Quality Assurance Programme

Quality Assurance Programme

UKIRT image quality, thermal stability etc. are monitored and maintained by a regular, scheduled series of “quality assurance” nights, with approximately one such night per month. This page details the measurements undertaken on these nights. 

Web Copy of this Page

This page is https://about.ifa.hawaii.edu/ukirt/telescope-information/quality-assurance-programme/

You should refer to the web copy rather than a printed version (which can’t be guaranteed to be up to date).


A WFCAM QA night is generally split over two nights and includes the following:

  • Top end Tilt (provisional)
  • Wavefront Sensing
  • Skydip for emissivity
  • Pointing test (after implementation of the wfs results)

Top End Tilt

(This is provisional): Run the two top-end tilt tests and adjust/save the parameters. The instructions are here:

WFCAM Secondary (M2) tilt alignment

Wavefront Sensing

The WFS procedure for WFCAM is described fully in the following page:

Wavefront sensing with WFCAM

You should run this in the middle of the night; it takes about 3-4 hours. 

Skydip for emissivity

Execute  the MSB entitled SKYDIP TO SOUTH (which is in the ukirtcal menus). Should be done only if photometric and only in the second half of the night, with the moon not up. Make sure the dome is in position before starting the first Observation in the MSB. 

Pointing Test

The TSS should perform a pointing test on the night after the WFS has been done. This takes about 45 minutes with WFCAM. All you need to do is run the “long_ptest” command in the TCS. See below for details.   Follow the instructions in:

 Wiki page about carrying out the Pointing test observations or
 QA/engineering night procedures in the UKIRT TSS guide.

Please send notification that a pointing test has been done to Tom Kerr. 


A Cassegrain QA night includes the following:

Wavefront sensing 

Refer to the WFS web page for details of running this instrument. The full WFS test takes four hours. Do not do this in the first few hours; wait until things have settled and cooled down. The WFS system and observing protocol are documented step-by-step on:

Conducting the Wavefront Sensing Observations (Cassegrain)

The WFS data reduction protocol are documented step-by-step on:

Reducing the Wavefront Sensing Observations

For instructions on doing WFS with WFCAM, please refer to the staff wiki.

Pointing test

The TSS should perform a pointing test. This takes about one hour (60 CMC stars should be observed with an exposure time of one minute).  Follow the instructions in:

 Wiki page about carrying out the Pointing test observations or
 QA/engineering night procedures in the UKIRT TSS guide.

Please send notification that a pointing test has been done to Tom Kerr. 

Port and Tilt (CGS4)

Send the (P=0, K, 2) Observation from the CGS4 aperture/PA measurement MSB from the queue. then adjust port through a range of values, peaking up at each, and set to the one which gives the best signal. Then run through a similar range of tilt and peakup at each, and find the best signal. Note that you can keep manpeak running while changing port/tilt positions, no need to stop it every time.

Port and Tilt (UFTI and UIST)

NB: Only needs doing when secondary swapped back after a WFCAM run, or when the instrument has been off the telescope.

The MSBs you will need are in U/EC/1. Titles as follows:

InstrumentMSB name in U/EC/1
UFTIUFTI port & tilt
UISTUIST port and tilt

Most of these have a target component set to RA, Dec 0,0. You will need to disable observability to get them to come up. Skip the slew and the TSS will go to an appropriate star. The following notes describe the principles of the measurement.

With dome lights on and mirror covers closed, exposures are taken in the J filter; the camera is aligned when the signal is at a minimum as then the camera is looking directly at the covers and not getting light reflected off the dome.

Turn on all lights (top and bottom levels); make sure that if people are around they will not affect the light path to the camera; remove the window cover from the camera.

Run up telescope with tel_dev (not _sim), datum dichroic and move dichroic to appropriate port for camera (west for UFTI) by “pstart”, and e.g. “port west”. Keep mirror covers *closed*.

For either UFTI or UIST, do the usual array tests via the QT/queue/sequence console. Then fetch and send the UFTI or UIST port and tilt MSB from U/EC/1. Take measurements at the listed port values, for a tilt of zero:

   tilt: 0   0   0    0 0    0    0 0    0 0    0 0    0
port: 0 -50 +50 -100 0 -150 -200 0 -250 0 -300 0 -400

The routine is: take an exposure with dichroic at 0,0; move dichroic, when dichroic is settled, take another exposure. There is some hysteresis so move the dichroic from positive to negative values, or always go through (0,0) (no need to actually take repeat data at 0,0, just set the dichroic there). The command is e.g. for moving port, “poffset 0 -50”. Set to minimum signal value e.g (0,-190), then cycle through tilt, e.g.:

   tilt:    0  -20  +20    0  +40    0 +100  -80  +70  -60 +100
port: -190 -190 -190 -190 -190 -190 -190 -190 -190 -190 -190

set to minimum signal value e.g (+15,-190), take another frame to check for low signal.


You can either analyse the data in the raw or reduced data directory, via this sort of kappa command:

    for UFTI data: "stats clip=3 f20001216_00014.i1"  for /raw or
"stats clip=3 f20001216_00014" for /reduced

or use a Gaia ARD region and look at the mean.

Plot the mean value against port or tilt offset and get a minimum by linear extrapolation from the outer points to center, or by fitting a parabola. The data will look something like:




x x x


When you have got a final value for port and tilt, record them in:  http://www.ukirt.hawaii.edu/cgi-bin/wiki.pl/UKIRT_QA_night_Logs

Focus offset measurements

Please log the results of focus measurements in http://www.ukirt.hawaii.edu/cgi-bin/wiki.pl/UKIRT_QA_night_Logs. Send a note to Tom and Russell at the end of the night pointing to this location.  See http://www.ukirt.hawaii.edu/telescope/QAnight/FF.html for the current and past focus values.

MSB Names and Locations

The MSBs you will need for these measurements are in U/EC/1. Titles are:

InstrumentMSB name in U/EC/1
CGS4CGS4 fine focus K band 40l/mm
CGS4CGS4 fine focus K band echelle
UFTIUFTI fine focus
UISTFOCUS: Imaging focii measurements; 3 cameras
UISTFOCUS: IJ/JH/HK Spec focus measurement

Most of these have a target component set to RA, Dec 0,0. You will need to disable observability to get them to come up. Skip the slew and the TSS will go to an appropriate star. The following sections describe the details for each instrument/mode.


The procedure is to measure the image FWHM at about 8 or 9 points, 0.2mm apart in Fine Focus setting, centred on the current default setting of Autofocus Fine Focus for the instrument. You should jump between positive and negative offsets in building up the coverage. Autofocus is run throughout, and allowed to settle after each change of focus setting.

Find a faintish star near the zenith (11-13th mag). Fetch/send the MSB titled “UFTI Fine Focus” and then slew to the selected star. Run the programme through to “Observe”, autofocus and wait until it has settled and then take the Observe and stop asap. Use View…Pick Object (zoom it out until the zoom scale is x3) to measure the FWHM, either on the quck look gaia or on the DR side (the DR uses QUICK_LOOK anyway). Then adjust the focus to the next setting and go on to the next observe. Record the results on the form given in a later section. (The FWHM gets a bit weird when the images are well out of focus.) Strehl would probably be OK and ought to be less vulnerable to thin cloud. The measurement range of fine focus for UFTI should be from 0.6 or 0.8mm to 2.2 or 2.4mm, in steps of 0.2 mm, and you should bounce from side to side of this. 


Same MO as for UFTI. Generally a star which doesn’t saturate UFTI will be very comfortable with UIST.

Please check: 
1. Imaging 0.12″ 
2. Imaging 0.06″ JHK 
3. Imaging 0.06″ LM 
4. IFU 
5. (Pol in Imaging 0.12″) 

General note: The 0.12″ imaging focus and the spec focus should always be the same, so generally its only necessary to do the former.

Imaging mode

Same as UFTI:
1. UIST Movie; Measurement range should be between -1.5 and +2.5.
2. Use 9th mag CMC and the MSB uses ~5 x 3sec exposures in H-band
3. For LM, you’ll have to cd into ORAC_DATA_OUT, run kappa, and sub adjacent frames (the sequence does a sky-sub pair).

Note: It is measurements taken away from the in-focus ff setting that ultimately constrain the ff value (even though these can be a bit blobby), since changes in ff near focus have little effect on the stellar psf.

Spectroscopy mode (if you have time to do it)

1Choose and point to an A or F-type star, near the zenith around 8th magnitude. This allows you to use an exposure time of around 20 seconds. 
2Set up an observation that uses an E-W slit and the HK grism with 20 sec exposures. Probably best to use the 4-pix slit, since this is wide enough (hopefully) to minimise the effects of any minor seeing changes. The sequence in the programme doesn’t matter since you will only be running movie. (Could just use the standard star observation as it is in the template library.) 
3Load and run the observation, slewing to the standard star (or a near-cmc). Run the sequence to the imaging acquisition pause and start Movie.
4Set the Fast Guider mode to autofocus. The TSS now uses XOFFSET to move the guide telescope and put the four images produced by the Autofocus lens array in roughly the right place on the CCD for autofocus. The shift is roughly (9, -3) arcsec (is this still necessary??). 
5Start Autofocus (this also does fast guiding). Centre/acquire the star for spectroscopy on the correct pixel in the usual way (with uist_pick?). Once the star is acquired, continue the sequence to the next break (just before you’d normally start observing). However, now run movie again so that the Movie-Gaia displays the star’s continuum. 
6On the Bottom End Controller screen set Fine Focus to -1.5. Allow autofocus to converge again after the change (this may take a little while after a big shift). [Note – the location is this: Bottom Eng Engineering screen -> other displays -> bottom end (1st column in this window).]
7When happy with the autofocus convergence, record the “signal” from the stellar continuum. To do this we suggest using “X-Y profiles” in the Movie-Gaia window. Drag out the box across the entire spectrum and watch the signal over several cycles of the display; write down the MAXIMUM VALUE you see.
8Change “Fine Focus” to +2.5 (yes – the opposite end of ff travel) and repeat steps 6 and 7. Note that re-acquisition of the star shouldn’t be necessary after each FF change (since the star should still be down the slit).
9Take measurements at alternate +ve and -ve offsets about the nominal (which is around 0.8mm) to build up the coverage (alternate in this way to avoid any possibility of hysteresis). At each ff value, record the peak signal strength and enter this and the ff value into the excel spread sheet on the UKIRT PC. Measurements about 0.25mm apart should be ok – NOTE: ITS THE ff SETTINGS FARTHEST FROM FOCUS THAT LEAD TO THE BEST MEASUREMENT, NOT THOSE NEAR THE FOCUS WHERE CHANGES IN THE ff VALUE HAVE LITTLE EFFECT ON THE SIGNAL.

Having plotted up the data, please then report the FF value at peak signal into the wiki and email notification to Russell and ukss.

IFU mode

Essentially repeat the steps for a normal spectroscopy focus run. However, when faced with the multiple spectra in Movie, drag out the XY-profiles box over all the bright/central spectra. Although moving away from focus spreads light into adjacent spectra, it also spreads light along the slit axis, so ALL the spectra in the IFU movie will drop in strength.

Finally, note that current ff values for all UIST modes are listed on the UIST engineering web pages.


The CGS4 protocol is as follows: 

1Choose and point to a star with K around 10, near the zenith. Use one of the brighter faint standards, for example. This allows you to use 10-20 second exposure times.
2In the CGS4 sms screen (TSS side), set CGS4 to 2.1 um, 1st order, 2×1 sampling (say; it’s irrelevant), NDSTARE. Note that at 2.1 microns you won’t include the thermal end of the K window, which can mask signal changes due to focus. The slit width can be 1 or 2 depending on seeing: Tom Kerr recommends 2 pixels since it doesn’t make much apparent difference.
3Set the Fast Guider mode to autofocus.
Start Autofocus (this also does fast guiding). Peak up on a suitable CGS4 row using “Peakup” and 10-20 second exposures. Record the mean of a series of peak counts.
5On the Bottom End Controller screen set Fine Focus to 1.0 below the nominal. Allow autofocus to converge again after the change. This may take a little while after the big shift. Note – the location is this: Bottom Eng Engineering screen -> other displays -> bottom end (1st column in this window).
6When happy with the autofocus convergence, again record the mean of a few readings on the movie display.
7Change “Fine Focus” to 1 greater than the original value (yes – the opposite end; build the coverage up in alternating fashion to avoid any possibility of hysteresis) and measure the peak mean again.
8Continue filling in the steps until all have been covered.

This whole process takes 15 or 20 minutes.

Data Analysis

By Hand

Plot the means of each set of signal numbers vs the Fine Focus setting at which they were secured. Draw a smooth curve through the data points (smooth through ragged data points, don’t join them). Select three or four levels at which the data are reasonably well behaved (i.e. not too near the peak, where seeing has the biggest effect on signal level, not too near the background level) and determine the Fine Focus values corresponding LHS and RHS of the plotted curve; average these to find the centre. The different centre values should agree to 0.01 or 0.03 depending on how ragged the measures were. The mean is the new estimate for the CGS4 Autoguider Fine Focus setting. (One gets surprisingly similar results by just interpolating linearly between the data points to get the LHS and RHS values.)

Using the Spreadsheet

Run the FINE FOCUS CALCULATOR Spreadsheet by double-clicking it on the desktop of the summit PC (this is a shortcut to the actual sheet, which is stored in C:\UKIRT\Utilities).

Enter the mean values determined above against the fine focus setting. Unless you know what you’re doing with EXCEL, don’t try to expand the tables to allow for more focus positions; let me (AJA) know if this needs to be done.

The spreadsheet does a fit to the values you enter. If any value is missing, leave the corresponding entry empty in the table, and it will be ignored. 

Copy the coefficients of x(squared) and x as reported in the chart window into the box to its right (you can drag the equation around if you need to). The sheet calculates the best-fit focus position and reports it in the green box. Please document this value in the “history” sheet, which you can get to by clicking on its tab at the bottom of the excel window.


This should be measured at the end of every night indicated on the ukirt schedule as “em”. It is performed with the echelle, so as to achieve the required resolution. Instructions are available on the staff wiki. Please be careful to move the dichroic to the CGS4 (north) port before taking sky and dome data. 

UIST Grism sensitivity check

A few times a year we should observe the same source with the HK grism, just to keep an eye on its transmission, etc. So, if you have a spare ten minutes please run the “HK Grism sensitivity check” MSB in the EC1 QA nite programme. Its set for just the one target, g-129 (RA 03hrs, Dec +18) so it should only come up in the QT if the target’s available. The target is faint (10th-11th), so imaging acquisition on the target itself (rather than a nearcmc) should be fine. Its set for just one quad, with 60sec exposures. 

CGS4 Position Angle Calibration

Protocol is as follows: TSS runs a two-row peakup for one and two pixel slits: Peakup is run twice, on two rows well spaced on the array – 140 and 200 for example.

If significantly nonzero angles are found (note – image rotator can set to accuracy of about 0.1 degree, and with the 2-pixel slit successive determinations can vary by 0.05 degree in good conditions):

TSS saves the slit position angle – SAVE_PA <SLITNAME> 

where SLITNAME is something informative such as 1pixdec00. Inform Tom Kerr by email that the offset has changed; he will adjust the settings in the cgs4 config file.

Until the config has been changed, using the affected slit will require the offset to be recalled from the saved version, until the new values are made the default:ADJUST_PA <SLITNAME>

Aperture Measurements

Note – enter all aperture determinations on the wiki, at:  http://wiki.jach.hawaii.edu/staff_wiki-bin/wiki/UKIRT_QA_night_Logs,  and send a note to Tom and Russell that you have done so.

The MSBs you will need are in U/EC/1. Titles as follows:

InstrumentMSB name in U/EC/1
CGS4CGS4 aperture/PA measurements
UFTIUFTI (full and sub) apertures
UISTALL UIST aperture measurements (2 cameras)

Most of these have a target component set to RA, Dec 0,0. You will need to disable observability to get them to come up. Skip the slew and the TSS will go to an appropriate star. The following sections describe the details for each instrument/mode.


ALL APERTURES (including coronograph and polarimetry) can be measured from the same imaging sequence, apart from the 0.06″ imaging aperture. This is because all we’re trying to do here is establish the aperture values that put the star on a specific pixel on the array. The UIST apertures MSB includes a separate 0.06 imaging Observation. The pixels used for imaging, spectroscopy, etc. are all listed in the UIST engineering web pages. Look here for the most up-to-date values – don’t trust numbers found elsewhere! 

If pushed for time, do the imaging apertures (because for spectroscopy and IFU we use upick to get onto the reference pixel).

Note: for the Imaging 0.12″ and 0.06″ apertures, use the full array and the same reference pixel (see the table in the above eng. web page).

There is a “UIST apertures” sequence in U/EC/1; load it, skip the slew and go to a 9-10th mag A- or F-type star. The sequence does 3-second exposures, which will be enough for good centroiding on the star image. Stop at break and run Movie. Then follow the steps outlined below (under Camera Aperture Determinations) to centre the star on the appropriate pixel on the array. 

Apertures should be measured for the following (all but the latter are done in the first Observation in the MSB)

  • Imaging – 0.12″ (APER UIST_im12)
  • Spectroscopy. Determine 4-pixel slit reference pixel. (APER UIST_sp)
  • IFU. (APER UIST_ifu)
  • Imaging polarimetry. Measured WITHOUT the polarimetry prism or waveplate in the beam; so with the same UIST imaging observation used to measure the other apertures, just centre the star on the pixel value for Imaging Polarimetry. (APER UIST_pol)
  • Coronograph: again, just measure this using the same imaging Observation at the pixel specified for Coronograph 6-pix wire (NOT Coron with Pol). (APER UIST_ cor)
  • Imaging – 0.06″ (this one uses the full array but 0.06 pixels; it is therefore a separate observation in the MSB) (APER UIST_im06)


Use the U/EC/1 MSB for CGS4 instrument aperture measurement. If there is time to do more than J and K (which are the important ones), then edit the wavelength and resubmit/query/fetch. Also if there is time to do several slit PAs, edit them as appropriate; a range of 0, -15, -30, -60, -90, -120, -150 is useful to cover. Using autopeak, each setting will take a few minutes. 

The full list of required aperture measurements is given in the form in a later section.


Essentially these are obtained as described for UIST. The MSB is called “UFTI (full and sub) apertures”, in U/EC/1; and although it does a series of observes, you can choose to use that or to just run movie. For the full array, centre the image on pixel [533,488] (full array) and [257,257] (subarray). For UFTI polarimetry, centre on [470,850] – note, this is WITHOUT the prism in the beam so just use the same imaging sequence as for the full array. (The UFTI pol aperture will then put the e- and o-beam images of the star on pixels 470,750 and 470,950 [H-band estimate, based on known beam separation].)  The running order should be Full array, Polarimetry, and subarray. Note that the pol slot is well above the centre of the array. 

Camera Aperture Determinations

In the TCS command window:

1. Use the PORT command to position the dichroic for the instrument

2. Use the APER command to select the correct instrument aperture (need to change this for each UIST mode)

3. Send the telescope to a CMC target (if necessary)

4. Use the COLLIM command to adjust the telescope pointing model

5. Turn on autoguiding

6. Click on the appropriate TCS Tab to display Instrument Apertures

7. Use the AOFFSET and AZERO commands to adjust the instrument aperture values until the target is in the correct spot on the array. For UIST see the UIST engineering web pages for the most up-to-date values. For UFTI, see above. Directions: for UIST, AOFF 20,10 takes the star down 20 and right 10 on the gaia display. For UFTI, AOFF 20,10 takes the star right 20 and down 10. There is a spreadsheet which can help you to quickly get onto the pixel – /home/observer/adamson/procs/APERTURECALC.ods (runs in open office).

8. The TCS status display will show the instrument aperture values (X and Y). Enter the numbers in the wiki: (http://wiki.jach.hawaii.edu/staff_wiki-bin/wiki/UKIRT_QA_night_Logs) and email ukirt_ss and Russell so that the OT configuration files and TSS values can be updated. 

Additional notes for Polarimetry apertures – really for CJD’s benefit…

UFTI : the current measured APERTURE offsets between imaging and imaging pol is: +5.7,-32.5

If you do need to check the imaging to imaging-pol offset, then in EC1 there is an MSB containing two obs, one for the normal, full-array instrument aperture, and a second for the UFTI pol aperture. UFTI pol targets should ALWAYS be placed in the top-half of the array, so the difference between the two apertures should be about 30″ (i.e. toff of 0,-30″ should move a target from the imaging aperture to the pol aperture, so that the two images of the star [the e- and o-beams] will be well-positioned in the centre of the two upper pol windows).


UIST Image Rotator Center-of-Rotation Measurement

Only do this if UIST has been disassembed during engineering and the center-of-rotation is therefore unknown.

This procedure determines the center-of-rotation of the UIST image rotator.

  1. Prepare UIST by running a sequence that has a suitable target and a UIST Target Acquisition component.
  2. When the sequence pauses at the breakForMovie location, run Movie from the sequence console.
  3. Use the APER UIST_AX command on the TCS command-line interface to select the special UIST aperture for this procedure. The star position will move to a new location on the array, but that is ok for this test.
  4. The idea now is to rotate the UIST image rotator through its full range of motion (-90 deg to +90 deg) and record where on the array the star is located at each position angle. A suggested form for recording the data is in the appendix of this page. The image rotator can be moved from the <TBD> EPICS screen. The star location can be determined using GAIA’s Pick Object tool. 

UIST Flexure Measurement

Only do this if the telescope or UIST structure has been changed and the flexure is therefore unknown.

This procedure determines the UIST flexure terms to be used in the TCS pointing model when a UIST aperture is selected.

  1. Prepare UIST by running a UIST read noise sequence.
  2. From the File->Load menu item, load the “UIST_flexure.exec” file.
  3. Run the sequence from the top.
  4. At the break in the sequence after the “set OBJECT” command, run the “long_ptest_uist” command on the TCS command-line interface. This command will cycle through a list of 60 sky positions. At each position, do the following:
    1. TSS: push the star into the guide box using the buttons and turn on autoguiding. 
    2. Observer: Click the “Run from highlight” button. 
    3. TSS: hit return to go to the next star. 

Dichroic Vignetting Check

Only do this if the dichroic has recently been changed and the vignetted region is therefore unknown.

  1. Use the PORT command to select a port of your choice (north, east, south or west).
  2. Use the APER AXIS command to select the AXIS aperture.
  3. Send the telescope to a CMC target.
  4. Use the COLLIM command to adjust the telescope pointing model such that the star is approximately centered in the guide box.
  5. Turn autoguiding OFF.
  6. Switch the guide camera to acquisition mode.
  7. Use the REFR OFF command to tell the TCS to not compensate for the dichroic refraction. The star will move outside the guide box, but that is OK. Do not move the star back into the guide box.
  8. Use the VIGN OFF command to allow the TCS to move the crosshead into the vignetted regions.
  9. Use the TOFF command to offset the telescope to determine the vignetted regions.The principle is to move the star image until it starts to ‘split’, then back off until you have a minimal or non-existent ‘partner’ image. Note: this is difficult for the E/W ports, as the acquisition mode on the guider creates a ‘streak’ off to the right, so some of the ‘split’ images will be buried somewhat in this. Also, when the star image goes into the guide box (but no companion split image) you are off the dichroic (because there is no longer any refraction due to the dichroic); move back towards the dichroic to find the area where the ‘split’ starts to occur. Apply offsets as follows: EAST and WEST portsTOFF X 0NORTH and SOUTH portsTOFF 0 YAs an example – the vignetted regions as of January 2001 were:East port-146.5 < X < -125.5 & 124 < X < 141.5 arcsec West port-134.5 < X < -117 & 132.5 < X < 153.5 arcsec North port-163 < Y < -141.5 & 108 < Y < 125 arcsec South port-150.5 < Y < -133.5 & 116 < Y < 137.5 arcsec
  10. Record the telescope offsets at which vignetting occurs (double images in the guider draw window). There is a form at the end of this page to suggest a method for recording the data.
  11. Set the telescope back to the zero offset position (TOFF 0 0).
  12. Go back to step 9 and repeat the procedure for the other 3 ports.
  13. When you are done, it is very important that you use the VIGN ON and REFR ON commands to tell the TCS to avoid the vignetted regions and compensate for the dichroic refraction.
  14. Send the results to r.kackley@jach.hawaii.edu

APPENDIX: BACKGROUND to Focusing protocols

Fast Guider

The Fast Guider has two basic modes of operation: Normal and Autofocus. 
The two basic modes use different lenses, located at a pupil plane inside the Fast Guider, to form an image on the guider CCD. One is a single lens, forming a single star image on the CCD for normal (fast) guiding. The second is a 2×2 lenselet array which forms 4 star images on the CCD. (Normal Guiding has subsidiary modes, “focus” and “acquisition” which are currently rarely used. These differ from Fast Guiding mainly in the binning of the pixels from the CCD and in the field of view which is read out.) 

The optimum arrangement derived and employed for Fast Guiding is to bin the native pixels of the CCD into 3×3 “superpixels” around 0.”93 arcsec square: we then read out a 4×4 array of the superpixels and use the signal on the central four and that on the outer twelve to determine the location of the centroid of the image. 

In fast guiding the the secondary mirror feedback loop tries to hold the centroid at the point defined by the inner corners of the four central pixels, by tipping and tilting the secondary mirror with its piexo actuators. In auto-focussing we use the fact that the radial separation of the four images is a measure of the telescope focus position, so by measuring this separation over a period of time and comparing it with a nominal value a focus correction can be derived which is then applied to the telescope secondary mirror Z position. In principle this could be done as fast as the CCD is read out; in practice the piezos do not have enough throw to correct for the focus excursions actually seen, so the focus corrections are applied via the hexapod, which is a lot slower (~Hz) than the tip-tilt system (~10s of Hz).

We normally average the 60-Hz focus correction measures for periods of 2, 4 8, 16 or 32 seconds. (In this process we also determine the Root-Mean-Square (RMS) variation in the numerous computed focus corrections which were combined to get the average; this quantity (“Zrms”) is a measure of the seeing.) The nominal value for the radial spacing of the images is actually just set by the CCD array pixel spacing: in autofocus mode we read out an 8×8 array of the same 3×3 superpixels used in fast guiding. This comprises four 4×4 “fast-guide” type sub-arrays, one for each subimage. The reference spacing is then just that of the four centrepoints of the central quartets of pixels, and for each readout of the array we compute the displacement of the centroids of the sub-images from the reference positions; its radial component measures the current defocus (and the mean X-Y component measures tip-tilt, just as in fast guiding, so that this process is still available).

In autofocus mode the larger readout area slows the process somewhat: whereas in fast guiding we can read out at 100 Hz (or in fact a good deal faster), in autofocus mode the standard readout rate is 60 Hz. When in autofocus mode the guider systems determine focus corrections averaged over the chosen time and send these corrections to the hexapod, which moves the secondary in the Z direction in such a way as to bring the centres of the four subimages onto the four centrepoints of the four subarrays of superpixels.

 Guider Fine Focus

The single and quadruple lenses are both carried in the same lens wheel, which can be moved towards and away from the CCD. This adjustment is called “Fine Focus” on the Botttom-End Control screens and is a measure of the lens wheel position (in mm) relative to an arbitrary zero. Each scientific instrument has a slightly different optimum telescope focus setting. But as we have seen the guider in autofocus mode can only correct the telescope focus to bring the images into coincidence with the reference points on the CCD. 

However the radial spacing of the images on the CCD is a function of the overall focus of the telscope and guider system, including the lenslet array. Thus moving the lenslet array towards and away from the CCD also changes the spacing of the images. This enables us to use differnt telescope focus settings for the various instruments: we just select for each an optimum distance of the lenslet array from the CCD, i.e. an optimum setting for the guider “Fine Focus”. These values are referred to as the Autoguider Fine Focus Offsets.

Note that the Fine Focus setting is different when in autoguider and Normal Guide mode as the lenses used have different focal lengths. 


Selecting a variable to be measured to determine focus quality is non-trivial, but image central intensity in one form or another works well. Strehl ratio on UFTI/TUFTI images should be fine and at least nominally independent of transparency. In the case of CGS4 we use the signal in the selected row as displayed on the Movie screen. (This is not independent of transparency; a normalised signal (the ratio of the central row to the two adjacent rows, perhaps?) is possible but not currently available.)

Appendix: Forms (Use is optional, since no-one ever used them…)

Full CGS4 Aperture Measurements


Recorded by:
GratingPAInstap xInstap y
40l/mm, 1st order0







150l/mm, 2nd order0














40, 2 pixel0

40, 4 pix0

Ech, 2 pix0

150, 2 pix0

40, 1pixel, 1.25um0

40,1 pixel, 1.65um0

40, 1pix, prism (use lower beam)0

Form for CGS4 fine focus

UT Date

 Signal Values 
FF setting123456Mean










Best focus setting: 

Form for CGS4 slit offsets

DateObserversOffset (2pix)Offset (1pix)Saved as

UFTI Fine Focus


Recorded by
FF settingFWHMFWHMFWHMFWHMMean FWHM for spreadsheet












FF settingFWHMFWHMFWHMFWHMMean FWHM for spreadsheet












Best focus setting: 

UIST Image Rotator Center-of-Rotation Measurement


Recorded by

Star Location
Position Angle (deg)
X (pixels)
Y (pixels)







Dichroic Vignetted Regions


Recorded by

+RA offset (arcsec)
-RA offset (arcsec)+Dec offset (arcsec)-Dec offset (arcsec)
Inner edge
Outer edgeInner edgeOuter edgeInner edgeOuter edgeInner edgeOuter edge





Full CGS4 Aperture Measurements