Change page style: 

Operating Phoenix

This page is partly based on the Phoenix Guide written by Tom Geballe with help from Bob Blum and Ken Hinkle. The Guide itself can be found in the Phoenix computer (/net/phoenix/home/aspp/OBSERVERS/guide/)

Troubleshooting procedures are partly based on the documentation written by Nigel Sharp for ABU/WildFire at Gemini South.

Start of the night procedures:

Starting from scratch: usually, the instrument will be left ready by the day crew before the first night of the run. However, if it is necessary to start from scratch, here are the instructions:

  • log in the Phoenix computer as aspp (contact someone on the staff for the password). This will bring up all the relevant windows. Instrument Control is in the top left, Instrument Status in the top right; an IRAF window is in the bottom left and the ximtool and a console window share the bottom right.
  • In the Instrument Control window type  (the UNIX prompt should look something like phoenix1 <2:23pm> {77}):
    • startwf
  • answer the questions carefully; this will start WildFire, the instrument control software. If everything works, the prompt should be now a '%'
  • From here on, just continue as for a normal start.
  • Do not type any command in the Instrument Status window unless instructed to do so (you don't want to upset WildFire).
  • See below for troubleshooting (instrument hangs, terminating an integration, not getting signal on the detectod, etc)

Normal start of night:

  • If WildFire is up and running and the instrument power is on, in the Instrument Control window type
    • activate
      • this will turn on the array electronics. "activate" will ask for the bias setting - enter 0.32. It would return a number which should be 0.300+/-0.002
    • setbias 0.32
      • to adjust the actual bias to 0.300
  • Check the instrument temperatures. The output will show in the Instrument Status window. The cold heads ("LN2") should read ~40K; the detector should read ~31K; and most of the dewar ~55K
    • status t
  • Check the position of the instrument elements (filter wheels, grating, etc).
    • status s
  • Create the data directories for the night. There are two data disks available (phx0 and phx1). At the beginning of the run, is a good idea to check the disk space.
    • mkdir /phx0/yyyymmmdd
      • where yyyymmmdd corresponds to the UT date for the end of the night, using the standard three letter code for the month
  • Set the data directory.
    • setdir /phx0/yyyymmmdd
  • Once the TCS and the VII are up, connect the instrument to the telescope. THIS IS VERY IMPORTANT - otherwise, you get mangled headers and no offset control!
    • tcpon
  • Finally, if you want to change the parameters set, use:
    • ped
  • On the IRAF display, frequency increases/wavelength decreases upwards (larger pixel values), and the slit is accurately aligned E-W, although it appears tilted on the array when seen in image mode.

Phoenix TV "guider": quite useful to see if you get the object in the right place

  • Turn ICCD all the way counter-clockwise
  • Turn on MAIN POWER
  • Turn on POWER (switch above the ICCD pot)
  • Push the INTEN ON button
  • With the current configuration (default instrument angle = 180deg and Phoenix on port 5), N is up and E is to the right on the TV
    • turn the ICCD down when slewing to avoid damage from bright stars
    • use the neutral density filter (ND on) when observing bright targets
    • turn the guider off at the end of the night
  • We are now using the OTR to select the observations, and all header updates for classical programmes are done after the fact. There is no e-obslog for Phoenix, classical observers can keep their own log or use the paper templates available in the Phoenix folder above the astronomer's workstation.

    Use of QMon by classical observers is DISCOURAGED - blank headers are easier to update.

    Typical overheads:

    Readout single image ~25sec
    Readout sequence of 10 images ~17sec
    Each lnrs 1 sec

    Focus and guiding:

    Focus: if using the J9232 filter, remember to ask the SSA to use the Phoenix_J setup (works for both WFS). Otherwise, use Phoenix_P1 or Phoenix_P2 as appropriate.

    Guiding: the telescope guiding system will compensate for differential refraction between the guiding wavelength and the instrument wavelength. Central wavelength and CassRot position angle are now read directly from the OT into the TCS. Just check as usual to see if they are correct.

    Note that there is significant flexure/"flopsure" in the Phoenix imaging mode, which results in the image of the slit moving on the array. This effect has been measured to be 3.5pix perpendicular to the slit (up and down the array) between elevations of 85 and 30deg (~1pix/hour), suggesting that recentering at least once per hour may be advisable during a long observation. The flexure parallel to the slit is about 7pix between elevation 85 and 30deg.

    Observing scripts:

    Useful numbers:

    • Saturation on the array: 8000 counts
    • Plate scale 
      • imaging mode: 0.055 "/pixel
      • spectroscopic mode: 0.085 "/pixel
    • Spectral coverage: ~1500 km/s or 0.005*(central wavelength or frequency)
    • Highest advisable exposure is 3000 counts.
    • At J, H, and K limit exposures to 1800sec. 
    • At M limit exposures to 30sec, and use coadds (4 to 8, so the final exposure time is a compromise between accumulating signal and quality of the sky subtraction)
    • For long exposures (more than 10min) set the number of low noise reads to 8 (using the lnrs command or as one of the ask options). 

    Generic observing caveats:

    • There is NO CLEAN WAY to stop an exposure. Therefore if you set a long exposure time, be VERY sure that the instrument configuration is correct. Be VERY sure you have not set a long exposure/many coadds if that is not what you want.
    • If by any reason you need to terminate a long integration, refer to the troubleshooting procedures below before hiting the dreaded CTRL-C.
    • There are three different positions for 2- and 3-pix slits, depending on the wavelength of the observations. Check here for the list or type help slit in the WildFire window. Note that the flats must be taken with the correct slit in position!
    • The minimum exposure time is 1sec. Use the ND filter when imaging bright stars to avoid saturating the array. Heavy saturation will leave an afterimage that takes quite a long time to disappear (there has been much debate where taking a series of short darks helps with the persistence, but for all that is worth it does seem that it goes away with time, not with exposures).

    IMPORTANT NOTE ABOUT OFFSETS: the offsets defined in the OT are NOT passed to the telescope. Dither patterns other than the default observing sequences below must be manually executed FROM THE PHOENIX COMPUTER (if you ask the SSA to apply them, there will be NO indication in the image header of the updated telescope position). Offsets are also relative to the last position, and done in N-E orientation (not p and q as for the facility instruments).

    The following scripts are in /home/aspp/OBSERVERS/blum/

    The source_phx.tcl script (usage: source_phx) will source all the scripts. It has to be sourced first (usage: source_phx) will source all the scripts. It has to be sourced first (usage: source /home/aspp/OBSERVERS/blum/source_phx.tcl).

    The individual scripts can also be sourced independently. If you add new scripts, put them into source_phx.tcl so when starting/restarting wildfire, a minimum of work is needed to source the observing scripts.

    • gcal.tcl
        usage: gcal t_exp(sec) nflats ndarks
        default values: 3 1 0 (a single 3 sec flat, no darks)

        Generic script for taking flats and/or darks. It assumes that the filter, Lyot stop, slit, and grating are already set. The user is queried to start, then it pauses after taking flats before taking the darks. See below for the generic flats/darks setup and procedure.

        • If uncertain about exposure time, try gcal 1. This will take a single 1 second flat exposure and display it on the Ximtool window. Then change t_exp so that individual flats have ~3000 counts.
        • If darks are taken, the Lyot stop is left in the dark position and must be repositioned before continuing with the observations.
        • Very important: TURN OFF the GCAL QH lamp as soon as the flats are finished. The average life for those lamps is about 40hrs and, although the spares are not hard to come by, changing the bulb in the middle of the night will result in substantial time lost! 
    • zcenter.tcl
        usage: zcenter x y angle scale
        zcenter x y angle
        zcenter x y
        default values: 128 504 90 0.055 (centre of array, slit EW, image scale)

        Set the (x,y) pixel values, slit PA and pixel scale for the starting position (slit center). The script also set these as global variables to be used by recenter, abba, and slitscan. It must be run at least once (any time the slit changes or after bringing up WildFire) so the other scripts can access those variables. 

        Note that this script MUST be run if the slit is not oriented E-W; otherwise the recenter and abba scripts will not offset properly.

    • recenter.tcl
        usage: recenter x y

        Given the x,y coordinates of an object in image mode, will center the object on the slit starting position defined by zcenter. It may be necessary to execute the command multiple times to get the object in position within 0.5pix. The script queries whether to disable the WFS during offsets (needed for moves larger than 5arcsec).

    • abba.tcl
        usage: abba ampl n

        Execute n a-b-b-a sequences along the slit, with a separation ampl in arcsec between ab positions (e.g. abba 5 2 will to 2 abba sequences offsetting +2.5 and -2.5arcsec from the starting position - usually the centre of the slit). If requested, will disable guiding during offsets.

    • abbaperp.tcl
        usage: abbaprep ampl n

        Same as abba, but beam b is off the slit in a perpendicular direction by ampl arcsec.  

    • slitscan.tcl
        usage: slitscan ampl m n

        Do m positions along the slit separated by ampl arcsec. For example, slitscan 4 3 2 will execute 2 sequences of 3 exposures separated by 4 arcsec along the slit (-4, 0, +4), then return to the starting position.  

    • immode.tcl
        usage: immode

        Set the slit to open and the viewer to image mode.  

    • spmode.tcl
        usage: spmode

        Set the slit to 8 (4-pix wide) and the viewer to open.  

    • mgo.tcl
        usage: mgo n

        Allows n images, without specifying "pics"

    • Useful observing setup commands:

      cover open: light path to the Science Fold mirror,
        close: light path to the Phoenix internal hollow cathod lamp
      viewer dark, open (grating), image, lyot
      slit open: for imaging
        an integer number to select the corresponding slit
      help filt: lists filters in filter wheel
        cvf: lists filters in CVF wheel (there are no CVF filters)
        slit: lists slits in slit wheel
        grat: explans the grating parameters
        lyot: lists the various lyot stops
      filt an integer number to select the corresponding filter in the filter wheel
        open to select the open position in the filter wheel (used with a filter in CVF)
      cvf an integer number to select the corresponding filter in the CVF wheel (note that although cvf 7 is listed as a blank, it actually contains the L2870 filter)
        nd to select the neutral density filter in the CVF whell (attenuates by 100x)
        open to select the open position in CVF wheel (used with a filter in filter wheel)
      grat (number) (units) (offset): sets the grating position; see below for an explanation
      settime a real number to set the exposure time
      eask edits the list of parameters that are queried by ask before an observation;
      la = list/ask; l = lists only
      ask enter the new values for the observation parameters
      toffset X Y; moves the telescope X arcsec E, Y arcsec N
      north X ; moves the telescope X arcsec north (or south, east, west)
      increases the voltage in the motor to coach a "sticky" mechanism to move. USE WITH CARE! If repeated issues of kick do not work, it may indicate a more serious problem. kick should NOT be used as "normal operational procedure" (i.e., whole night long).

      Observing commands:

      set the fiducial position of the slit. Required at least once at the start of the night or every time the PA of the slit is changed. Reccomended when changing filters (particularly to the L and M bands)
      given the X,Y coordinates of a target, will offset it to the fiducial slit position as defined by zcenter.
      go start the integration with the current setup (set by ask
      observe equivalent to an ask followed by a go with an option to abort the operation after the ask. 

      GCAL flats:

      Flats are taken with the facility calibration unit GCAL and the following (approximate) configurations:
      Note: use these values as indicative only.

      Filter Lamp Diffuser ND filter Exptime
       QH  optical  open 10sec
      L3100  QH  optical  open  7sec
      J7799  QH  optical  ND1.0  2sec
      K4667  QH  optical  ND1.0  1sec
      K4578  QH  optical  ND1.0  1sec
      H6420  QH  optical  ND1.0  1sec

      Note that depending on the grating setting, it may be necessary to increase the exposure time/remove the ND filter (e.g, if too close to the edge of the filter).  If in doubt, start with an 1 second exposure, ND=1.0mag, and work your way towards ~2000-2500 counts average.

      To bring up the GCAl window, select it from the Gemini menu in the GuestA account. Click the Mechanism Control Commands button and select the appropriate setup. Don't forget to hit APPLY. It may be necessary to datum all mechanisms at the start of the night.

      To reach S/N ~400 you need 15 images with the above level  (2.0-2.5k).

      Very important: turn off the QH lamp as soon as the flats are done. The average life for those lamps is about 40hrs and, although the spares are not hard to come by, changing the bulb in the middle of the night will imply in substantial time lost! 

      The GCAL spare bulbs are in the Instrument Lab, in a cabinet located against the wall to the right of the door, at the back of the room. Remember to use the cloth gloves to manipulate the bulbs. To change any of the lamps, the telescope has to be taken to the zenith and the GCAL bottom cover unscrewed. The lamp positions are clearly identified (here is a picture). 

      Telluric standards:

      The rule here is simple: the brightest, the better. My usual procedure has been to obtain the telluric standard after the target, matching in airmass, and using the list of bright OBA+early F stars provided at the NOAO Phoenix webpage. Select the appropriate sidereal time bin and choose a star that is off by ~0.1 airmass (higher or lower depending if the star is rising or setting) from the matching value (that will compensate for the acquisition overhead).  There is an OTR library (GS-LIB-PHXCAL) that contains the subset of stars in that list that can be used (a few were doubles or had no good guide star). Select the appropriate star, copy the observation to the programme, then copy and paste the instrument configuration from the science target. Please do not queue directly from the LIB programme  (this will leave an instrument configuration component there that the next observer will have to remove).

      Some programmes may require a KIII spectrum for wavelength calibration (some spectral regions do not have enough atmosphere absorption lines to provide the calibration). A list of bright KIII stars is available at the telescope (in the Phoenix - a green paper - folder, in the shelf above chamaeleon).

      Acquiring the target in the thermal IR:

      The usual procedure is to position the star in the slit using the same filter as used for taking spectra.  However, the Phoenix filters have different indices of refraction and are individually wedged to avoid fringing, so the telescope must be positioned slightly differently for different filters in order to get starlight through the slit. Another problem is that the slit position does not reproduce accurately.  In the thermal IR, one often cannot see the target on a Phoenix image, because the background is too high, so the best procedure appears to be

      • put the target in the slit using a short wavelength filter (usually K4308)
      • change to the thermal IR filter and offset the telescope.

      The offsets determined so far are (for cass rotator at 90deg; i.e., EW slit):

      K4308 --> M2150:  offset -0.10", -3.80"  (0.10 W, 3.80 S)
      K4308 --> M2030:  offset -1.25", +3.40"  (1.25 W, 3.40 N)
      K4308 --> L3290:  offset   +0.9", -3.20"  (0.9  E, 3.20 S)
      K4308 --> L2734:  offset -0.35", +2.78" (0.35 W, 2.78 N)

      For arbitrary rotation angles, the offsets are

             dra  =  x cos(theta-90) + y sin(theta-90)
             ddec = -x sin(theta-90) + y cos(theta-90)

      where theta is the Slit PA, and the x's and y's are as given above (i.e., -1.25 and +3.40 for K4308-->M2030)

      To detect the object in the thermal IR image, one must use the neutral density filter (nd) in the cvf wheel, and one usually must take pairs of offset images and difference them. If the object is brigth enough in the optical, it is useful to mark the position on the TV monitor at which the starlight goes through the slit using a grease pencil.  Remember that positions will be different for different filters. The observer can then move the telescope so that the (visible) object is at the marked location, and if desired touch up the position in the slit by examining pairs of subtracted images.

      Another procedure, when the target is bright enough in L and M, is to use the Lyot central mask to cut the background. Set the lyot to position lyot16, configure the instrument to image the slit, measure the slit centre at column 128, then issue the zcenter command to define the fiducial position for that filter. After that, set the slit to open to directly image the field and use the  recenter command to obtain the offsets to move the star into the slit position. Don't forget to set the viewer and lyot to the correct position once ready to start taking the spectra.

      Image field and slit orientations:

      The default Cass rotator position (shown in the TSD as IPA) is 90deg. At this position and Phoenix on port 4, the field orientations are as follows:

      TV monitor if using the old monitor flipped on its side, N is up and E to the right
      if using the new (white TV set) one, N is to the right and E is down
      Phoenix image N is up and E to the left
      Phoenix spectrum slit is E-W, E to the right
      Wavelength increases towards smaller pixels

      The IPA (field orientation) is now set when the observation is selected from the queue. If other than 90degrees, don't forget to run zcenter with the appropriate parameters, otherwise the observing scripts will not work.

      Grating setup procedure:

      To set the grating positions, use the grat command in WildFire. The following syntaxes are equivalent:

      grat 2154.5 K 2000 drives the grating to 2154.5 Kaysers (cm-1) with an offset of 2000 encoder units

      same as above, but in microns

      grat -14567 drives the grating to the encoder number (no offset given)

      The encoder values are relative, and therefore may change between runs if the instrument is warmed up or if the mechanism is datumed by any other reason. However, using encoder units allow for a much more precise repositioning of the grating than the other two. Also, if the mechanism is datumed the difference between old and new encoder values should be a constant: find it for one setting and just add to the remaining.

      2004Dec19: Bob Blum wrote a script which uses a fit to the encoder/wavenumber values obtained for the last three semesters of queue and that seems to work fairly well. It lives in the usual place in the Phoenix computer and can be run as

      enc XXXX.X

      where XXXX.X is the desired central wavenumber. The script will return the corresponding encoder number. There may be an offset to be added, which has to be checked on the sky, but will be valid to all subsequent grating positions.

      Obtaining the grating setting for an observation can be tedious and is time consuming. Best strategy seems to be do it for as many settings as possible in the first night of a run or if the weather is too bad for anything else. 

      • Acquire a bright star, center it on the slit as usual.
      • Set the grating to an approximate position using any of the first two forms of the grat command above. The offset value may change from run to run, but it is usually around +1400;nbsp; to +1800.
      • Obtain one spectrum, extract (see below) and plot. Frequency increases/wavelength decreases towards higher pixel values. Now compare your spectrum with the appropriate section in the Infrared Arcturus Atlas to determine the approximate central wavelength. The atlas contains both a K star spectrum and a spectrum of the transmission of the Earth's atmosphere (which is what is detected by looking at an A star).
      • Type status s. The current encoder position will be listed on the Instrument Status window. 
      • Determine how many microns/wavenumbers you want to shift to get the correct wavelength centered on the array. The full spectral range corresponds to about 2000 encoder units. Increasing encoder units drives the grating to higher frequencies.
      • Don't forget to write down the final encoder value!
      • The grating can never be moved back to precisely the same spot. Always take flats before moving to a new grating position!

      Quick look data reduction (in the Phoenix computer):

      • Start IRAF on the IRAF window that comes up automatically when login in.
      • Load twod and apextract
      • If the target is bright enough compared to the background, a single file may be used. In this case skip irdiff and go directly to apall. Otherwise you need a pair of offset observations.
      • epar irdiff
        • exten = 3
        • prefix = the root filename (e.g., if the files are data123, data124 and so on, prefix=data)
        • pic = the name of the difference file to be created. The default (junk9999) is good enough.
      • irdiff p1 p2, where p1 and p2 are the beginning and ending image index. 
      • appal junk9999  - all questions probably are answerable as the default. This will create the 1D file ( with the sky subtracted positive spectrum in junk9999.
      • splot to examine the resulting spectrum.

      Quick look data reduction (from the Guest account):

      • Start IRAF as described in the Observer Setup page.
      • display the first spectrum for the target and measure the approximate background value (just put the cursor somewhere away from the spectrum and/or atmospheric lines)
      • epar sntest2
      • the parameters are the root filename (eg, 2004dec19), the file numbers for a pair of spectra (eg, 181 and 182), the path for the files (the easiest way is to cd to the night's directory in the scratch disk), and the background level measured above. The size of the aperture for extraction can also be changed if needed.
      • the script will subtract the spectra one for the other, displaying the result (fl_display=yes); then extract the spectra interactively or not (fl_inter=yes). It calculates the signal by extracting the mean counts in the spectrum (it gets the mean and the stddev, then recalculates the mean excluding values at more than 3sigma - this eliminates the atmospheric lines); and gets the noise from the variance plane of the extracted spectrum (which is why it needs the background added back). The S/N computation uses the correct values for coadds, gain and read noise and tends to be quite accurate as long as there is a decent continuum to start with.
      • So, a typical command line would be:
        • sntest2 2004dec19 181 182 bglev=5. fl_inter-
      • A quick look script to evaluate the signal on flats is also available. Given a single flat, the script will return the mean counts and estimate the number of flats required to reach S/N=400. It will also warn if the counts are too low or too high. Just type
        • snflat 2004dec19_0103

      End of the night procedures:


      • To do all at once. Very important! Please don't forget
        • end_of_night
      • This script will:
        • Turn off the array electronics.
          • deactivate
        • Protect the dewar window from dust and contamination
          • cover closed
        • Physically shield the detector from light. A deactivated array remains sensitive to light (the resulting image is negative).
          • viewer dark


        • Toggle off "Monitor" in mmon
        • There is no need to exit the OT during a run (or to exit the Guest session), but I usually minimize all the windows to avoid any extraneous typing during the day.


        • Double check if all lamps are off, and the shutter closed.

        Troubleshooting: What to do if...

          The following procedures are the most commonly needed to restore WildFire after different  levels of system failure. Re-booting the Phoenix computer and/or cycling the instrument or DSP box in the computer room are not usual WildFire procedures and should be done only if the symptoms described below are satisfied. The procedures are listed in order of increasing severity, so unless a specific condition has occurred (e.g., power has been off in the DSP box), try the less severe first.

          There are four main levels of troubleshooting (some minor recovery procedures are listed here):

          First, identify the appropriate hardware elements (thanks Gelys for checking those out!):

          • The Instrument Control Computer, 'phoenix1' in Gemini South, is the workstation located at the centre of the summit console. The CPU itself is in the computer room (the double door at the back of the Console room) third rack from the left.
          • The DSP box is a black box in the computer room, just below the Phoenix1 CPU, third rack from the left. It is labeled "HEURIKON" and the power switch is near the name. In normal operations, there will be one green power and one (out of four) green status lights on. There are also two lights marked 'RCVR'. Both will be red if the instrument is powered off (no connection). If the instrument is working and one  or both lights are red, there is a problem with the optical fibers connecting to the instrument.
          • Instrument Power: The Phoenix electronics in Gemini South is with the instrument, in one of the side ports of the ISS,  under the telescope. There is NO remote power switch down at the Console room.

          Now, identify your problem:

          1. If the array warms up because the cold heads stopped (whatever the reason), and the problem does not seem to be fixable soon, the procedure is simply to deactivate the array (use the deactivate command), and leave WildFire running. You will not be able to take any useful data, but normal operation should resume after the system cools back down. Do not cycle the power on the DSP box or the instrument because of cooling glitches.
          2. If the WildFire prompt ("%") does not come back after issuing a command (no prompt appears after repeated pressing of the <return> key), read the WildFire control not responding section below.
          3. If any part of the WildFire system has crashed, but the prompt is still available in the Instrument Control window (either the % or the 'phoenix1' UNIX prompt), go to the warm restart procedure.
          4. If there is some problem with the data taking, but you know for sure that the power has not gone off to either the instrument or the DSP box, use the warm restart procedure. Some of the situations can be: the Instrument Computer has been rebooted in a civilized way (for upgrades, relocation, network changes); any of the WildFire ancillary or helper processes (e.g., the Instrument Status window) have died; WildFire itself has crashed or had been shutdown using the exit command.
          5. If the warm restart procedure does not work, try a cold restart.
          6. If the instrument power has been turned off, intentionally or not, follow the cold restart procedure. Note that is very much advisable to exit from WildFire beforehand if a power outage is scheduled.
          7. If the cold restart procedure does not work, try the deep freeze procedure.
          8. If the DSP box power has been off, a complete rebbot is required. Follow the deep freeze restart procedure. In some cases, more than one sequence might be required to solve the problem. If you run into such a case, please contact someone from the support personnel before flipping the power switches a dozen times. 

          WildFire control not responding:

          Again, the symptom for this one is that you do not have a prompt at the Instrument Control window after hitting <return> a couple of times. 

          • First, check if you are not in the middle of a long integration. Just scroll up the window and look for a go or observe command or any of the observing scripts listed above.
          • If you are in the middle of a sequence, the telescope offsets normally, then the observing script seems to hung after issuing a "Turning guide off" message AND the guide loops never closed again, you have found a TclTK/VII intermitent bug. First, be sure that BOTH the conditions above are satisfied. Second, ask the SSA to close the guide loops again. Once you are guiding, hit Ctrl-C ONCE (and only once!) in the WF window. You should see a message "Turning guide on" and the sequence should continue normally. Be VERY careful with this problem - hitting Ctrl-C in WildFire will crash the session and you will have to go for a cold restart procedure.
          • If neither is the case, check if the WildFire 'saver' process is running: the software is smart enough not to allow you to start one integration until the data from the previous one has been saved. Go to any UNIX window and type 'ps ax | grep save'. If it shows a 'saver' process, the problem is somewhere else, see below. If the process is not there, go to the Instrument Status window and type 'saver &' to restart the process and adopt that window for the output. If the saver process died during an observation, and as a consequence WildFire hung up, restarting the saver will cause the image to be safely read out and the control should return. Just go on observing.
          • If the saver is running, you will need a cold restart. Type CTRL-C once in the Instrument Control window (which at the moment will have no prompt), then go to a UNIX window and type 'hung' or 'hung2'. If everything works, the control process will crash after a few seconds, isuing a few error messages and leaving a clean UNIX prompt. Now, turn the instrument power off and go to the cold restart procedure.

          WildFire warm restart:

          Use if WildFire has been shutdown because of an exit command, a crash or a computer reboot, but there has been no power outages.

          • Check if all ancillary windows are gone (Instrument Status) and that the UNIX prompt is present and responding in the Instrument Control window.
          • At the UNIX prompt, type 'g phoenix'
          • Read the messages displayed to check if the instrument recovers cleanly. 
          • Restart WildFire with the 'startwf' command, answer the questions, check the observing setup and go on observing.
          • If it instrument startup fails, hangs, complains or does not work, go for the cold restart procedure

          WildFire cold restart:

          Use if the instrument power has been turned off. This could have been intentional, due to a power outage, or as part of the recovery procedure for a WildFire hung.

          • Make sure you can turn the instrument power back on.
          • Check if you have a functional UNIX prompt in the Instrument Control window (type <return> a couple of times)
          • Now type 'startwf' and answer the questions carefully. The DSP box has not been off, but the Instrument power does, so answer 'y' to the second one. 
          • Turn on the instrument power, answer the remaining questions and check if everything comes up correctly. 
          • If it does, check your observing setup and go on taking data. Otherwise,  proceed to the next level.

          WildFire deep freeze  restart:

          Use this if the DSP box power had been off or if the previous levels of troubleshooting did not work.

          • Turn off the instrument power (if it is not off already)
          • Turn off the DSP box (if it is still on) and wait for the electronics to discharge. At the first attempt, let is sit for a minute. If you have to repeat the procedure, wait longer.
          • Turn the DSP box back on.
          • Now reboot the Phoenix computer: press the Stop and a keys at the same time, then type b at the prompt. Leave the instrument power off until the computer is back on line.
          • Once the computer is back up, log in (username: aspp, ask some of the support staff for the  password), run 'startwf' from the Instrument Control window, answering yes to all questions and turning the instrument power back on when requested.
          • If the deep freeze restart does not work on the first try, repeat the whole procedure, leaving the DSP box off for more time. You may have to do it more than twice, but if by the third or fourth time the problem did not show any  sign of improvement, please consider contacting someone of the support staff.