Dear all

Below is Ed Hawkin's report on light leaks in UES.  These have long been
a hazard e.g. Linda Smith reported losing 1 hour in Feb 99 when the black 
cloth blew off.

Ed has reduced the light leak by a factor of over 1000, to a negligible
level.  Outstanding action is to plug the 2 large bolt holes with 
something more permanent than black tape.

Chris


-----------------------------------------------------------------------------

UES Light Leaks Report
----------------------

Light leak tests were carried out on the UES detector on 03/09/99 and 07/09/99 
to try and find the causes of the stray light entering the detector. Until 
this point the detector had been operating with a black cloth covering the 
entire unit and this had stopped the light from entering but not solved the 
problem.

Initially when the cloth was removed and a 5 second exposure taken the CCD 
immediately saturated showing that there was a large amount of light entering. 
This was traced to 2 large bolt holes on the front of the casing, above the 
cryostat interface. When these were taped up most of the light was stopped 
from entering. Many further tests were carried out and various much smaller 
sources of leak were found and taped up. At the end of the tests with the dome 
lights on and no cloth covering the detector the leak was an average of 8 
counts per pixel per second above the bias. When the cloth was introduced this 
was reduced to a negligible level.
As it is thought that the full moon is ~10000 times fainter than the dome 
lights, the leak with no cloth is reduced to <~0.001 counts per second per 
pixel, which is an acceptable level. 

Despite all the tests there is still an unknown source of leak which seems to 
project a polygonal shape onto the CCD, which disappears when the cloth is 
introduced at the front end of the detector casing.

Conclusion: With the cloth on the detector works with virtually no stray 
light (< 1/10000 previous level), but if the cloth were to be blown off 
during observing it should still function well (< 1/1000 previous level), 
although it would be better to eliminate this unknown source.


Table of leaks (with dome lights on)
--------------
                                Bias-subtracted
Source                          counts /sec /pixel   Solution
---------------------------------------------------------------------------
Bolt Holes on face of detector       ~ 15000         Currently taped up but
box                                                  needs something more 
                                                     permanent.
 
Cryostat interfaces                     ~ 80         Taped up - not much else
                                                     can be done really. 

Hatch above slit unit                    ~ 1         Not a major problem.

Black casing behind CCD interface        ~ 0         No leaks.

Screw-holes for preamp box              ~ 10         Currently taped - should
                                                     be checked and rescrewed?

Westinghouse TV camera casing            ~ 5         Not a major problem.

Unknown (polygonal pattern)              ~ 8         Still present if black
                                                     cloth not covering
                                                     cryostat interface.
                                                     Needs investigating.
 
(The CCD was run in TURBO mode, bias level ~ 790 counts)