Light leaks and dark current

[roger smith x294]

Everyone,

You might be interested to know about some experiences 
with excess "darK" current which we have had in various
spectroscopic applications over the last 9 months.  These 
have occurred in locations where we have not had the 
luxury of a dark slide in the "right place" to distinguish
between dark current and light leaks.



1) Light emission from fiber optic communication links.
   ---------------------------------------------------

The CTIO array controller (Arcon) is mounted on the
CCD dewar and is connected by a fiber optic data link.
To the eye this looks dark, but this is due to the long
wavelength of the light rather than the opacity of the
link.  CCDs see the ~950 nm light very nicely.  An image
of the Arcon with fibers glowing can be seen via Mosaic
at....

http://ctios2.ctio.noao.edu/www/arcon/arcon.fiber.html

I took this image with another Arcon/CCD system plus a home
grown lens/shutter assembly.

Not only is there light gushing from the fiber transmitter
/ connector, but also gross light leakage through the fiber
cladding.  You can appreciate the extent of the light loss
by comparing the brightness of the transmit and receive fibers.
Bends or very small pressure on the cladding produce much 
greater leakage.

This kind of thing is very bad inside a bench mounted echelle
spectrograph! Even conventional spectrographs can have surprising 
problems if the light seal is poor between the dewar and camera 
as is often the case.  Wrapping the CCD dewar and spectrograph 
interface in black cloth fails if the light source is inside! 
Glowing fibers sometimes pass by small cracks in the spectrograph
casing and thus intermittently cause a surprising increase in 
dark current.

In the lower right corner of this image you can also see a 
mysterious light source behind the power cable's socket on the
front panel.  I heartily recommend using a slow scan CCD 
camera as a means of searching for undesirable light sources
in critical applications.

I have also heard tales of things like luminous paint in bench 
spectrograph rooms.  A really dark room is not trivial hard to
achieve.


REMEDY:

We have found simple ways of dealing with fiber light leakage:

- Use only fibers with black cladding.  All other colours
  leak.  Tom Ingerson (who did the follow up work on this
  after I took the initial images) found that beige, orange,
  grey and brown plastic cladding are all transparent at
  the operating wavelength.

- Install bolts in the threaded holes in the base of the
  transmitter and receiver modules.  Lots of light eascapes here.

- Paint plastic surfaces of transmitter/receivers black.

- Use a rubber gasket at the front panel feedthrough.

- Slide a black plastic/rubber sleeve down over the bayonet
  connector once it is mated.  A small rubber O ring serves 
  nicely as a retainer: slide it up the conical strain relief
  on the back of the connector to trap the black sleeve over the
  connector.

- Use a low dark current CCD camera to check your work.




2) Glowing windows:
   ----------------

  For a long time we obtained disturbingly high dark currents
  for one of our Tek 2K's.  This was our lowest noise CCD and
  thus the one we wanted to use for photon starved observers
  ...of course.  The dark current had several puzzling 
  characteristics it reached an abrupt floor when one varied
  temperature or operating volatges. It also suffered from 
  extremely high level, and slow decay after installation 
  on the telscope (not seen in the lab) and minimum dark current 
  was unusually high, about 8-9 e-/pixel/hours (MPP clocking, 
  after 3 days in the dark).  Was this from over-illuminating 
  the CCD?  That shouldn't be a problem for MPP mode.  Lab tests 
  showed that power cycling wasn't the culprit.

  Alistair Walker and Jorge Bravo recently confirmed a suspicion 
  we had for some time that the **dewar window** would glow for 
  many hours (days even) after being exposed to light.

  Alistair purchased a new window (UV grade fused silica) which we are
  pleased to find does not suffer from this problem or certainly not
  to the same extent.  Even more importantly, the final level of dark 
  current level has been reduced from 9 to 2 e-/pixel/hour.  Pixels 
  are 24 um square.

  LESSON:   When making critical dark current measurements, try 
            replacing your window with a metallic blank, or putting
            a cover directly on the CCD.

	    A dark slide incorporated into the dewar on the vacuum side
            would be very nice!


My thanks go to Tom Ingerson for investigating the fiber light leak 
extensively and for putting my image of the Arcon fibers in the Mosaic 
data base.


Roger Smith,  CTIO