These instructions are based on a set of notes taken during observing runs on 14 Oct and 11 Nov 1996. Xiaopei Pan was the PI, and he showed me the startup procedures, which I recorded as best I could. There is no guarantee that they are complete or correct, especially since the instrument itself is in a state of flux. Nevertheless, it is hoped that they will be useful to future observers.
These instructions cover only the most basic setup and a few of the error recovery procedures. A very concise description of the startup and operation is availabile in /home/pti/writeup/startup.txt Additional information is available in a set of files in the /home/pti/writeup directory on horatio. Links to these documents are provided below.
The data returned are in the form of a single fringe visibility measurement per observation. Earth's rotation enables visibility measurements to be made in a series of points in the (u,v) plane, but reconstruction of an image is not possible. Future enhancements include implementation of a "dual-star" mode in which the fringe tracking is done on a bright (K<5) star within the isoplanatic angle (~30 arcsec) of the science target, which can be considerably fainter (K~15?). In addition, a third aperture is being brought online, although it will not be possible to use all three simultaneously.
PTI consists of several discrete subsystems: a pair of siderostats (flat, steerable mirrors) and their associated telescopes, a star acquisition system with a CCD camera for each siderostat, a set of delay lines used to equalize the path length from the source to the beam combiner, metrology systems for accurately measuring the lengths of the delay lines, a star tracking system with a tip-tilt mirror (?), the beam combiner itself, an infrared array detector used to sense the ~2 micron light, and a rather complicated computer setup used to control it all.
PTI is currently housed in a main building containing the delay lines, beam combiner, detector, and warm room, plus a small outbuilding ("hut") for each siderostat+telescope. Each hut is approximately 50 m from the main bulding, and is optically connected to it by a pair of PVC pipes of ~5" diameter which is used to block airflow across the light beams. For single-star (a.k.a. primary) observations, only one of each pair of pipes is in use; the second pipe is used to carry the beam of a secondary star during dual-star observations.
Switching between these filters is not completely trivial, since in addition to changing the filter wheel in the dewar, also it requires changing the delay-line stroke and detector readout timing, replacing the prism in the spectrometer, and realigning both the spectrometer and white-light optics in front of the dewar. The details are given in the file /home/pti/writeup/H-99.txt . Note that it may be a bad thing to switch store data taken at different wavelengths in the same directory, since it can make the reduction more difficult.
One may independently choose the length of time the detector integrates
between readouts. Frame times of 10 and 20 mS are available.
Because sensitivity of PTI measurements is typically limited by the detector
readout noise, and because the detector spends a smaller fraction of its
time reading out when used in 20 mS mode, 20 mS mode allows PTI to observe
stars ~3 times fainter than it can in 10 mS mode. The catch is that
the longer frame time reduces the bandwidth of the fringe tracker loop,
so it doesn't work well when the atmospheric seeing is fast. Switching
between 10 and 20 mS mode is done by running a commandline script, rebooting
the fringe tracker, and resetting the amplitude and frequency of the delay-line
stroke. Detailed instructions are in /home/pti/writeup/20ms.txt
.
Here is a map of the optics room at PTI.
^ ^ ^ ^ | \ | Delay | / | | \ | Lines | / | | \ |___________| / | ______________| \ ___________ / |______________ | | | __________________ | | __________________ | | | Secondary Bench | | Metrology | | Primary Bench || | | | | | | || | | Dewar | | | | Dewar WL || | |__________________| | PC | |__________________|| | |___________| | | | |_______________________ _______________________________| liquid Nitrogen tank
Maintainence of the dewar should be done only be people experienced with such things, because although it is basically pretty simple there are a number of things that can go badly wrong. The vacuum must be maintained by pumping through a valve on the side of the jacket when the dewar is at room temperature. Typical dewars can go about a week between pumpings before the vacuum goes "soft" and the nitrogen boiloff rate becomes unacceptably high. When the valve between the dewar and the pump (or the outside air) is closed, and there is a good vacuum in the jacket , liquid nitrogen can be added to the can through the fill hole on the top surface of the jacket. There is a funnel and a set of hand dewars (and gloves!) for this purpose. Don't put the funnel directly into the secondary dewar; put the notched metal cylinder on first to leave an exhaust hole for the nitrogen gas.
An LED display next to the dewar shows the temperature of the detector in degrees Kelvin. If it's close to room temperature when you start adding nitrogen, it should take an hour or so to reach the ~77 K operating temperature. If the temperature reads 77 K, then all is probably well.
Note that the secondary dewar, which is located on the secondary beam combiner bench and is larger than the primary dewar, is used for dual-star mode only. If the secondary dewar is cold (check its thermometer), then it's probably worth topping off the nitrogen in that one also. Generally it's good to avoid putting array detectors through many thermal cycles as this causes pixels to go bad.
Each hut produces two beams, the Primary beam and the Secondary beam. The Secondary beam is only used for Dual Star observing. The Primary beam path is on your right as you follow the pipes into the hut.
Flip up the shiny cloth cover over the optical bench and find the Primary corner cube. It's about an inch in diameter and is attached to a black aluminum stand about 3" high. Don't confuse it with the Secondary corner cube, which is gold-coated. Place it against the stops set up for it on the optical bench, labeled "Primary". Then remove the plastic cover over the end of the primary PVC pipe and inspect the window to make sure it's clean.
visible light goes to star tracker ^ ^ | | light from S hut | | --------------------> ---------->\ | N+S light for light from N hut Dichroic IR light v visibility measurement --------------------> -------> beam ------------>\ combiner | | | | N+S v | light fiber | for fringe | | tracking | v v detector<-- prism <-- / (in Dewar)Here is a crude map of the Primary bench. Locations of things you will need to adjust have asterisks.
______________________________________________________________________________ | visible light goes up here | | | | Questar star tracker | | | | | | | | compensator | | /Di- | \ coupler | | /chroic \ | \ \ | | / beam OAP | | combiner single | | || stops for > \\ mode | | corner cube* > OAP \\ fiber* | | 2 corner fiber | | cubes* \ ends | | \ here | | _____ align WL | | / \ spectral *here _ | | / \ /\/*fil-\\ l| | micro-| | | | ||WL* |> prism / \/ter \\ a| |ND scope | | | dewar | ||lens \ \goes s| |filter _ | | \ / ___ spectrometer WL \ /here e|_| * | | | | \_____/ --- fold * source \/ r +block |_| | |______________________________________________________________________________|
For each siderostat (North and South), put the alignment target, which is a piece of cardboard with a hole in it mounted on an aluminum stand, against its stops on the fringe-tracker table. Now adjust the mirror which sends the laser beam to one of the huts. This mirror is located on the optical bench close to the end of the delay lines, and right next to the end of its corresponding PVC pipe. It is one of four, and is labeled. Adjust the mirror by turning the two adjustment screws, either by hand or using the picomotor controller on the N side of the table. Try to get the return beam centered on the hole in the target. To use the picomotors, set the swich on the box to the adjustment you want to make and make the adjustment with the "a" switch on the remote controller. Repeat for the other siderostat.
Caution: Do not adjust these mirrors unless you see some return light (try translating the template left-right or up-down slightly to see where the return is). One rotation of a knob causes about a 12" translation of the beam at the hut, making it easy to get "lost" by adjusting so far that the light doesn't make it out the pipe at the end. If that happens, it's a pain to get back. If you don't see anything, first verify that light is indeed heading into the input side of the pipe toward the outside. Then remove the cardboard template you just installed in the lab, head back to the hut, and install the full-size template. This is an inverted-T bracket with target sides labeled "Primary" and "Secondary." Put this template in place of the corner cube with the "Primary" side facing out and use it to get the beam roughly centered. This will require running back and forth between the hut and the lab, unless you have a friend to help via the intercom. For the fine tuning, you can put the corner cube back and follow the instructions above.
Once the laser beams are aligned, go to each hut and remove the corner cube target and store it out of the beam. Get a segment of PVC pipe from rack on the back of the hut and carefully install it between the end of the long pipe and the optical bench. The hut end of the pipe should rest between a pair of screws on the optical bench. Pull the shiny cloth around the pipe to insulate the bench as well as possible. The South Hut has two pieces of cloth that fit aroun the pipe outside the hut wall that should be velcroed or taped on to keep air from flowing through and degrading the seeing. On the South Hut, put the section of wall back in place on the south side for the same reason.
Aligning the metrology optics for each of the long delay lines takes several steps. Move the delay lines to the back. Go to the center optical table in the lab. Flip the cardboard target down over the upper of the two ~2" mirrors which face toward the delay-line carriage, and tweak the rotation of the beamsplitter, which is a 1" cube on a small plexiglas-covered table between the mirror and the carriage, so that the laser spot is centered on the target. Then flip the target back out of the beam and put a piece of paper in front of the optical detector, which is mounted on the table with the beamsplitter. Using the white plastic knobs, adjust the top mirror so that the return laser spot is aligned with the outgoing spot on the detector. Then remove the paper and adjust the bottom mirror to maximize the signal on the oscilloscope. Since the laser is bright enough to saturate the detector in normal operation, making it difficult to see where the strongest signal is, you might want to put a neutral-density filter in front of the detector for this adjustment.
In the worst case, the alignment is completely off and there is no signal. If this happens, there is a set of instructions for recovering the signal in /home/pti/writeup/dlalign.txt
Once the signal strength is normal, send the delay lines to their fiducial positions (using FidSet at the GUI) to recalibrate them. We will do this in the next step.
On harbinger, cd to /home/pti/bin. Type 'beginNight' to create a data-directory link. Then type 'g &' to start the GUI. This launches the main control panel, labeled "Palomar Testbed Interferometer", which has a bunch of buttons that toggle the visibility of control windows for the various subsystems.
The next step is to establish the fiducial (home) positions of the siderostats and delay lines by driving each system against a limit switch. The encoders and metrology systems will monitor changes in position relative to the fiducials. Note that the fiducials for a subsystem are lost whenever the corresponding CPU is rebooted, making it necessary to these procedures if that happens.
Click "Pri N" under Siderostats to pop up the north siderostat control panel. Click "Off", then "Home" to send the north siderostat to its home position. Then click "Pri N" on the main panel again to hide the north siderostat control panel. Repeat this procedure for the south primary siderostat.
NOTE: On these panels, it is sometimes necessary to precede each click of a top-area action button by a click of the "Off" button to get the system to recognize a click.
Click "Psv" to get the passive delay line control panel, titled "Long Motor Only Delay Line". Click "Fid Set" button to send it to its fiducial position, then hide the control panel by clicking "Psv" on the main panel again. Repeat this procedure for the active delay line ("Act").
On the "Long Active Delay Line" window, check the "Show detailed status" box to expand the window. Open the Fringe Tracker window and look at the Frame Rate and the Desired stroke. Their values will depend on whether you are in 10-ms or 20-ms mode, and on on the observing wavelength; for example, for 10-ms mode at K there will be a Frame Rate of 100 and Desired Stroke of 3260, for 20-ms mode at K there will be a Frame Rate of 50 and Desired Stroke of 2934. If you want to change the integration time, now is a good time to do it. Otherwise, copy the Desired Stroke into the "Amplitude (nm)" field in the Long Active Delay Line window, and click the appropriate Frequency(Hz) button below that field to enter the Frame Rate. Then click "Send Stroke", followed by "Stroke Set".
The white-light source is a tungsten lamp controlled via a power supply located below the optical bench. Flip the solid orange "DC out" swich to turn on the WL source. Turn the voltage all the way up (~4.5 V) so that the current through the WL source is at the limit set by the current adjustment knob on the power supply. Then block the laser and look at the monitor. Adjust the two white plastic knobs on the mirror tip-tilt stage next to the WL source to put the WL spot on the crosshairs. These knobs are sensitive.
Note: This procedure is also described in /home/pti/writeup/ftalign.txt
Commands on the PC can either be typed carefully (the system is case-sensitive) or entered by typing the Alt key followed by a number. The Alt key macros are listed on a piece of paper next to the keyboard. You have to hit the "Enter" or "Carriage-return" key after typing each command or Alt macro. Note that the PC periodically prints out some status info, and this might come out in the middle of the command you're typing. Don't worry about that.
Type the following commands:
The PC will show a pair of centroid numbers, X, and Y, every detector readout. Adjust the white-light lens mount (it's on the optical bench, and is labeled "WL") so that the X and Y values are 0 +/-1. This will happen when the center of the bright spot is precisely on the 5th row down, and the 5th column from the left (the correct pixel is marked with an "X" on the display). The units of the centroids are 1/30 of a pixel.
The next step is to align the spectral axes. Do this by putting the narrowband infrared filter in front of the white-light source. The filter is taped to a small aluminum stand, and it fits on the little shelf behind the lens mount closest to the WL source. It is necessary to increase the voltage on the white-light source to ~4 V to compensate for the light lost in the filter. At the PC, type the command
Note: This procedure is also described in /home/pti/writeup/ftalign.txt
Make sure there are no strong lights on. Remove the narroband filter from the beam. Type "ftABCD" on the PC. Turn on the multimeter by turning its rotary switch two click CCW (to "off") and then two clicks CW (to "DC", which is labeled as a V with a solid and a dashed line above it. Turn on the WL source and adjust it so that the meter reads something like 2.5 V. This should require around 0.7 V on the WL source. (If the meter reading fluctuates erratically over a wide range then the WL source is probably too bright and the sensor is out of range.) Peak up the signal using the white fine adjustment knobs on the fiber positioning stage. The little bargraph display on the meter is convenient for this kind of tweaking.
Additional information is available in /home/pti/writeup/fiber.txt
Then turn the white-light source back on, put the corner cube in its position at the edge of the table, and type
NOTE: Because of the way the software is written, ftCalLow and ftCalHigh must be done in this order, and in pairs. So, for example, if you mess up the ftCalLow you must do the ftCalHigh before you can come back and do the ftCalLow again. Then you must repeat the ftCalHigh.
Now move the corner cube out of the beam. It's best to store it with the front (optical) surface facing down so it won't accumulate dust.
Note: This procedure is also described in /home/pti/writeup/ftalign.txt
Unblock the boresite laser and remove the ND filter from the beam. Open the "North" acquisition camera window and click "Laser Boresight". Do the same for the south. The laser spots should be visible on the two TV monitors to the left of harbinger's display. If there's a problem, click "Off" and "Laser Boresight" again. When the process is finished, block the laser.
CPU Rack behind sound curtain ________________________________ | | | | | Sequencer| Delay | Fringe | | Computer | Line | Tracker | | | Computer | Computer | | | | | | | | | | | | | | | | | | | | Acceler- | | | | ometer | | | | Oscill- | | | | iscope | |________________________________|
Set the white-light source to ~1.6 V. Remove the paper cover from the pair of corner cubes, and verify that the laser is blocked. Back at the horatio console, click on the "PriN" and "PriS" under "Star Tracker" to bring up a pair of control windows. Position the windows with the mouse so that both are visible. Click the "Chop Calib" button on each window, one right after the other (don't wait for one to finish before starting the second - the two chop calibrations should be done simultaneously; in practice, what is important is that they finish nearly simultaneously). When both calibrations finish, it's a good idea to repeat the chop calibration, though the reason for that is somewhat mysterious. Then put the paper cover back over the pair of corner cubes and turn off the white-light source.
You are now ready to begin observing. YIPEE!
To start observing, click "Off", and "Primary Observe". This should cause the system to start its automatic observing mode, in which it cycles through the stars in the starlist, performing all tasks without user intervention (until something goes wrong!) Do not close the Observing window until you have sent the list to the sequencer, or the GUi will crash.
If you wish to modify the starlist, click "Create star list" again. Modify the starlist in the obvious way. The only trick is that you must stop observing (click "Off" in the Primary Observe window) before sending the new starlist to the sequencer.
There are some simple scripts, called 10MS and 20MS, in /home/pti/bin. They modify configuration files in the appropriate way. Once one of these scripts has changed the configuration, it is necessary to activate the changes by rebooting the fringe tracker CPU (ft0), and resetting the stroke on the delay line. The reboot is done in the usual way: hibernate the system by checking the Hibernate box in the Sequencer window, then rlogin to ft0 with username delay, password lineline, and hit Ctrl-X. When the terminal beeps and tells you that the remote host has closed the connection, the CPU should be back up.
Note that the calibrations for the fringe tracker have to be done after the reboot. This means the CalLow,CalHigh sequence for the fringe tracker. You'll probably want to redo the FidSets for the delay lines. Put in the new values of the stroke amplitude and frequency (check the Primary Fringe Tracker window for the correct values), then do SendStroke,StrokeSet for the Long Active delay line.
Additional information is available in /home/pti/writeup/20ms.txt
For each hut, turn on the air conditioner and/or dehumidifier as appropriate, remove the PVC pipe extension and store it, cap the end of the pipe to cover the window, flip the shiny cloth down and velcro it in place. Put the covers back on the siderostats (with the little silver tubes fitting over the pins). On the south hut, put the south wall panel back in place to provide insulation.
The data acquisition system saves raw data to the /home/palomar/data directory, which is actually a symbolic link to another directory which is created by the beginNight script. cd to /export/data and make a subdirectory there whose name is the data (yyddd, where ddd is the U.T. date at the start of the observations). Use the 'cp -rp' command to copy the night's data there, and then delete the orginal data. Follow the instructions in /home/pti/writeups/cdBurning.txt to create two CDs, one for taking back to Caltech and one to stay at PTI.
Note that an on-call system has been set up in which experienced PTI observers are assigned to serve as telephone tech support people. The schedule, which tells you whom to call on a given night, is on the web. Please contact your on-call volunteer after you've exhausted the procedures in the writeups and before calling Mark Colavita.
Computer crashes: The most troublesome component of the PTI is probably the computer system. This consists of three cardcages in the observing room, at least one cardcage in each hut, and several PCs which serve as dumb terminals connected over serial lines. The cardcages can contain several CPU boards each. Typically, one CPU board drives one subsystem component. If you see the realtime clock on the control window for a particular subsystem component stop, the corresponding CPU is probably down.
If it becomes necessary to reboot the whole system, follow the directions in /home/pti/writeup/reboot.txt
It's usually the cardcages that cause the most trouble once things are working. The usual procedure is to reboot the offending CPU. Before doing this, it is necessary to 'hibernate' the system. As I (vaguely) understand it, this is necessary to prevent shared memory areas from being corrupted by the rebooting CPU. Additionally, the CPUs must be booted in sequence, and for any given attempt there is a significant chance of failure, so it might be necessary to try several times. There is a file in /home/pti/writeup called 'reboot.txt' or something similar which gives details. There's also a set of instructions posted at eye level on the wall to the right of the cardcages. Good luck.
Fringe Tracker Window Hang: If the system still seems mostly alive but the digital clock on the Fringe Tracker window has stopped, it probably means that one or more of the processes running on fit FT CPU has a problem. You have to use the PC terminal in the beam-combiner lab to get it going again. Type a ^C (ctrl-C) to get a prompt, then schedShow to see a list of the scheduled processes. If any of them are in "panic" status, re-enable it with schedTaskEnable "taskname". Here taskname will be something like "ftTask Med", for example. The computer should respond with messages like "... cal start". Repeat the schedShow command as needed to confirm that there are no more processes in panic state, and then type ftOff and ftTaskZip. Finally, do another, schedShow to confirm that everything is well. Note that this procedure may work with the other CPUs as well.
PAQ (Primary Acquisition) Errors: These fall into a general class of errors which are often caused by the presence of a large offset value. So the first step is to go to the siderostat control window and and check the "Show detailed status" box. The idea here is to clear out any offsets that are larger (in an absolute-value sense) than ~10,000. To do this, click "Set Offsets" to get a new panel on which a there are buttons used to clear the offending offsets. Since this panel disappears when you do this, you'll have to repeat the procedure once for each offset you wish to clear.
If a PAQ error persists, try hitting the "Hardware Reset" button, which becomes visible when the "Show details" box is checked.
If none of the above works, try repeating the Laser Boresite procedure, clicking "Use Laser Boresite", and then clearing the siderostat offsets again.
NOTE: Watch out for NaN values in the offsets. These are trouble, and should be cleared out quickly whenever they are seen.
Don't step on the big black beetles. They stink.