|
During the cooldown
in early May, 2007 and during the first night on the telescope (5/20/07),
a number of serious issues were uncovered that required an immediate warm-up
and unscheduled servicing. These included:
- Detector unable
to reach nominal operating temperature
- FW3 sticking
- Stellar images
elongated EW in F/5 camera images
Additionally, a couple
of minor desirements had been left over from the previous warm servicings:
- M7 hermetic connector
leaking
- Needle valve leaking
Mimir was warmed,
moved into the clean room, opened and inspected. Present were D. Clemens,
B. Taylor, and A. Pinnick.
Findings and
Actions:
- Regarding the
detector temp problem: During operation, the temperature sensor
on the 2nd stage got as cold as 14.8 K, while the sensor just 8"
away on the 2nd stage side of the split copper block holding the sapphire
disk barely made it below 30K. This pretty clearly located the problem
to the mechanical interface from the getter to the copper mount disk
for the triplet of 1/8" copper rods that go to the split copper
block. Once Mimir was opened, we found that the 1/4" screw that
holds the copper disk to the getter was loose and unable to make proper
thermal contact. A lock washer was added to the flat washer on the 1/4"
screw and the screw was reinserted and tightened "robustly."
- Regarding FW3:
We opened the filter bay and rotated FW3 by hand. No interferences were
felt, but the wheel was grossly out of balance, permitting back rotation
after motion. One dummy filter cell was removed from the portion of
the wheel deemed too heavy, which produced an excellent balance for
the wheel. No back rotation was evident after removing the dummy cell.
Nevertheless, the 94 mil thick spacer used to register the FW stack
against the back reference surface was hand lapped to 90 mils thickness,
providing an additional 4 mils clearance between FW3 and its neighboring
bulkhead. This will also cause the POL wheel to give up 4 mils clearance
against its bulkhead, but upon rotating POL no interferences were found.
The filter bay cover was also inspected for wear by FW3, which would
have indicated interference, but none was found. The bay cover was reinstalled
and all wheels were operated satisfactorily.
- The M7 Mil Spec
14-12N hermetic connector was desoldered and replaced with a spare.
- The needle valve
and its flange were removed from the warm bulkhead, and replaced with
a stainless steel blank flange.
- Regarding the
elongated stellar images:
PV images from
the same night showed the dome in focus and no elongation of either
the dome parts or the pupil mask elements. The elongated stellar images
viewed through the F/5 camera showed no change in elongation properties
with location in the image, so we believe that no lenses were broken.
April did report the nominal F/5 camera position, determined via balance
of edge vignetting, had jumped some 60 motor steps, coinciding with
the image elongations. This seemed to imply that perhaps one of the
F/5 lenses had slipped or tilted away from its nominal position.
To check this,
the camera block was taken out of Mimir and the F/5 lens cell stack
mostly removed. The most likely culprit was L5, the final ZnS lens
which is both thick and has strong power. Upon removal from the camera
block, this lens cell was dismantled and the lens inspected. A partial
ring of scoring was noted on the convex side (sky side) of the lens.
This most likely was caused by the lens meeting the ID of the lens
cell front face. This should not happen, as the 1 mil thick kapton
"fingers" between the lens and the cell front face should
prevent interference. The presence of the scoring ring on the lens
indicated that either the lens cell had a burr on the ID lip and/or
that the kapton fingers had thinned and allowed the lens to meet the
cell ID. Once the lens and cell are in contact, stiction could prevent
the lens from cooling into its proper location, introducing both a
tilt and decenter. Though there is no guarantee, the scoring on this
lens is consistent with the optical elongation noted for the F/5 camera.
Two mitgations
were introduced. First, the ID of the lens cell was polished/deburred
using 600 grit paper. Second, another kapton piece was added to provide
a second set of front surface "fingers" to share the load
with the existing set. This piece only affects the front surface --
it was cut to be short enough along the optical axis so as to not
ride under the twin axial reference feet in the cell. The cell was
reassembled and the freedom of the lens to move under the assist of
the axial spring was checked. The motion appears satisfactory, but
given the thickness of the lens, more spring force may be necessary
in the future. For now, the spring force is adequate to not allow
the lens to float free when the cell is shaken moderately strongly.
The L4 cell was
also pulled from the block, the rear C-ring popped and the axial spring
action checked. All appeared well. The lens was not removed from its
cell, but showed no obvious problems. Both L4 and L5 cells were returned
to the camera block.
The block was
turned over and the L1 and L2 cells were removed. L1 was the other
possible suspect for the image elongation, due to its strong power
and lead location. This cell's C-ring was removed and the axial spring
action checked and found to operate satisfactory. The wave washer
under the C-ring was either cut or broken at one bend location and
when the cell was assembled, the broken ends of the wave washer overlapped,
providing a nasty line contact with the lens. This washer was pulled
and the two broken ends cut back 40-50 mils to clear their overlap
when installed. The ends were also deburred to prevent excess stiction
with the lens. The wave washer and C-ring were reinstalled and the
L1 lens cell deemed operational.
The L2 lens cell
was also opened at the C-ring and its axial spring action checked
and found to be satisfactory. We were unable to pull the L3 lens cell
without excess force. It would have been possible to pull L3 by removing
the ball keeper cylinder, but given how hard the ball keeper cylinder
is to put back in, and how unlikely it would be for L3 to cause the
image elongation (L3 is a virtual flat), we elected to inspect L3
in place.
None of the lenses
(L1-L5) showed any cracks, oil, or coating delaminations, though L5
did show a coating change outside the partial ring scoring on the
convex surface. This coating change could signify removal of the coating
by scraping from the ID edge of the cell.
The L1 and L2
cells were returned to the F/5 camera bore in the camera block and
the block was returned to the camera bay, including carefully setting
the hold down springs to the same compression values they had at removal
time. The motion of the camera block was checked by sending the camera
block to home and to the PV location and back. No scoring or interferences
were noted or heard.
- The detector unit
was opened and inspected. No aluminum shavings or dust was found. Similarly
the inside of the camera bay was inspected when the camera block was
removed for F/5 work and the bay bulkheads showed no new scoring and
the bay base showed no new aluminum chips or dust.
Summary: The
detector temp problem is very likely cured. The FW3 "sticking"
problem was probably actually back rotation due to poor balance. Rebalancing
and spacer lapping are likely to have cured the FW3 problem. After closing
up Mimir and pumping for one day, the leak rate was checked and found
to be 5.0x10^-5 torr per 30 minutes. This is 1/2 of the previous value,
so we conclude that the connector replacement and needle valve removal
have reduced the leak rate. The F/5 camera optics problem was expected
to be difficult to find and diagnose, based on the available image data.
We believe the problem is traceable to centration and tilt problems during
cooldown for one lens, and suspect L5 is the most likely culprit. Mitigations
were applied to L5 and its cell. We will not know if this worked until
Mimir is cold, sometime later this week.
|
