The flexure of the instrument was measured primarily
by the motion on the CCD of a fixed 100 m pinhole in the slit mask. The
instrument was rotated on its axis with the instrument rotator module.
The instrument module was held in the test stand. While in the test stand
the instrument and rotator module could be rotated in elevation between
zenith and the horizon, much the same way they will when held in the telescope.
The angles were varied by increments of 45 degrees in each axis. The spectrograph
was operated in all three modes for a complete series of tests at all angles.
The results of the tests are summarized on Figures
The results show that pointed at the horizon we have
a ± 3-pixel motion of the spectra on the detector. At 45 degrees
elevation the motion was about ± 2 pixels, and at zenith ±
1 pixel. The requirement was that spectra not move more than 2-pixels in
a 2-hour exposure (30-degree change in elevation or rotation). Except at
the zenith this amount of motion is well within the requirement. At zenith
the motion would be contained within a 2-pixel circle, but the total path
length of the motion could be much more. The flexure is moving at twice
the angular rate of the rotation in the bearing and with the telescope
near zenith, the rotation can change a large amount for a small zenith
Move and repeat tests
Automatic scripts were written to move all axes multiple
times forward and back checking both the repeatability of the stages and
the keywords that drive the axis from software. Bob will
write more on this
The back focal distance of the camera was measured
in the Optics Lab with a microscope. The on axis images looked good, as
did spots to the edge of the 4-degree field. A problem does exist for which
we do not have a complete explanation. The back focal distance as measured
by the microscope, is 0.008" back of where the optical design predicts
it to be. This can be partially explained by the fact that elements in
the final triplet of the camera are likely in contact rather than spaced
at 0.003" as specified in the optical design. The reason for the spacing
is that the coupling used in this triplet is grease, and the method used
to grease elements together does not allow for spacer shims.
Double pass interferogram
A double pass interferogram was done through the
camera and results agree well with what the design would expect.
A star test of the camera was done using a 50 m
pinhole at the focus of a collimating telescope. The image size was as
expected, although the image quality was less than what we had expected.
After considerable investigation, we concluded that the limiting factor
in the image quality was the primary mirror of the collimation telescope.
Spectra was taken using the continuum source and each of the line lamps. Both a 100 m pinhole and a 0.005" slit were used. The spectra are shown as Figures 4 and 5
The results show that all the orders are present and that we have adequate image quality over the full field. As the calibration source does not light the slit with the same illumination as the telescope, the collimator was stopped down with a pupil size aperture stop at the center of the field.
The only surprise is that the cross dispersion of
the spectra is about 5% less than we would have expected from the optical
design. The explanation may be that the spacing of the prisms is slightly
different than the design or that the index of the prisms is slightly different
than we specified.
Low dispersion mode
A slit mask with line of 0.005" holes was used with
the three calibration source lamps to put low dispersion spectra over the
8-minute long field. The spectra is shown as Figure 6. As expected the
spectra is at a 9-degree angle to the rows of the CCD, which is a feature
of the optical design. The image quality looks ok, but is a bit hard to
judge as off axis the pupil is not properly illuminated for the same reason
as was explained previously.
A matrix of 0.005" inch holes was installed in the
slit mask and imaged on to the CCD with the continuum source. As the matrix
we were able to fabricate did not cover the entire field, the slit mask
wheel was moved, so that in three images we did cover the field. Figure
7 is a composite image of the dots across the whole field. To the extent
we can tell, there are no problems showing with the image that would indicate
a problem with the optics or the spectrograph.
Calibration system tests
We have run the calibration light source in the spectrograph,
bouncing light off the hatch. The intensity of the source seems to be adequate.
Integration times vary from 2 to 60 seconds depending on the line lamp
source, neon being the brightest and copper the least bright. We can take
a spectra of the continuum light source in Echellette mode in 1-second.
We do not expect the continuum source to be uniformly illuminated to better
than a few percent, thus observers will need to use either a dome or sky
source to flat field their data.
Bob will write more on this
The dewar has a hold time of approximately 17 hours
when the spectrograph is in the storage position. We have seen no evidence
that this hold time is very effected by motion of the spectrograph in either
rotation or elevation.
We have had a problem with something condensing out of the vacuum onto the CCD. What we believe this condensation, which we have seen at various times, is from several sources. Some of the original sources may have been from materials that we used in the dewar system, but have since removed. The dewar was cleaned very thoroughly after the possible contamination sources were removed and after some initial deposits on the engineering grade chip, nothing was observed.
More recently when the science chip was re-installed,
we did see some contamination re-appear. We believe this was water contamination
introduced when we did the QE tests with the radiation source. The window
we use for this test leaks at a fairly high rate, and over the course of
the 14 hours of tests, it is possible we introduced an amount of water
vapor. Continued pumping and thermal cycling of the dewar has eliminated
this contamination source and we are currently not seeing anything condensing
on the CCD.
TV system tests
A matrix of dots were imaged with the TV system,
both on the guider mirror and on the slit mask. We are able to focus the
camera on the 4-minute square guider field and see features over that full
The glycol connections have been changed out to more
robust stainless steel ones, and the system does not leak
Electronic box tests
The cooling system in the boxes has been run with
the boxes sealed. The coolant was 5°
F lower than the ambient temperature of the room, and the electronics stabilized
at 5° F warmer
than the ambient temperature of the room.
Instrument control and software tests
Bob will write more on this
Data tacking system tests
Bob will write more on this