diameter 3.5 and 7.5 arcmin respectively. They will significantly vignette
the Cass autoguider.
Fig. A8.1 - Proposed location of ring baffles.
Neil installed a cardboard version of baffle 2 (580 mm) on 20 July 2011
(standing on the primary-mirror covers to insert it into the
fog baffle).
That night, Ian took 2 sec ACAM images of the sky through a Sloan
R filter (bin 4*4, fast readout), at elevations
10, 15, 23, 30, 40, 50, 60 deg higher than the moon
(runs 1660351-362). The moon was 6 nights from full,
i.e. ~ same brightness as on 12/4/11 (Appendix A1).
On each image, the brightness of (1) the uniform moonlit sky background
(minus bias) and (2) the peak brightness of the scattered-light feature
(minus uniform background) were measured (as they were on 12/4/11).
The brightness of the moonlit sky on all exposures >= 40 deg
from the moon is SKY = 250 counts per binned pixel
per sec of exposure.
The value of SKY is a proxy for the brightness of the moon (which depends
strongly on lunar phase).
The sky brightness rises to 1.6 SKY at 30 deg from the moon,
2.0 SKY at 23 deg,
3.0 SKY at 15 deg and 4.4 SKY at 10 deg, broadly similar to the dependence seen
on 12/4/11 (the exact variation depends on how much dust and cloud are present).
The scattered light intensity (minus sky background) is 18 SKY at
10 deg from the moon, 3.6 SKY at 15 deg,
and 0 at angles >= 23 deg. The values at 15 and 23 deg
are similar to those seen 12/4/11, that at 10 deg is rather higher.
On exposures made at the same elevation as the moon, but 15 and -15
deg away in azimuth, the scattered-light intensities are 2.8 and 4.8 SKY,
i.e. there's some evidence for a departure from circular symmetry.
No scattered light is present on an exposure taken at the same
elevation as the moon, but 90 deg away in azimuth, i.e. with
the dome completely shadowing the telescope from moonlight.
The moonlit sky brightness there is not significantly different
from SKY.
Conclusions:
- The intensity of the scattered light at 15 deg from the
moon is similar to that seen on 12/4/11, so the baffle has not suppressed this.
-
The baffle may have suppressed the lower-level scattered light
at elevations > 30 deg, present 12/4/11, but not 20/7/11.
*** Report pending *** (see Annemieke's MSc thesis).
Observing conditions: clear
Observing conditions: clear, some dust
The variation of sky brightness with radius from the moon (3 days from full)
was measured
for each of several combinations of cardboard baffles.
The baffles are below called:
- 'ACAM' - the multi-component baffle left in from
the previous night
- 'turret' - a 3-plate baffle positioned inside the Nasmyth turret,
with each plate suspended from the one above it
- 'fog-baffle' - multi-plate baffle inside a card cylinder, which can
be inserted into the fog baffle
Fig. A10.1 - Multi-plate cardboard baffle for insertion in the fog baffle.
Setup: Sloan r filter, readout speed = fast (bias level = 850), 4*4 binning, 0.5-sec exposures.
Summary results are given below (measurements from individual images
to be added later).
ACAM, turrent and fog-baffle baffles in position
Observations were made at angles between 4 and 40 deg from the moon.
The sky brightness at large angles from the moon is SKY = 1000 counts / sec
/ binned pixel. This provides a reference brightness for the measurements
made tonight. The sky brightness at 10 deg from the moon is 24 SKY.
No scattered light is evident on any of the images. This is the first
time that any combination of baffles has eliminated the scattered light
entirely.
ACAM and turrent baffles in position
The fog-baffle baffle has been removed.
Observations were then made at angles between 4 and 40 deg from the moon.
No scattered light is seen at any angle.
The sky brightness at each angle is slightly less
than that measured above, which is hard to understand, but
might be due to the changing elevation
of the moon.
ACAM baffle in position
The turret baffle has been removed.
Observations were then made at angles between 8 and 20 deg from the moon.
Strong scattered light is present, with the usual appearance,
and intensity ~ 8 SKY at 10 deg from the moon.
No baffles
The ACAM baffle has been removed.
Observations were made at angles between 8 and 20 deg from the moon.
The sky background has increased by a factor of two (to 18 SKY at 10 deg).
The intensity of the scattered light has not changed at any angle,
suggesting that the ACAM baffle was not effective in blocking
the scattered light. However, it was probably vignetting the beam
in the near-pupil space before the filter wheels, hence the reduction
in the intensity of the background.
Fog-baffle baffle only
The fog-baffle baffle has been re-installed.
Observations were made at angles between 4 and 50 deg from the
moon. The intensity of the background is approximately
unchanged (22 SKY at 10 deg), but the intensity of the scattered light
is reduced by a factor of two relative to having no baffles.
Conclusions:
- The turret baffle seems to have had the most dramatic effect,
but the fog-baffle baffle seems to significantly reduce the scattered
light.
The ACAM baffle had no effect on the scattered light.
The fog-baffle baffle was removed by the ops team during the day.
Observing conditions: clear
Ovidiu measured the intensity of scattered light vs radius from the moon
(4 days from full)
on (1) ACAM, with the mirror petals closed (runs 1671257 - 261),
(2) ACAM, mirror petals open (runs 1671276 - 294), and
(3) the AG3 acquisition camera at the GRACE Nasmyth focus,
with the mirror petals open (runs 1671295 - 302).
Results for ACAM
Setup: primary-mirror petals open,
no cardboard baffles in ACAM (apart from focal-plane annulus)
or telescope, Sloan r filter, readout speed = fast (gain 1.9 electrons/count),
4*4 binning,
0.5-sec exposures.
Results (measurements from individual images will be added later):
The sky background is SKY ~ 600 counts / sec / binned pixel at
large radii (< 65 deg) from moon, consistent with the prediction
in Appendix A1 above.
The sky background rises to 8 SKY at 15 deg from the moon,
14 SKY at 10 deg and 23 SKY at 4 deg.
The scattered-light intensity is 0 at 2 - 6 deg from the moon,
rises to 14 SKY at 10 deg, drops to 1 SKY at 20 deg, 0 at 25 - 30 deg,
and is <~ 0.3 SKY at angles > 35 deg.
This is similar to the variation seen in Fig. 2.
Ovidiu also made similar measurements with the mirror petals
closed. The sky background then varied much less with angle from the
moon, as one might expect, and for angle < 30 deg was similar to
that at 10 deg from the moon with the petals open (at 50 deg, it fell to
about one quarter of this value).
The intensity of the scattered light,
however, was almost identical to that
measured with the mirror petals open, confirming that the scattered light
does not enter via the primary mirror.
Results for AG3, at the Nasmyth focus
The measurements were repeated with the AG3 camera (field of view 3 arcmin,
scale 0.41 arcsec/pixel)
in GRACE to test whether
the scattered-light problem affects observations there.
Setup: Nasmyth flat in, no filter,
AG3 readout speed = fast (gain 0.9 electrons per count),
binned 1*1.
Observations were made at angular distances 6 - 40 deg from the moon.
Results (measurements from individual images will be added later):
(1) The moonlit sky background
(minus bias level) is SKY = 100 counts / sec / pixel at large radius
(40 deg) from the moon.
The surface brightness rises to 8 SKY at 15 deg from the moon,
and 15 SKY at 10 deg, similar to the values measured by ACAM,
suggesting that the smooth background seen in the ACAM images
is indeed dominated by moonlit sky, rather than a smooth scattered-light
component.
(2) The variation of intensity with angle from the moon shows no
peak near 10 deg, suggesting that the scattered light seen in ACAM is not
present at the Nasmyth focus.
There is a significant gradient in the sky
background on all exposures (<~ 30% from one side to the other)
but no structure similar to that seen in the central 3 arcmin of
the ACAM images.
Conclusions:
- With the cardboard baffles removed, the variation of intensity
of the scattered light with
angle from the moon is again similar to that seen in Fig. 2 (sanity check),
with no scattered light at 2 - 6 deg, or at 25 - 30 deg.
- There is significant scattered light out to at least 60 deg from the moon.
- The scattered light does not enter via the primary mirror (confirmation).
- At the Nasmyth focus (AG3 in GRACE) there's no evidence of scattered
light on the detector.
- The AG3 observations suggest that the smooth background illumination
in the ACAM images is dominated by moonlit sky rather than including a
significant scattered-light component.
Observing conditions: clear, moon (2 days before full) up all night.
Kevin constructed a 9-annulus 'Chinese-lantern' baffle:
and this was installed in the WHT below-Nasmyth
turret by Servando and David, ~ 22:20 - 23:30.
Results
(1) The baffle does not vignette the ACAM field of view (radius 4 arcmin)
at all (runs r1713207-8).
(2) We observed 7 guide stars (at radii 8 - 11 arcmin = autoguider r
= 0 to 40000) before installing
the baffle. Four were observed at one sky PA, and 3 at a sky PA 180 deg away,
to constrain possible departures from radial symmetry in any vignetting,
(r1713181-187).
For each guide star, we recorded the apparent mag, autoguider seeing,
autoguider exposure time and autoguider S:N, then took a short
exposure (e.g. 1 sec) with the autoguider CCD (AG6), and recorded the
run number, exposure time, peak counts, FWHM and total counts (imexam a).
All the guide stars were easily acquired after installing the baffle,
and there
was evidently no major vignetting,
although cloud precluded a more
accurate assessment.
On a subsequent clear night (2/12/11), we re-observed the same guide stars
(r1726234-240), and
there is no evidence of vignetting (>~ 10%) at radii 8 - 11 arcmin in
the patrol field.
(3) The scattered-light intensity vs radius from the moon
was measured before (r1713188-206) and after (r1713209-240) baffling.
Before baffling, the curve looked like that shown in Fig. 2
on this page,
although there seemed to be ~ zero scattered light at radius
> 25 deg.
After baffling, the curve changed as follows:
Scattered light for radius 6 - 20 deg is flat at about
one third the intensity of the peak seen in Fig 2.
At radius > 20 deg, there is no scattered light at all.
So the low-level scattered light seen before at radius
> 25 deg wasn't present either before or after baffling,
which is odd, but there's a dramatic improvement
for radius 6 - 20 deg.
Marie repeated the measurements on 13/11/11 (moon 4 days after full),
and her data (r1713716-725)
look similar to those of 8/11/11, except that the scattered
light is present at radii ~ 9 - 22 deg rather than 6 - 20 deg
(why?).
Conclusions
The baffle has significantly reduced the intensity of the scattered
light, but the scattered light remains a serious problem.
The baffle has been left in the telescope.
Observing conditions:
Observing conditions:
Observing conditions:
Conclusions:
The observed radial gradient in the flat fields does not arise from
light scattered within ACAM. Rather, it is present in the focal plane,
Fig. B1 - Layout of main WHT components in profile view (provided by Kevin).
The baffle of the secondary mirror is not included.
The fog baffle has diameter 710 mm (sky end). The next component along
the light path has diameter 635 mm.
There's a 1 - 2 cm (?) gap between the primary mirror and the bottom
of the turret / fog baffle.
The distances from the Cassegrain f/1l focus to the bottom and top of
the fog baffle are 5.0 and 6.5 m respectively.
The distance from the f/11 focus to the bottom of the A&G box is 150 mm.
The nominally unvignetted field at the f/11 focus (1994 Observers' Guide)
has diameter 19 arcmin (253 mm).
The fog baffle probably partially vignettes fields with diameter >
11? arcmin (KMD 3/6/11).
The fog baffle has been in place for many years, but used to be removed
for certain Cass instruments e.g. FAST? to reduce thermal emission.
Sketching additional rays representing moonlight entering the fog-baffle
(rim in yellow) after grazing (1) the edge of the secondary mirror,
or (2) the inside of the top-end ring, or (3) the outside of the top-end
ring, suggests that the scattered light will be blocked from
entering the fog baffle if the angular distance from the moon
is <~ 4 deg, or between ~ 19 and 29 deg.
These numbers match rather will the observed dependence on the angular
distance of the moon (Fig. 2).
Fig. C1 - ACAM mounted at the folded Cassegrain focus.
Fig. C2 - ACAM transparent view
Fig. C3 - Light path through ACAM
With filter wheels out,
the aperture allowing access to the inside of ACAM is about 90 mm wide.
Each filter wheel is held in place by a long steel bolt through
its centre.
It's important that at least one of these bolts,
together with the wheel hub, is always in place,
otherwise there's a small risk that the structure could deform
(advice from Kevin).
