[Aavso-photometry] CCD Linearity testing

Thu Jul 9 16:24:07 EDT 2009

Hi Tom

Beside doing "ramp-up" followed by a "ramp-down" cycle, one can
also do a "ramp-up" and bracketing every "linearity" exposure by a
mid-long fixed length exposure. Doing so makes it easy to monitor
the actual output of the source. If a change in output is present, one
can scale the "linearity" exposures by the mean of the neighbouring
fixed length exposures. This should also get rid of a part of the influ-
ences if RSI or RBI is present at all. But to avoid RBI its always best
to use blue to at most green light as it doesn't penetrate the CCD
deep enough. An other option to get rid of a changing RBI effect
would be to fill the traps before every exposure by overexposing the
CCD with a indesicant lamp that is not heat-blocked. RSi is pro-
bably not an issue as most CCDs don't show it and its effect is usu-
ally pretty smooth over hours anyway.

If someone is using a halogen lamp or an indesicant bulb for this, its
a good idea to let it burn for some time (up to an hour) until it reached
a thermal equilibrium to get a more constant output before starting
the measurements.
Doing such measurements not when someone is switching on and
off electrical devices in the house and not around early morning,
noon, or early evening when the electrical network is often heavily
loaded should also make the output more constant.
But using a current and temperature stabilised LED or even a pulse
count modulated LED is probably always the best way. Changing the
number of pulses should be better than changing the pulse width
as long as the pulses don't get too close together to influence each
other.

Many will do the linearity test exposures with an attached optics
like e.g. a camera lens. But this will usually introduce severe vig-
netting and therefore an uneven illumination over the chip which
is of disadvantage if one wants to evaluate all pixel of the chip.
But there are several other setup that might give a more even illu-
mination of the CCD chip:
One is to add a stack of white paper directly in front of the camera.
This method has the advantage that one can control the illumination
also by the number of papers in the stack:
http://www.ccd.com/images/ccd105j.gif
If one has a lightbox with a diffusor, one can use it without an op-
tics directly in front of the camera. For a cubic light boxdesign the
diffusor should be pretty evenly illuminated. At ING (where the cubic
lightbox design might originate from) they use it even without a dif-
fusor for CCD calibration measurements.
The savest is probably to use a relativ long tube with a pinhole at
the top end that is well blackened inside and point it to an uniform,
plain, white surface or lightbox. The shading due to the tube avoids
changes in stray light from the room and its blackening minimizes
internal reflections off the chip and back agin.

The most simple way to handle general non-linearities is to imple-
ment a global look-up table. Many pros do it this way.
In other cases the "curve" shows a piece wise linear behaviour as
e.g. for the SDSS CCDs that is probably caused by the amplifiers
during read out. If a chip shows this behaviour one could implement
also a piecewise linear polygon instead of a LUT.
If one wants to do it as perfect as possible and as every pixel has
its own linearity curve that is slightly different from the others (espe-
cially when the response starts to get non-linear), one would have
to do it on the pixel and not on the entire CCD level. Hot to very
warm, and cold pixel should do probably deviate from the mean
the most not just in regards of dark current but probably also in re-
gards of QE and non-linearity. For these its also most probable
that they don't behave well with scaled darks. So an other approach
would be to treat the bulk of well behaving pixel the same and do a
special processing for just the most deviant ones.
For an example of the variations of these linearity curves see e.g.:
http://www.jca.umbc.edu/telescope/Glossary/figs/non-linear_point.gif
This plot makes clear that if one wants to extend the linear range
into the non-linear range its proably best to do it on a pixel level
as every pixel has its own saturation level it converges to. This plot
also explains why one does see often see a pattern of vertical
darker stripes in overexposed flats (the pixel with a less than usual
full well and saturation level will leave darker read-out tails in the
overexposed flats).

Tom, do you have your own old pages somewhere else now ?
ST-7 CCD A non-linear detector ... and hopefully a way to make
it linear
http://overton2.tamu.edu/aset/krajci/st-7/st-7.htm
Repaired ST-7E CCD Bench and field testing show that it's now
ready for photometry
http://overton2.tamu.edu/aset/krajci/st-7-new/st-7-new.htm
Test your CCD camera for proper performance (gain, ...)
http://overton2.tamu.edu/aset/krajci/st-7/test-procedures.htm

You probably also know the following ones:
Tim Abbott CTIO - In situ CCD testing
http://www.ctio.noao.edu/~tmca/CCD/docs/cookbook/top.html
Chapter on Linearity
http://www.ctio.noao.edu/~tmca/CCD/docs/cookbook/node11.html

ING - CCD Quality Control
http://www.ing.iac.es/%7Eeng/detectors/QC/qc.htm
ING - La Palma Technical Notes
http://www.ing.iac.es/Astronomy/observing/manuals/man_tn.html
No 111 Design and use of a light source for linearity measurement
in CCD-cameras
http://www.ing.iac.es/Astronomy/observing/manuals/ps/tech_notes/tn111.pdf
No 108 Design and use of a novel flat field illumination light source
http://www.ing.iac.es/Astronomy/observing/manuals/ps/tech_notes/tn107.pdf
No 107 Design and use of LED light sources for QE calibration
http://www.ing.iac.es/Astronomy/observing/manuals/ps/tech_notes/tn108.pdf

UMBC-JCA - Users Guides
http://www.jca.umbc.edu/telescope/UsersGuides/UsersGuides.shtml
Dynamic Range & linear response of the ST-8E
http://www.jca.umbc.edu/telescope/Glossary/CCD_SBIG_ST-8E.html#Range
http://www.jca.umbc.edu/telescope/Glossary/figs/non-linear_point.gif
http://www.jca.umbc.edu/telescope/Glossary/figs/dIdt_vs_I.gif
Temperature dependant bias and dark frame characteristics of the ST-8E
http://www.jca.umbc.edu/telescope/Glossary/Dark.html
http://www.jca.umbc.edu/telescope/figs/DarkVsExp.gif
http://www.jca.umbc.edu/telescope/figs/RateVsT.gif
http://www.jca.umbc.edu/telescope/figs/BiasVsT.gif

mirametrics - Mira Tech Note
http://www.mirametrics.com/library.htm
Pixel Response Effects on CCD Camera Gain Calibration
http://www.mirametrics.com/tech_note_ccdgain.htm
Centroiding Noise and Precision Astronomical Photometry
http://www.mirametrics.com/tech_note_centroidnoise.htm
Correcting CCD Bias using the Overscan Method
http://www.mirametrics.com/tech_note_overscan_bias.htm

Apogee - CCD University - CCD Test Methods
http://www.ccd.com/ccd105.html

SDSS DR7 - Imager CCD nonlinearity
http://www.sdss.org/dr7/instruments/imager/nonlinearity.html

Clear skies
 Wolfgang

-- 
Wolfgang Renz, Karlsruhe, Germany