Clio is a 3-5 micron imager and coronagraph built to exploit the unique sensitivity and resolution of the MMT deformable secondary AO system. A particular area of focus for the instrument is to provide sensitive detection of giant planets at 3.8 and 4.8 microns. The instrument was designed and built by a team at Steward Observatory including Melanie Freed, Andy Breuninger, Ari Heinze, Suresh Sivanandam, and Phil Hinz. The camera is currently being commisioned and is currently available on a collaborative basis. Inquiries into the availability of the instrument should be made to Phil Hinz.

CLIO Upgrade

The detector for Clio is an InSb 9809 ROIC in a 320×256 format. Pixels are 30 microns on a side.

We operate the detector by taking exposures set by the sky brightness and coadding them in the memory of the computer to achieve the required integration time.

The detector reads out through 4 independent channels which have slightly different gain. As of April 2006 the gain for the detector in electrons / ADU are as follows:

Ch1: 81.86 ± 0.58

Ch2: 83.20 ± 0.66

Ch3: 91.79 ± 0.79

Ch4: 93.55 ± 0.76

The channels read out in successive columns, so Channel 1 is column 1, 5, etc.

These results were derived using 8 different exposure times 25 frames each. A window was used that removed the first and last 10 rows and columns. A 5-sigma clip of the data was used to remove hot and dead pixels. The errors are 1-sigma errors from the fit. The fit was error- weighted by the standard deviation of the mean of the variance measurements of each of the 25 images.

The approximate full well of the detector is 40,000 DN or 3.3 million electrons.

The approximate read noise of the detector is 700 (?) electrons.

For the April 2006 run the DSP code being used is tim.lod. It takes 59.1 ms to clock through the array in integrate while read mode (IWR) This means that the true integration of the detector is the set integtration time +59.1 ms. For the default lod file it is not possible to have an effective integration time of less than 59.1 + 5 = 64.1 ms. If shorter frame times are needed another DSP code can be loaded. We expect that the different DSP code as well as the effective integration time will be handled more transparently in a future software upgrade. For the moment, Caveat observor!

The measured duty cycle or efficiency of the April 2006 DSP IWR code is 92 % for an effective integration time of 100.1 ms (software integration time of 41 ms), obtained by having the software request images in blocks of 20. One frame of dead time exists for each block in the April 2006 configuration. Additional dead time occurs each time a frame is displayed to DS9, so increasing the number of coadds in routine data taking in multiples of 20 is recommended.

The saturation time of the detector when blanked off at the second filter wheel was approximately 10 seconds on April 06, 2006. This was for a detector temperature of 56.4 Kelvin. This is approximately 300,000 e-/s for a well depth of 3 Me-.

The AP bits were set all low as well as the PWA bits (AP000 PWA00) in an attempt to minimize detector glow and thus dark current. It appeared to help somewhat. The detector is set up to read through all 4 channels in IWR mode.

Clio has two optics trains creating either an f/20 or f/35 final focal ratio. A pupil imaging lens is used as well for initial alignment.

There are two aperture wheels in Clio. The first is at the image plane (field stop wheel) which is the Cassegrain focus. The second is at the image of the telescope secondary (pupil wheel). Each of these are six position wheels accepting 1 inch diameter masks. The wheels contain the following masks. The listing is clockwise from the home position.

Last updated March 2006.

1. Small pinhole (home) (50 microns?)
2. Blanked off
3. Open
4. Large rectangle (size of f/20 field)
5. Small rectangle (size of /35 field)
6. Large pinhole (1 mm?)

Last updated July 2006

1. f/20 stop (home) (4.95 mm)
2. M band phase plate
3. L band phase plate
4. Oversized Stop (5.6 mm)
5. f/35 stop (3.8 mm)
6. f/20 stop smaller (4.6 mm)

There are two filter wheels in Clio placed just after the pupil stop. Each wheel is capable of holding 8 one inch filters. The depth of the receptacles is approximately 0.33 inches. As of November 2006 The filter wheels contain the following filters. Listings below are listed in clockwise order from the home position.

1. Blanked off (home)
2. MKO M’
3. Barr M
4. x0047
5. 3-5 micron (Janotech)
6. Barr L’
7. Polarizer (Meadowlark Stock# C001151 Part# VL-100-IR-R)
8. Open

Electronic versions of the wavelength sensitivity of three of these filters are shown below:

A postscript version of the above graphic: clio_plots_v3.ps

The associated text files for the filter transmission curves are in this tarball: clio_filters_v3.tar.gz

1. Blanked off (home)
2. 0.1% transmissive ND (Janostech P/N FX90-0099)
3. 1% transmissive ND (Janostech P/N FX90-0098)
4. 10% transmissive ND (Janostech P/N FX90-0097)
5. 3.4 micron filter (JDSU N03401-6E)
6. Open
7. Barr Ks
8. Barr H + 5mm thick BK7 window

Expected sensitivity calculated from approximate sky, telescope and detector background are the following. These are the theoretical numbers we should be achieving. Actual sensitvity for the June 2005 run was approximately 0.5-0.7 magnitudes brighter for M band, and 1-1.5 magnitudes brighter for L’ band.

Assumed dark current was 0.5 Me- /pix/s. These calculations are for f/20.

5 sigma expected AO point source limits (Vega magnitudes)

5 sigma AO surface brightness limits (Vega magnitudes/sq. arcsec)

If we are not using AO we are limited to the following under median seeing

5 sigma non-AO point source limits (Vega magnitudes)

April 2006 photometric calibration, in brief:

We give the photometric calibration in terms of the counts per second received from a star of magnitude 10.0. The errors on these calibrations, given in magnitudes, are the sum in quadrature of the error on our own measurements and the photometric uncertainties on the Legget et al (2003) standard stars we have used.

April 2006 photometric calibration

Estimated throughput:

L’: T = 0.67

M: T = 0.42

April 2006 photometric calibration, in more detail:

Note: to see the June 2006 photometric calibration, scroll down past all April results.

On the night of April 8, 2006, the star HD 106965 was observed in L’ and M bands for photometric calibration, and on the night of April 10 the star HD 162208 was observed for the same purpose. Eight images of HD 106965 with 15 coadds, and 1.0596 sec of true exposure per coadd, were obtained in the L’ band, while 8 exposures of 150 coadds with 0.1596 sec of true exposure per coadd were obtained in M band. For HD 162208, the data consisted of 8 exposures in L’ of 10 coadds with 1.0596 sec of true exposure per coadd, and 8 exposures in M band with 90 coadds and a true exposure of 0.1596 sec per coadd. There was light patchy cloud in the sky at the time of the HD 106965 measurements on April 8, although the target appeared to be cloud-free throughout the observations. There was some concern, notwithstanding, that invisible faint cloud could have affected the measurements of HD 106965. This concern is not borne out by the data. The images of HD 162208, made in very clear conditions, show unusually bad AO correction, which is probably indicative of bad seeing. The photometry of this star does show evidence of an unusually wide halo.

Legget et al 2003 record HD 106965 to have a brightness of L’ = 7.311 +/- 0.010 and M’ = 7.32 +/- 0.04. For HD 162208 they give L’ = 7.125 +/- 0.011 and M’ = 7.05 +/- 0.02. We reduced our data and obtained photometry with apertures of 30.0 pixel and 10.0 pixel radii. We standardized the calibration to counts/sec for a star of mag 10.0. The results are as follows, where the errors given are the image-to-image rms, and do not take into account the Legget et al photometric uncertainties:

April photometric calibration, normalized to cts/sec for star of mag = 10.0

Note that we have made the approximation that any color term between M and M’ is negligible. This is true. The actual value of the color term for different stellar spectral types is given in Legget et al; it is usually less than 0.005 mag.

Examining the table above, we see that the 30.0 pixel radius photometry in L’ is very consistent, and both measurements have low fractional rms, indicating that there are no invisible clouds affecting HD 106965, and that, bad though the seeing was during the HD 162208, it is not bad enough to throw much light outside a 30.0 pixel radius. Thus either value can be adopted as a good calibration for measuring L’ magnitudes with a 30.0 pixel radius. We choose the HD 162208 value because of its lower noise.

The L’ calibrations with the 10.0 pixel radius show the effect of the very bad seeing during the HD 162208 observations: the smaller radius photometry has a large fractional rms and is too low to be consitent with the HD 106965 result. Thus, for L’ photometry with a 10.0 pixel radius, we prefer the HD 106965 measurement.

The M band calibrations with the 30.0 pixel radius are marginally consistent, but both have a very large fractional rms. This is because these stars are fairly faint at M band, and intense noise from the bright sky introduces significant error into the large aperture measurements. Thus we do not recommend trying to perform photometry at M band using a 30.0 pixel aperture. If, for some reason, a 30.0 pixel aperture is required, we recommend using the weighted average of the HD 106965 and HD 162208 values: 5067 +/- 434.

The M band calibrations with the 10.0 pixel radius have much lower fractional noise. The seeing at M band is always much better than at L’, so the HD 162208 10.0 pixel measurement may be fairly good. Still, we adopt the HD 106965 result because it has lower noise and is slightly brighter, suggesting the HD 162208 one may still be slightly affected by noise.

These choices result in the table of final photometric calibrations given under the heading ‘April 2006 photometric calibration, in brief’ above.

Measured Photometric Sensitivity from images of GJ450:

Using an April 2006 integration on GJ450 that lasted 5355 sec, we obtain a background-limited, 5-sigma point source detection limit of about L’ = 17.1 - 17.65. This is based on the HD 166208 calibration mentioned above.

Scaling this to 1 hour we get L’ = 16.67 - 17.22.

June 2006 photometric calibration, in brief:

We give the photometric calibration in terms of the counts per second received from a star of magnitude 10.0. The errors on these calibrations, given in magnitudes, are the sum in quadrature of the error on our own measurements and the photometric uncertainties on the Legget et al (2003) standard stars we have used.

June 2006 photometric calibration

June 2006 photometric calibration, in more detail: The only star on which the above calibration is based is HD 162208, observed in good seeing and clear weather. The reason for the reduced counts relative to the April calibration is a pupil misalignment problem that plagued almost all of the June data. It is not certain that the degree of the pupil misalignment remained constant, but there is evidence that it was not zero on any of the science observations made before it was identified and fixed, and there is no evidence that it changed. I think a long integration on Altair by Eric Mamajek at the very end of the June run is the only good science data that was taken after the misalignment was identified and fixed. This misaligment reduced our sensitivity and slightly broadened our psf in one dimension, but other than that did not cause any harm. That is, the data remain good scientifically, albeit somewhat less sensitive than they might have been.

The Ks band calibration is based on the L’ magnitude in Legget et al. HD 162208 is an A0 star, so the assumption that Ks-L’ = 0.0 is not unreasonable. Allen’s Astrophysical Quantities gives Ks-L’ = 0.0 for an A0 star.

July 2006 photometric calibration, in brief:

We give the photometric calibration and its uncertainties using the same convention described above. Two photometric standard stars were measured during the July run, HD ?? on the night of July 11, and HD 203856 on the night of July 12. Only the July 12 data has been analyzed so far.

July 12, 2006, photometric calibration based on star HD 203856

September 2006 photometric calibration, in brief:

We give the photometric calibration and its uncertainties using the same convention described above. A single photometric standard star was measured during the September run, HD 203856 on the night of September 10.

September 2006 photometric calibration based on star HD 203856

December 2006 photometric calibration, in brief:

Calibration based on star HD 22686, observed on the night of Nov 30.

December 2006 photometric calibration based on star HD 22686

January 2007 photometric calibration, in brief:

Calibration based on star HD 22686, observed on the night of January 3.

January 2007 photometric calibration based on star HD 22686

During the January run atmospheric extinction calibrations were also performed for the L’ and M-bands.

January 2007 extinction calibrations in units of mag/airmass

April 2007 photometric calibration, in brief: The stars HD 106965 and HD 136754 were observed on the night of April 10, 2007. A new dichroic was installed in Clio before this run, apparently improving the throughput noticeably. The mean values in the table below are straight averages of the values from the two stars observed, corrected to a consistent airmass of 1.03 using the January airmass slope values and assuming the Ks slope is the same as that for L’. The uncertainties on the mean values are set equal to the mean square sum of the individual errors, or the difference between the values for the two stars, whichever is greater.

April 2007 photometric calibration

Even though, remarkably, the 30 pixel M-band result shows a lower uncertainty than the 10 pixel result, we still recommend using a 10-pixel aperture for M-band photometry, unless the source is very bright. For L’ and Ks, as usual, we recommend a 30.0 pixel aperture.

During the April 2006 run, crude measurements of the sky and instrument emissivity in L’ and M were carried out. Dome flats and 2 airmass sky flats (which include thermal emission from telescope structure and instrument) were taken along with corresponding darks to subtract off the zeropoint. The emissivity was estimated by finding the ratio between the sky flux and the dome flux since the dome has an approximate emissivity of 1. To disentangle the sky from instrument emissivity a science exposure with an airmass of 1.15 was used. Note, however, the darks were right taken after sky and dome flats, so even though the science exposure was closest in time to the darks, the zeropoint may have shifted significantly. Therefore the emissivity measurements are good to about 10%. The preliminary results are shown below:

Emissivity Measurements

From the L’ data, one can estimate the instrument emissivity to be 10%, if the L’ sky emissivity is 6%. However, the M-band data is not consistent with the above emissivity. M-band data implies a sky emissivity of only ~10%, which gives an instrument emissivity of 20%, when the expected value is 15-20%. One explanation for this is the saturation of atmospheric lines at higher airmasses where the atmosphere is no longer optically thin. The simple minded assumption that the emissivity scales as airmass is no longer true. Therefore, since the L’ window is devoid of strong lines, the L’ emissivity measurements are likely to be closer to the truth.

The Clio computer is located in the blue electronics case in the observing chamber. The camera can be run on any remote linux computer by ssh.

At the beginning of the night:

1) Login to clio using the command “ssh obs@clio.mmto.arizona.edu -X -Y -C” Make sure you have talked to the previous observer or email Phil Hinz to get the password.

2) Setup a directory for the night within the Data/ directory. The normal convention is to create a directory for each night, labeled by the UT date. For example, on the night of June 10 I would login, “cd Data” and then type “mkdir n060611”

3) Type “runclio” to start the various programs. “runclio” is a link to mirkwood.sh in the Programs directory. This in turn is a shell script which runs mirkwood (the clio control program), ds9 (the image display program), Cliomotors.tcl (the filter control program), starnudger-Clio.tcl, (the telescope paddle control), and statcurrent.tcl (the statistics display program).

4) In the xterm, at the prompt, type “init”. It will ask you which lod file to use. Choose the default (tim.lod).

5) Type “image”

6) Type “set” ??

7) Type “query ao” to turn the AO header info on. NOTE: IF THIS IS NOT TURNED ON YOU WILL NOT HAVE TELESCOPE INFO IN YOUR FITS FILES

8) Set the data directory to the directory you made earlier.

9) Choose a prefix for your files.

10) Set integration, number of coadds and other settings. You are now ready to begin observing.

Most of the control of the Clio camera is done with the text program in the xterm. This program has commands divided into a set of menus. The majority of the observing is done from within the “image” menu.

Images are acquired by the Clio electronics and transferred to the computer memory, where they are coadded to create a single FITS file. Unless you have a really bright star you should set the integration time according to the sky brightness. The typical rule of thumb is to set the integration time to correspond to 2/3 of the well capacity. As of April 2006 the dynamic range for Clio is 10,000 at zero flux to 50,000 at saturation. You should then set the number of coadds to a reasonable value to achieve the desired integration time in each position. For L’ the typical times are 1200-2000 ms and 5-10 coadds. For M a typical setting is 100-200 ms with 50 coadds(??).

It is important not to set shorter integration times than necessary. There is significant read noise associated with each frame which can degrade the signal to noise if many short exposures are taken.

To set the nod vector, type “nod off” and fill in values in arcseoncds for the desired throw. The values are for the star movement on the array where +RA is equivalent to a star movement of +y and +DEC is equivalent to a stat movement of +x. (Yes this is obscure and we hope to fix this soon).

You can set up more complex nodding or dithering patterns with a command file . . .

The telescope can be commanded to move through the starnudger program. As the name implies, the arrows will nudge the star around on the array. Thus pushing the right arrow with a value of 1 will cause the star to move right 1 arcsecond or approximately 20 pixels.

If the starnudger program has failed, kill it by searching for its PID number “ps -elf | grep starnudger” and then type “kill 1300” where 1300 should be replaced by the PID. Restart it by changing directory to Programn and type “./starnudger.tcl”

As data is taken it will automatically be displayed in the ds9 window. To see fainter objects you will want to offset the star off the array and then type “dark”. This takes an average background frame. Now type “use dark”. This is a toggle (true or false), so when you need to take another dark you will not need to type “use dark” again.

Not that the dark frame is saved separately and is not subtracted from the actual data. It is only used for display purposes.

Information about the currently displayed frame can be seen in the “statcurrent” window. This can be used to check sky brightness, check for the number of saurated pixels, or estimate the size and location of the star (this is currently still being completed).

The motor control program for Clio is a tcl script which controls four motors: the field stop (not used for most observations), the pupil stop (DO NOT TOUCH), and filter wheels 1 and 2. Se the listing earlier in this document for which filters are loaded. It is recommended that you home the filter wheel each time before going to a new filter.

DO NOT, EVER, PRESS “HOME ALL WHEELS”! This begins a self-destruct sequence in Clio. Actually, it resets all the wheels. This is bad for the pupil wheel which is fine-tuned when Clio is mounted. The current program is an engineering interface which will be replaced soon.

Nodding should be carried out every 0.5-2 minutes to ensure good sky subtraction.

Flats are typically acquired by pointing at blank sky and taking frames, followed by putting a blank in the filter wheel and taking another set of frames. The difference should provide reliable flats.

In brief: The Clio plate scale is 0.048574 +/- 0.000090 arcsec/pixel. Raw Clio images are mirror-flipped. In processing they can be mirror-reversed right-to-left and then rotated 272.98 deg - PA + RA clockwise to get North up and East left on the images (here, PA means parallactic angle and RA means the setting of the MMT instrument rotator).

In more detail: The plate scale in April 2006 was determined by Ari Heinze to be 0.048574 +/- 0.000090 arsec/pixel, based on observations of the double star HD 100831. The double star HD 115404 was also observed, and from it a consistent calibration of 0.048620 +/- 0.000323 was obtained. The uncertainties are completely dominated by uncertainty in the true seperation of these binaries, not by Clio measurement error. Note that the small difference between the values for the two stars (less than 0.15 sigma) may indicate that the uncertainties are overestimated.

In normal Clio operation, with the MMT instrument rotator set to 0 degrees, +ALT on the telescope is approximately to the left on the Clio detector. Raw Clio images are mirror-flipped with respect to the sky, so that to get ordinary correct astronomical images with North up and East left Clio images must be mirror-reversed in processing, and then rotated. If a Clio image taken at 0 degrees rotator angle is mirror-flipped right-to-left, telescope +ALT is then approximately to the right on the image, so a clockwise rotation of 270 degrees is required to get telescope +ALT to be up on the image. The exact angle required was determined from the two double stars mentioned above. For HD 100831 an angle of 272.98 +/- 0.15 degrees was found, and for HD 115404 an angle of 273.00 +/- 0.20 degrees. Note again that the uncertainties are dominated by uncertainty in the true position angles of the double stars, not Clio measurement error. Also, as before, it appears the uncertainties may be overestimated. A further rotation, dependent on parallactic angle, is required to get North up on the images. If the parallactic angle (PA) is defined in the usual way as the celestial position angle of telescope +ALT, then the full rotation required to get North up and East left on Clio images that have been taken at telescope rotator angle 0 and then mirror reversed right to left is 272.98 degrees - PA, clockwise. If the rotator angle is not zero, the full rotation required becomes 272.98 degrees - PA + RA, where RA is the rotator angle. It is wise to test any rotation scheme on a star with a real companion to guard against bugs and insure that the PA values being used are accurate. Data sets containing real companions of known seperation and PA, including the binary stars used for this calibration, are available from Ari Heinze.

In brief: The Clio plate scale is wavelength dependent. The following values were measured using double star HD 100831:

Ks: 0.049066 +/- 0.000090 arcsec/pixel

L’: 0.048590 +/- 0.000090 arcsec/pixel

M: 0.048681 +/- 0.000090 arcsec/pixel

Raw Clio images from this run are mirror-flipped. In processing they can be mirror-reversed right-to-left and then rotated 272.8674 deg - PA + RA clockwise to get North up and East left on the images (here, PA means parallactic angle and RA means the setting of the MMT instrument rotator). The accuracy on this angle is about 0.15 deg.

More detailed comments: The same star was used for the main April 2006 photometric calibration. The vast majority of the uncertainty in the above calibrations is from the 61” measurements of the double star’s seperation in arcseconds. The internal precision of the MMT measurements is at least four times better. This explains the agreement between the above L’ plate scale and the April result with far less variation than the errors would imply. Ari Heinze hopes eventually to use this star as a primary astrometric standard, so that all Clio astrometry can be linked to it, providing a fourfold increase in Clio’s ability to measure changes in the relative positions of companions to stars from run to run. This will not improve the calibration accuracy relevant for comparison of Clio astrometry with that from other instruments. The calibration uncertainties relevant for such comparisons will remain those quoted above.

No double stars on the list of 61” astrometric calibrators were measured during the July 2006 run. However, the double star GJ 896AB was observed, and as soon as it can be tied to the primary standard HD 100831 as discussed in the June Astrometric Calibration section, an accurate calibration should be available for July 2006.

An alteration in the Clio software made September 2006 the first Clio run for which raw images are NOT mirror-flipped. Raw Clio images taken at rotator angle 0.0 may be rotated clockwise 92.577 deg to get +EL up on the images. From this point the further rotation needed to get North up depends on the parallactic angle. The uncertainty on the rotation required to get +EL up is about 0.15 deg.

Using the double star HD 223718, the L’ plate scale is measured to be 0.048404 +/- 0.000135 arcsec/pixel. The difference between this and the June run is probably due to the different calibration star and not to an actual change in the Clio system. The difference between this calibration and the June one is 0.000186 +/- 0.000162 arcsec/pix, so the two calibrations are consistent within their uncertainties. If the Clio-specific astrometric system discussed in the June entry can be created, the September 2006 calibration will be tied to those of other runs with much greater precision. As with the June run, the astrometric calibration is found to be wavelength dependant, with the following values:

Ks: 0.048896 +/- 0.000131 arcsec/pixel

L’: 0.048404 +/- 0.000135 arcsec/pixel

M: 0.048563 +/- 0.000130 arcsec/pixel

These all give fewer arcseconds per pixel than the June calibration based on HD 100831, with the difference ranging from 0.24 % (M) to 0.38 % (L’). The ratios of Ks/L’ and M/L’, however, differ from the June values of these ratios by only 0.04 % (Ks/L’) and 0.14 % (M/L’). In no case are the June/September differences so large as to be inconsistent with the uncertainties.

It appears that due to the refractive elements in Clio the plate scale, though not the rotation angle, has a statistically significant dependence on wavelength. Also, as expected, changing 61” calibrator stars changes the astrometric calibration by much more than the internal precision of the MMT/Clio measurements, but not by an amount inconsistent with the uncertainties on the 61” measurements, which are higher. The ratios of calibrations for different wavelengths experience less fractional change on switching standards than the calibrations themselves, which is what we would expect.

These are preliminary notes. Please consult Phil if there is any confusion.

• Unwrap the tarp around Clio and evacuate the dewar with a turbo pump. You will need a 3/4” slip fitting. The fitting is mangled and needs some rough handling to get it to work.
• Fill the outer vessel of Clio with Nitrogen. Cool down will take ~12 hours.
• Once the rest of Clio is at nitrogen temperature fill the inner vessel with liquid nitrogen.
• Wait a half hour and fill the inner vessel again.
• Locate the solid LN2 tube (the stinger) and insert it into the inner vessel.
• Unpack the electronics rack and locate the solid nitrogen pump.
• Plug the solid nitrogen into the electronics rack and the rack into the back of the chanber.
• Uncoil the umbilical cord from the back electronics cover and hook up the pump. You will need a nw16 o-ring and clamp on one side. The other side attaches to the stinger with a hose clamp.
• Start the pump. Cool down to 55 K will take around two hours.

• The electronics rack should be positioned under the back edge of the primary mirror.
• Locate the cryostat mounting bolts and allen wrench in the electronics drawer.
• Attach the cryostat to the telescope using at least three 3/8-16 bolts on each side.
• Attach all the cabling in the umbilical to the connections on the cryostat.
• Strain relief the umbilical by using velcro straps attached to the bottom rail of the cryostat.
• Arrange the cable so that it will hang down when the telescope is at horizon.
• You should now be able to read the temperature of the cryostat.

• Attach an ethernet cable to the computer and plug it in at the back of the chamber.
• Power on the computer and open the monitor in the electronics rack.
• Determine the hostname of the computer once it is booted up. It should be something like paw9.sao.arizona.edu
• Run the program astropci_load to load the Leach device driver
• Change the permissions on /dev/ttyUSB0 to allow rw access.
• Check that you can access and run Clio remotely.

Clio is put on the turbopump and when the vacuum gets down to about 7e-6 Torr, it is good enough for a typical Clio AO run.

The outer vessel is filled first, and this takes typically 10 minutes to get the cooling down going.

If the detector is the coldest surface in the dewar, the detector will attract all residual vapours!

When the outer vessel temperature is 150 degrees or below, you can fill the inner tube. The limit of 150 degrees is important, as this is the temperature that the getter in the dewar activates.

Fills at 11.30am 2008 Mar 20, but INNER not needed. OUTER requires only small topup. Inner vessel on pump at about 2.45pm, temperature change is 10 degs in about 10 mins to 65K.

Liquid N2 dewar should be filled every day, the solid N2 dewar every two days.

Liquid Nitrogen Dewar

An LN2 bottle is pulled over, and the metal rod inserted into the dewar. A solid metal rod from the top drawer of the Clio electronics rack is used to restrain a loop of the elastic filling tube. When the LN2 is turned on, this prevents the metal filling tube shooting out of the tank. HOlding it lightly in place until the cold vapour freezes te tube in place, it takes about 5 minutes to fill the LN2 tank. The sign of a full tank is LN2 overflowing and spurting out of the tank.

Solid Nitrogen Dewar

The pump is switched off, and the pumping tube is removed from the dewar, taking care not to lose the Teflon washer that seals the solid nitrogen dewar tank. The filling procedure is the same as for the Liquid Nitrogen tank, and the Teflon washer and pumping tube reattached. You may use a wrench to firmly fix on the pumping tube, and then switch on the pump again.

Temperature of the dewars is checked at the digital thermometer display on the Clio Electronics rack.

To store in 4th floor library:

• two grey plastic clio boxes
• one vacuum pump
• Clio electronics rack mount box
• Clio on the wooden castor stand.

Procedure:

• Disconnect cables at Clio dewar end
• Stop pump and remove solid LN2 pipe
• Unclamp the plastic hose from the solid LN2 pipe, and store it along with the small black stopper.
• Move all clio cables off the instrument lift out of the way.
• Put plastic cover on dichroic.
• Get wooden castor platform and put on instrument lift.
• Rise lift to Clio with about 1/4 inch from clio framework.
• Undo clio bolts, lowering onto the wooden platform.
• Remove MMT electrical extension cord.
• Unplug Clio network cable and put in drawer or back.
• Disconnect clear hose from top of the pump by removing the clamp and vacuum washer fitting.
• Refasten clamp on with the vacuum washer in place.
• Coil cables into the back of the Clio electronics box.
• Cover the Clio ends of the fibre optics with the dummy fiber optic connectors in the clear plastic case.
• Put Clio in elevator and take up to 4th floor library.