[Aavso-photometry] CCD Linearity testing

[Tom Krajci]

From: arne <arne at aavso.org>

>Tom is doing linearity tests, not flatfielding.  For linearity, you
>don't worry so much about flatness or QE peaks - what you are looking
>for are changes with respect to total exposure on each pixel.  LEDs
>are fine for this, with the possible exception being Residual Bulk Image,
>an effect that increases in strength with red light.  So you may get a
>different result with, say, red LEDs, than with white LEDs.  

I'm currently using a green LED.

Does this mean I should get a Blue, Red, and IR LED for more complete testing?

>Also, with
>linearity tests, you generally don't need to have the telescope involved
>at all - you can take the camera to the lab and work with it there.  This
>keeps the temperature constant, and you work with completely defocussed
>light, since there are no optics in the path.

Yes, bench testing is easier and more accurate...LED, and CCD...and not much else.

>One trick to guard against a systematic buildup of residual light is to
>increase exposure until you get total saturation, and then ramp down
>and see if there is any systematic difference between increasing exposures
>and decreasing exposures.  I usually run several sets of "ramp-up"
>and "ramp-down" cycles.

After Richard Miles pointed out that the middle of my run appeared to be 
the when the LED was most stable, I looked at that data more closely.  I 
did use a ramp-up/down exposure sequence, and at about the 0.1% level 
you can see a difference in the ramp up versus ramp down exposures...for 
the fainter levels, and at the faintest levels this affect appears 
greater, but the measurements also have greater scatter.

Also, I find that if I add about 0.0055 seconds to the stated exposure 
time...the curve flattens out at the faintest levels (to better match 
the rest of the curve/slope at medium ADU levels).  I can't prove that I 
have a small amount of shutter error, but I've seen this before on some 
CCD's I've evaluated, to include my Cookbook CCD...and for that one I 
know that it did have a shutter error (the fault of one version of the 
camera control software, not hardware).

Even with my crude/hasty test set up, it's surprising how this is 
accurate enough to show such subtle effects in the CCD's performance.

-- 
-------------------------------------------
Tom Krajci
Cloudcroft, New Mexico
http://picasaweb.google.com/tom.krajci

Center for Backyard Astrophysics (CBA)
http://cbastro.org/ CBA New Mexico

American Association of Variable Star
Observers (AAVSO): KTC http://www.aavso.org/
-------------------------------------------