• position the cursor over an unsaturated star with reasonable S:N
• hit 'r'
• read off the FWHM (last number on the line below iraf's plot of the PSF)

If the seeing is good (< 1 arcsec), it's worth focusing with accuracy ~ 0.02 mm (which correponds to 0.16 arcsec diameter at the f/11 focal plane). For fine-tuning the focus, you may want to iterate manually, rather than use focusrun:

• TO> focus 98.15 [and wait for focus to arrive]
• TO> glance acam 7

Defocus broadens an image obtained at the f/11 Cassegrain focus of the WHT by 8.2 arcsec (33 ACAM pixels) per mm of secondary movement. E.g. 0.8 mm of defocus (~ 26 pixels) looks like this:

5.3 - Focus offsets
Focus changes as a function of temperature, mainly because of the change in length of the telescope tube, but the relevant focus correction for this is applied automatically by the telescope control system (TCS), and is shown on the telescope status display. The TCS also applies a correction for changes in focus with telescope elevation, but this is not reported on the telescope status display.

By default, when the ACAM control system is started up, it enables automatic application of known focus offsets relative to that for Sloan r (#702), for filters mounted in the filter wheels, e.g. at the time of writing the offset used for Sloan g is 0.08 mm. The offset for each filter is taken from the ING filter database, and is displayed on the ACAM control gui (in grey characters, but this does not mean that the offset is not active). The relevant focus offset will be applied whenever a filter is moved into the light path, and the value of this offset will be reported (as 'df') on the telescope-control information panel.

The automatic focus offsets can be disabled with:

acam_focus disabled

and re-enabled with:

acam_focus enabled

New offsets can be entered by the observer at the ACAM control gui (click on the 'Edit' button) and are copied into the ING filter database, for use whenever they are next required (the old values are over-written, although there are some reference values stored in the filter database).

The focus offsets between filters depend mainly on filter bandpass and thickness, and are discussed on the ACAM filter focus offsets page. For convenience, we note here, for the new Sloan filters (#700 - 704), the predicted offsets in focus in mm (all relative to Sloan r #702), and the median focus offsets (and rms error) deduced from focus runs so far (as of April 2012):

In the unlikely event of requiring more than one filter in the optical path at a time, the automatic focus offsets should be disabled, otherwise the observing system will apply the sum of the individual filter focus offsets.

There are no automatic focus offsets when switching to (or between) filters mounted in the focal-plane slide (as opposed to the filter wheels). The focus must be determined on-sky as above, and must thereafter be set manually with the focus command e.g.:

• TO> focus 98.15

5.4 - Focus-offset changes when switching between ACAM and other instruments
If switching between ACAM and another instrument during the night (or indeed, from one night to the next), the operator or observer needs to reset the telescope focus to the value found during the focus run for that instrument. However it's also wise to keep an eye on the focus offset discussed above (df on the TCS display), to avoid accidentally inheriting the filter focus offset of the wrong instrument.

The safest way of switching is for the telescope operator to use the INSTRUMENT command (in the TCS control window). The INSTRUMENT command sets df to zero. So if the requested instrument is ACAM, it's then necessary to change or confirm the filter, e.g. with:

acamimage SlnR 1

to make sure that the correct df for the filter is applied.

Occasionally, when switching to ISIS to take an observation of a bright star, and when guiding is not needed and time is short, the INSTRUMENT command is not used, in which case the df is unchanged, and will usually be non-zero and wrong. One workaround here is to first set each ACAM wheel, and the ACAM focal-plane slide, to a position with no filter, and hence df = 0, suitable for ISIS.

Back to top

• 6.1 - Setting up the CCD
• 6.2 - Biases, sky flats, dome flats
• 6.3 - Target acquisition

6.1 - Setting up the CCD
The field of view in imaging mode is circular, 8 arcmin in diameter, as illustrated by this image of part of the moon:

The vignetting at the left edge is due to a small mis-positioning of the CCD in its cryostat.

At sky position angle 0, north is right, east is up.

The useful imaging area (+ overscan at left edge) occupies approximately 1 < x < 2030, 940 < y < 2950. The scale is 0.25 arcsec/pixel.

If both imaging and spectroscopy are to be used, it's better to window the CCD so as to include the useful area for spectroscopy as well, using:

TO> window acam 1 "[1:2148,800:3300]"

Most observers will not need binning (the scale is 0.25 arcsec/pixel), but it can be implemented with e.g.:

TO> bin acam 2 2

to bin x2 in each direction.

To change the readout speed:

TO> rspeed acam slow

or

TO> rspeed acam fast

6.2 - Biases, sky flats, dome flats

Bias frames
Bias frames may be obtained in the usual way (after making sure that the dome lights are switched off):

TO> bias acam or

TO> multbias acam nn
where nn is the number of biases required.

Note that the overscan strips lie at x < 50 and x > 2048.

Observers using small windows for their science targets (to minimise readout overheads) may want to define additional windows to sample the overscan strip, see Appendix 2 for impact on readout overhead.

ACAM biases should not be obtained simultaneously with biases on ISIS - this results in horizontal features on both sets of biases.

Sky flats, autoflat

• Ask the operator to open the dome and mirror petals and point the telescope at a 'blank' field.
• Insert the feed flat with:
  TO> agacam
• Check that the slit mask is out of the beam, and insert the required filter, e.g. Sloan r:
  TO> acamimage SlnR
• Check the count level with a glance exposure:
  TO> glance acam 1
  Typically, a few x 10^4 counts/pixel are required, but avoid saturation (~ 60k counts/pixel).
• Obtain sky flats with e.g.
  TO> multsky acam 10 2 "Sky flat R"
  (10 flats, 2 sec each). It's usually desirable to dither the telescope a few arcec after each exposure to allow subsequent elimination of any star images. The telescope operator usually takes care of the dithering (sometimes by running the spiral.csh script, which offsets the telescope a few arcsec a few times per minute, ensuring that no two flats are exposed at the same position on the sky).

Suggested sky-flat exposure times can be found in Appendix 4.

A script autoflat is available for obtaining several sky flats though each of a specified set of filters. To run the script, set the CCD readout speed, binning and windowing (window 1 only - the script will not work for other windows) as required for the science exposures, then:

cd /home/whtobs/acam/jmcc/under_development/autoflat

Then e.g.:

python autoflat.py acam 4 SlnG SlnR SlnZ

to obtain 4 flats, through each of the Sloan g, r and z filters. autoflat aims to deliver for each flat between 20k and 40k counts per pixel. It takes test exposures (using a small window, in fast readout mode) to monitor the sky brightness and sets exposure times (min and max 1 and 120 sec) accordingly, working through the filters in an appropriate order (i.e. not necessarily that supplied by the user). There is automatic dithering between exposures. If the script fails part way through, take the rest of the sky flats manually.

The autoflat script was provided by James McCormac.

Dome flats
Dome flats can be obtained by illuminating the dome with the top-end flat-field lamps (Appendix 4) and, ideally, pointing the telescope to a relatively featureless area of the dome , e.g. at elevation 45 deg (ask an ING member of staff to move the telescope - visiting observers must not do this).

6.3 - Target acquisition

Ask the telescope operator to slew the telescope to the target. At sky PA = 0, N is right and E is up on the images. At sky PA = 270, N is up and E is left. The target usually appears within a few arcsec of the rotator centre, which lies near x, y = 1098, 1890 (10/2009) on the unwindowed CCD.

If required, the operator will look for a guide star.

Check that the slit slide and filter wheels are in the correct position (usually two of them will be clear, one will have a filter in the beam).

Take a test glance exposure, check the image against the finding chart:

TO> glance acam 10 &

Then check that the telescope is autoguiding, and take the science exposure, e.g.

TO> run acam 900 &

or a series of science exposures, e.g.:

TO> multrun acam 3 900 &

will take 3 900-sec exposures (if dithering is required between individual exposures, use the multdither command).

Back to top

• 7.1 - Setting up the CCD
• 7.2 - Rotation / curvature on the CCD
• 7.3 - Biases, spectroscopic configuration, flat-fields and arcs
• 7.4 - Focusing the telescope
• 7.5 - Target acquisition
• 7.6 - Scripts
• 7.7 - Data reduction, and quick-look

7.1 - Setting up the CCD
The (fixed) format of a long-slit spectrum obtained with the V400 VPH grating in filter wheel 2 (currently the only option) is illustrated below (taken in 2009, click on the image for an enlargement):

The CCD window used above (outer boundary of the figure) has 1 < x < 2148, 800 < y < 3300, and can be set with:

TO> window acam 1 "[1:2148,800:3300]"

This is big enough to include the whole useful wavelength range in y, and also the whole 8-arcmin field of view in imaging mode.

In the spatial direction, the spectrum runs from x = 335 - 1725, corresponding to a slit length ~ 5.8 arcmin. Red is at the top, blue is at the bottom. A rough equivalence between y pixel value and wavelength is given on the ACAM spectrocopy page. For the wavelengths of prominent arc lines, see the ACAM CuAr and CuNe arc maps.

For spectroscopic programmes not requiring science imaging, a smaller window could be used e.g.:

TO> window acam 2 "[665:1385,800:3300]"

In imaging mode (for acquisition onto the slit) this still gives a field of view of 3 armcmin x 8 arcmin, sufficient to identify the field. NB if a measurement of the mean bias level during each exposure is needed, extend the window to the left edge (ideally) of the CCD (to include the bias strip, 1 < x < 50).

For slit widths >~ 1 arcsec (spectroscopic resolution >~ 4 pixels on the CCD), the CCD can be binned, to reduce readout noise and readout overheads, e.g.:

TO> bin acam 1 2

The spatial scale on the CCD is 0.25 arcsec/pixel, as in imaging mode, so it may also be useful to bin in the spatial direction, depending on the seeing and the angular diameter of the target, e.g.:

TO> bin acam 2 2

With this binning, and the window size specified in the previous paragraph, the dead-time between consecutive exposures is only 5 sec.

The readout noise per unbinned pixel is ~ 3 electrons in slow readout mode, 6 in fast. Set the speed with e.g.

TO> rspeed acam slow

7.2 - Rotation / curvature on the CCD

Arc lines: the central ~ 1 arcmin of the arc lines (slit images) is rotated relative to the CCD rows by only ~ 0.1 deg (clockwise).

(From June 2009 until adjustment of the cryostat in late 2010, the rotation was larger, ~ 0.7 deg clockwise.)

The arc lines are also curved (as can be seen by clicking on the above figure for an enlargement), with the y value at the centre of the line being slightly higher than the mean of the y values at the left (2.5 pixels lower) and right (4.5 pixels lower) ends of the line. The curvature arises as a result of the VPH (and its flanking prisms) being in an f/22 beam, rather than a collimated beam.

Spectrum: the spectrum is rotated anti-clockwise relative to the CCD columns, by about 2.0 deg.

7.3 - Biases, spectroscopic configuration, flatfields and arcs

7.3.1 - Bias frames
Bias frames may be obtained in the usual way (after making sure that the dome lights are switched off):

TO> bias acam
or
TO> multbias acam nn
where nn is the number of biases required.

Note that the overscan strips lie at x < 50 and x > 2048.

ACAM biases should not be obtained simultaneously with biases on ISIS - this results in horizontal features on both sets of biases.

7.3.2 - Spectroscopic flats
For spectroscopic flats, deploy the calibration-lamp mirror in the Cassegrain A&G box, switch on the tungsten lamp and insert neutral density filters with total ND = 1.5 (to prevent saturation):

TO> acamcal

TO> complamps w

TO> compnd 1.5

Then select the desired slit width and the VPH grating in the filter wheel, with e.g.:

TO> acamspec V400 1

(NB not acamspec V400 1.0, see the ACAM commands glossary) or

TO> acamspec V400 0.5

or, to include the GG495A order-blocking filter in the light path:

TO> acamspec V400 0.5 GG495A

equivalent to:

TO> acamslit ACAM_0.5

then, if the GG495 filter is in position 6 of wheel 2:

TO> acamwh2 6

Then expose the flat-field images, with:
TO> flat acam 1
or
multflat acam nn 1
where nn is the required number of exposures, and the last entry on the line is the exposure time in sec.
A 1-second exposure through a 1-arcsec slit typically gives maximum ~ 25000 counts per unbinned pixel in the red part of the CCD. Several exposures will be required to reach adequate S:N in the blue.

Then switch off the lamps, remove the ND:

TO> complamps off

TO> compnd 0

Troubleshooting: if your flat-field looks more like a light leak than a flat-field, check that the A&G mirror is in the correct position (i.e. acamcal, not agcomp!).

7.3.3 - Arcs
Arc exposures can be obtained using the standard CuAr and CuNe lamps in the A&G calibration-lamps unit. If LIRIS is mounted at Cassegrain, it's likely that only the CuNe lamp is available.

These lamps provide only weak lines in the blue, so two exposures are needed - a short one for the red end of the spectrum, and a longer on for the blue (no ND filter is required) e.g. for a 1-arcsec slit:

TO> acamcal (if not already deployed)

TO> complamps cuar+cune

TO> arc acam 1

TO> arc acam 15

TO> complamps off

ACAM flexes significantly as the telescope elevation and rotator position angle change. For accurate wavelength calibration, arc exposures should be taken before or after observing each science target.

7.4 - Focusing the telescope
The technique for focusing in spectroscopic mode is similar to that used in imaging mode (see Section 5):

1. Configure ACAM for spectroscopy, ideally with a wide slit (but it doesn't much matter).
2. Acquire a star, mag ~ 10 - 12 (e.g. one of the FT... stars in the TCS catalogue) at the centre of the ACAM slit.
3. Take a spectrum of the star at each of a series of telescope focus positions, using focusrun e.g.:

   TO> focusrun acam 9 10 97.9 0.05

   to take 9 10-sec exposures, initial focus 97.9 mm, stepping in focus by 0.05 mm.

4. Display each spectrum as soon as it has read out, and use imexam option 'j' (not 'r') to measure the spatial FWHM. Note down the focus position and the FWHM.
5. Find by eye the minimum in the trend of FWHM with telescope focus. The trend will often be noisy, due to variable seeing.
6. Set the telescope focus to this value with e.g.:

   focus 98.05

The net telescope focus is actually the sum of this value (and an automatic correction for temperature) and the filter focus offset df. df is shown on the ACAM mimic display, in the entry for the VPH, and is usually set to -0.16 (mm). When the VPH is in the light path, this value is also shown on the TCS information display (bottom line, to the right of the nominal telescope focus value).

In principle, the nominal focus offsets df (df = 0 for the Sloan r filter, by convention) take care of the change in telescope focus required when changing between filters or between filter and VPH. In practice, the measured df for spectroscopy has in the past been found to range from -0.11 to -0.22 (perhaps because of flexure in ACAM), and if carrying out both imaging and spectroscopy, it's a good idea to check the focus for both (and tweak the focus offest for the VPH accordingly).

An error in applied focus of 0.05 mm corresponds to a defocus in the focal plane of about 0.5 arcsec.

If switching between ACAM and another WHT instrument during the night (or day), please note the advice under Section 5.4 above about the risk of inheriting the focus offset of the wrong instrument.

7.5 - Target acquisition
ACAM has no equivalent of the slit-viewing camera used with ISIS. Targets are acquired by imaging the back of the slit and then the target, measuring their relative positions, and offsetting the telescope accordingly (LIRIS observers use a similar technique). The procedure is described on the ACAM spectroscopic-acquisition page.

Acquisition is normally carried out by the telescope operator, using the ACAM acquisition tool.

The image above shows a star acquired onto the centre of a slit (no disperser in the light path).

If the target is very faint, a blind-offset star will be required. Experience suggests that, to minimise acquisition overheads, a blind-offset star should be provided if the time taken to image a star-like target with S:N ~ 30 is > 3 sec. This translates, for 1-arcsec seeing, into magnitude limits R ~ 19.5 / 19.2 / 18.6 (for dark / grey / bright of moon), and for 1.5-arcsec seeing, R ~ 19.2 / 18.8 / 18.2 (for dark / grey / bright).

7.6 - Scripts
To save typing during the night, it may be useful to store some command sequences as scripts in the /home/whtobs/acam directory. This directory currently includes examples called acq (imaging mode for target acquisition), slit (positions a slit in the focal plane) and spec (positions a slit in the focal plane, inserts the VPH grating). To execute a script, just type its name, e.g. to execute a script called spec:

TO> acam/spec

Warning! These scripts send several offset values to the telescope-control system (TCS) in quick succession, and occasionally cause communication with the TCS to hang, which can be solved only by restarting the observing system (allow ~ 10 mins for this). This is under investigation.

7.7 Data reduction and quick-look

7.7.1 Quick-look spectrum extraction
Run the ACAM quick-look script with:

ecl> acam
(needed initially, to load ING's ACAM-related scripts)

ecl> acam_ql

and respond to the prompts, or give required parameters on the command line, e.g.:

ecl> acam_ql r1234567 985

extracts from r1234567.fit a spectrum near x = 985 on the CCD.

Additional parameters can be specified on the command line, e.g.:

ecl> acam_ql r1234567 985 s f y

's' indicates a stellar object (alternatively 'e' for extended). 'f' indicates a faint object (alternatively 'b' for bright - triggering a different optimal extraction). The final parameter on the line indicates whether or not the extracted spectrum is to be kept ('y' for yes or 'n' for 'o').

The names of the data, destination and scratch directories used can be changed using iraf's epar. If acam_ql fails to find the named file, it will report to the user the name of the directory in which it is looking.

The script extracts the spectrum, calibrates approximately in wavelength and intensity (using archival calibration data), and displays it in an iraf graphics window using splot (all the usual keystrokes available). Hit 'q' to exit from the plot and be offered the option of saving the spectrum.

Above is an example output from acam_ql. It shows an ACAM spectrum of a broad-absorption-line quasar at redshift 3.06. The prominent broad emission lines at wavelengths ~ 5000 - 6300 A are Lyα/NV, SiIV and CIV. The broad absorption lines just blueward of CIV (also visible in OVI, NV and SiIV) are due to ionised outflows with speeds up to ~ 15000 km/s. Features due to the earth's atmosphere at 5577 A (remnant of a sky line) and at ~ 7600 A (A-band absorption) provide a helpful check of the wavelength calibration.

acam_ql was developed by Javier Méndez, October 2012.

7.7.2 Alignment of sky lines with CCD rows
Accurate sky subtraction is complicated by the small relative rotations of the slit, dispersion direction and CCD, i.e. by the rotations of the sky/arc lines (approximately left-right) and the spectrum (approximately up-down) relative to the CCD rows and columns. For bright targets, this is not a problem, but for fainter targets (e.g. mag > 19) the CCD images need to be rotated (after debiasing and flat-fielding) to align the sky lines better with the CCD rows. Otherwise the sky will not subtract well, particularly in the red, where the sky lines are bright.

Until 2010 November, the rotation required (for binning 1 x 1 or 2 x 2) was 0.66 deg, but is now much less (following rotation of the CCD on its mount). This can be corrected with e.g. the iraf command rotate:

imgeom> rotate r1318756 rr1318756 0.66
Due to the curvature of the skylines no single rotation will give a perfect alignment.

The spectrum may then be extracted using the quick-look script documented in Section 7.7.1, or using iraf apall (below).

7.7.3 - Spectrum extraction using iraf apall
The apextract package may already be loaded (e.g. if you are using the iraf session running on the instrument-control screens). If not, load it with:

ecl> noao

ecl> twodspec

ecl> apextract

Then edit apall's parameters, using:

ecl> epar apall

E.g. the parameter 'background' should be set to 'fit'. Turn on the 'fit' and 'fit2d' options and set the sky area close to the spectrum (e.g. 10 pixels left and right). A background fitting with Legendre polynomials and order 2 or 3 seems to work best.

To end editing the parameters type ':q' (or ':go' to end editing and run the program directly). Then run apall:

ecl> apall r1234567[1] out=s567

To display the resulting (one-dimensional) spectrum:

ecl> splot s567

Acknowledgment
This section (7) is based on notes kindly provided by Paul Groot, who was one of the first users of ACAM in spectroscopic mode, following commissioning in 2009.

Back to top

On the day after your observing run, the ING daytime support astronomer should make sure that the default broad-band filter set is restored in ACAM.

Back to top

Header     Content                Possible values  

TELFOCUS   Tel focus (M2) 
FOCUSTMP   Focus corrn for temp
FOCUSALT   Focus corrn for elev
FOCUSFLT   Focus corrn for filter

ROTSKYPA   Sky PA
MNTPASTA   Mount PA at start
MNTPAEND   Mount PA at end

CAGMIRRO   A&G mirror deployed    LARGEFEED or SMALLFEED
PLATESCA   Plate scale at f/11    4.5 arcsec/mm (fixed)
           focal plane

INSTRUME   Instrument name        ACAM (fixed)
ACAMMODE   Observing mode         IMAGING, SPECTROSCOPY

ACAMSLIT   Position of slit mask  CLEAR, PINHOLE or SLIT
ACAMSLI    Component deployed     CLR, PIN (pinhole mask,
           in focal plane         slit width, or 4 characters
ACAMMASK   Deployed mask          Name of deployed mask

ACAMFSLI   Filter in slit unit    CLEAR, or name of filter in slit unit  

ACAMWH1    Wheel 1 position       CLEAR, or name of filter/disperser
ACAMWH2    Wheel 2 position       CLEAR, or name of filter/disperser
ACAMDISP   Name of disperser in   NONE, or name of disperser
           wheel
ACAMFILT   Names of filters in
           wheels 1 + 2

ACAMTFOC   Whether auto focus     ENABLED or DISABLED
           offsets enabled
ACAMFOFF   Telescope focus offset
           applied

and the usual detector-related FITS headers, including:

DISPAXIS   Dispersion axis        2 (fixed)
DETECTOR   CCD name               AUXCAM (fixed)

Full example FITS headers for science and calibration exposures obtained with ACAM (and with other ING instruments) can be found on the ING FITS headers page. Information about the accuracy of the UT given in the FITS headers can be found on the WHT time-stamp page.

The 'standard' window [1:2148,800:3300] serves for both imaging and spectroscopy.

w1, w2, w3 and w4 above indicate arbitrary 100-pixel (i.e. 25-arcsec) windows defined as follows:

w1 = [1100:1200,1900:2000]
w2 = [400:500,2400:2500]
w3 = [1500:1600,2400:2500] 
w4 = [400:500,1200:1300]

Telescope pointing:	< 180 sec (azimuth slew 1 deg/sec)
Target acquisition (imaging):	0 sec (the pointing rms is a few arcsec)
Target acquisition (spectroscopy):	~ 100 - 200 sec, using the acquisition tool, and assuming the targets are easily visible on a few-sec image (longer e.g. if blind offsets are required)
Autoguider preparation:	60 sec
Detector readout	See Appendix 2 above
Switch between science observing and spectroscopic arcs/flats:	20 sec
Movement of filter wheel	5 sec
Changing filter (i.e. removal/installation) in one of the wheels	~ 5 minutes (including movement of telescope to zenith)

Sky flats
Broad-band sky flats with ~ 30k counts per pixel (slow readout mode) can be obtained with ACAM just after sunset, using the following approximate exposure times (as measured by ING support astronomers during 2011-15):

Filter and      Time after    Exp    
reference no.   sunset (min)  (sec)  

SlnU (#700)          4          1

SlnG (#701)         16          1
                   
SlnR (#702)         23          4  

SlnI (#703)         18          1

SlnZ (#704)         15          1


BESU (#230)         10          1    

BESB (#231)          9          0.4  

SloGunG2 (#219)     11          0.4  

HARV3 (#32)         16          1.5

SloGunR (#216)      20          2    

SloGunI (#217)      24          2

RGOZ1 (#20)         27         15    

So to get evening sky flats through Sloan filters u, g, r, i and z, it's best to start with u, then z, and the order of the rest (g, r, i) doesn't matter much. For morning sky flats, reverse the order.

Exposure times for narrow-band filters can be estimated from the above numbers by scaling by relative bandwidth. Flat-fields through H-alpha filters with 50-A bandpass typically require ~ 3 sec, 15 min after sunset.

During twilight, the sky dims by a factor ~ 2 every 3 minutes - see the twilight sky brightness page.

To avoid contamination by background stars, track a 'blank' area of sky, and ask the telescope operator to dither successive exposures by a few arcsec, or do the dithering yourself, with commands of the form:
user "offset arc 0 10"

at the observers' user interface (the coordinates given are in arcsec, and are absolute, not relative).

Sky flats can also be taken when the sky is cloudy (as long as the telescope operator is satisified that there is no risk of rain). The exposure times required are similar to those recommended above.

A script autoflat (described above in Section 6.2) is available to take a series of flats through a specified set of filters.

Dome flats
The inside of the telescope dome can be illuminated for dome flats by any combination of five lamps (1 = 9 W, 2 = 25 W, 3 = 150 W, 4 = 5 = 500 W) mounted on the top-end ring of the WHT. The lamps can be switched on and off, and the % intensity varied, from the usual instrument-control interface. See the above link for details, and for advice about exposure-to-exposure variations in counts (resulting from e.g. a known 5-Hz flicker if lamp intensity is set to < 100%).

Before changing the illumination in the dome (e.g. by switching off the main fluorescent lamps on the dome wall), please make sure that this doesn't disrupt the work of anyone else, e.g. by plunging them into unexpected darkness.

The following exposure-time / lamp-power combinations yield dome flats with ~ 30k counts when using readout speed = slow (Lilian Dominguez, Raine Karjalainen, James McCormac, Ian Skillen, July 2011 - January 2016):

Filter and ref no.   Exp      Lamp numbers    % Lamp
                     (sec)                    intensity

Broad-band:

SlnU (#700)          30       4 5             100       
SlnG (#701)           6       2               100
SlnR (#702)           2       2                50
SlnI (#703)           3.5     1               100
SlnZ (#704)           2.5     1               100

BESU (#230)          17       4 5             100
BESB (#231)           2       3               100
SloGunG2 (#219)       5       2               100
HARV3 (#32)           3       2               100
SloGunR (#216)        3       2                50?
BESR (#233)           1.2     1 2              70?
SloGunI (#217)        0.8     2                50?
BESI (#234)           2       1               100
RGOZ1 (#20)           4       1               100

Narrow-band:

T4570 (#98)           1.1     3 4 5           100
Halpha 50-mm filter   3.5     3               100
T6589 (#142)          1.5     4               100
T6613 (#147)          1.1     4 5             100
T6631 (#148)          1.1     3 4 5           100
T6637 (#149)          2.0     3 4 5           100
T6709 (#157)          1.7     4 5             100
T6757 (#162)          1.5     3 4 5           100
T6770 (#164)          1.5     4               100
H6577_ANDV (#8346)    1.1     4 5             100
H6621_AND (#7434)     1.2     2 4             100

Ideally, the telescope should be pointed to a reasonably featureless area of the dome, but it probably doesn't matter much.

Note that on a sunny day (and depending on the relative positions of the dome and the sun) the illumination of the dome wall provided by the lower-power lamps might be swamped by sunlight leaking around the top of the dome shutter.