Subject: DIS throughput

From: Ed Turner

Submitted: Mon, 16 Oct 95 09:07:03 EDT

Message number: 7 (previous: 6, next: 8 up: Index)

I am forwarding the attached message from Jim Gunn to apo35-dis.  -Ed Turner


The following table is calibration data taken with DIS on June 13 1994,
on the AB79 standard +26 2606.  The exposure was taken at an airmass of
1.01, slitless, with the low-dispersion gratings.  The following table
gives the wavelength in Angstroms, lam, the counts AVERAGED over a 7
column swath in the blue and an 11-column swath in the red, cnt, the
AB79 monochromatic magnitude of the star ABnu, the assumed atmospheric
extinction, a, (magnitudes), and the computed quantum efficiency on the
sky with the atmosphere corrected for--i.e., the quantum efficiency
referred to the entrance of the telescope, qe.  The assumed dispersion in
the blue was 6.25 A/pixel, in the red 7.00 A/pix, and the gains 0.97
e/DN and 1.47e/DN, respectively.  The exposure time was 5 seconds. 

        BLUE SIDE (multiply cnt by 42.4 get electrons/A ) 

     lam    cnt      ABnu         a         qe
    3500      10    10.79       0.71        0.001
    3600      50    10.72       0.58        0.003
    3700     500    10.65       0.53        0.025
    3800    2000    10.39       0.48        0.077
    3900    3200    10.31       0.44        0.114
    4000    4300    10.14       0.38        0.127
    4500    6600    9.93        0.24        0.159
    5000    7300    9.77        0.18        0.159
    5250    6200    9.72        0.16        0.133
    5500    2800    9.67        0.15        0.060
    5750    1100    9.63        0.14        0.023
    6000     600    9.60        0.13        0.013

    3650     140    10.61       0.55        0.0068
    4046    4600    10.12       0.37        0.134

         RED SIDE (multiply count by 113.2 to get electrons/A )

     lam    cnt      ABnu         a         qe
    5000      20    9.77        0.18        0.001
    5250     350    9.72        0.16        0.016
    5500    1420    9.67        0.15        0.064
    5750    2250    9.63        0.14        0.102
    6000    2450    9.60        0.13        0.112
    7000    2660    9.51        0.08        0.124
    8000    1650    9.47        0.06        0.083
    9000     780    9.46        0.14        0.047
    9500     350    9.46        0.43        0.029
   10000     150    9.46        0.04        0.009

Thus the peak qe in the blue is about 16 percent, and in the red a bit
better than 12 percent. What do we expect? The measured reflectivities
of the mirrors are 0.88, 0.88, and 0.80 for the tertiary, secondary, and
primary, respectively (at about 6400, but these numbers should serve
well except below 4000 and in the infrared dip around 8000; the throughput
of the telescope is thus 0.61. The spectrograph throughput is compounded
of the following approximate numbers, which are roughly the same in the
blue and in the red:

0.90  dichroic
0.97  field optics
0.98  collimator
0.60  grating efficiency
0.92  camera (mostly secondary obscuration in the Schmidt)

The spectrograph throughput is thus approximately 0.47. The blue
CCD has peak qe around 0.60, that of the red about 0.50. Thus we
expect qe's of about 17 percent and 14 percent, respectively, at peak.

The performance is therefore roughly as expected except for the
precipitous drop in sensitivity shortward of 4000 A, the
half-sensitivity point occurring at about 3800 instead of the design
number of 3400.  The CCD qe was measured to decline from about 60
percent at 4000 to about 45 percent at 3500 in the spring of 1994.  A
series of tests to measure the ratio of the throughput at 3650 A and
4046 A (the two strong Mercury arc lines) of the blue-side optical
elements was performed in June 1995, and it was determined that serious
losses were occurring in both the main camera block and the field optics
group; those elements showed ratios of transmission of about 4 and 3,
respectively. For all the elements, the transmission ratio is about 
12:1, the other elements contributing negligibly.

Both of those elements have internal reflective surfaces.  All of the
purely transmissive optical elements including the dichroic had ratios
near unity.  Correcting for the known slight decline in CCD qe, the
ratio of the system qe at 3650 and 4046 should be about 1:16, .062, with
an uncertainty of perhaps 20 percent.  It was measured the year before
with the above data at .0068/.134 = .051, well within the measuring
error of the rather crude technique to measure the optical
transmissions.  It thus does not appear that the condition has worsened
over the past year, and that it does stem from bad reflective coatings. 

Don Loomis, the optician who constructed the optics for DIS, has
reported that it is possible to disassemble the cemented quartz
assemblies involved, and could do so with a turnaround of about two
weeks.  The coaters, ZC&R, have not yet been contacted concerning quotes
to recoat, nor for what might have gone wrong or might go wrong again,
nor for how much responsibility they might we willing to take for the
condition.  Coating the camera, in particular, is complex, and one could
expect a turnaround time of at least a month with another week back at
Loomis's for reassembly, and a week at APO for refocussing and
collimation.  Thus the spectrograph would be out of service for about
two months.  I would suggest that the fix be done during the summer
monsoon of 1996, with exchanging the detectors for more modern CCDs and
electronics to follow within a year.  With more modern detectors,
'super' coatings on the secondary and tertiary and a good coat on the
primary, one could expect peak qes of about 24 percent in the blue and
29 percent in the red (assuming 94 percent and 88 percent, respectively,
for the secondary and tertiary, and primary, respectively, and 65 and 80
percent for the CCD qes. 

APO APO APO APO APO  Apache Point Observatory 3.5m  APO APO APO
APO  This is message 7 in the apo35-dis archive. You can find
APO  the archive on
APO  To join/leave the list, send mail to
APO  To post a message, mail it to