Linearity of CCD output amplifiers

[Hopkinson, Gordon]

Posted to CCD-world:
-+-+-+-
Hello CCD world

This question opens up a bunch of issues.  Basically the output amplifier
should be good to 1%, probably 0.1% over most of the range.  We have
measured linearity that good on some of our programmes.  There will be a
non-linearity in the MOSFET response, but it should be small (it is more
important when binning pixels together which I assume you are not doing).

A 5% value will not be due to the amplifier.  It may be due to the off-chip
electronics (lots of things can happen there) it may also be due to a low
full well capacity.  We tend to define full well as the point at which
non-linearity drops to some percentage (usually 1% to 3% depending on the
project).  So, in effect the your 5% value defines a low full-well.

Low full well can be due to clock driver design.  It sometimes happens that
on largish chips the drivers don't have enough capacity to drive the area
clock capacitance.  In that case you see a lower full-well in the centre of
the image than at the edges.

Also measured full-well tends to be lower if you use spot illumination (e.g.
stars) rather than uniform illumination.  This is due to CTE effects.

Most likely your effect is in the off-chip electronics.

Best regards

Gordon Hopkinson

Sira Electro-Optics Ltd

> ----------
> From: 	Michael C. B. Ashley[SMTP:mcba at phys.unsw.edu.au]
> Reply To: 	CCD-world at cfht.hawaii.edu
> Sent: 	30 April 1999 02:30
> To: 	CCD-world at cfht.hawaii.edu
> Subject: 	Linearity of CCD output amplifiers
> 
> Posted to CCD-world:
> -+-+-+-
> Dear CCD gurus,
> 
> I am interested in knowing more about the non-linear behaviour of CCD
> output amplifiers. I have made measurements of the performance of an EEV
> CCD05-20 (770x1152 22.5 micron pixels), and have found that it is
> non-linear at the 0.5% level below one-quarter of full-well capacity, but
> deteriorates to 5% above this. Interestingly, the non-linearity can be
> well corrected using a 2nd-order polynomial below one-quarter of
> full-well, and a 3rd-order polynomial above this. There appears to be a
> distinct "break" in the characteristics at around one-quarter full-well,
> and I have been unable to find a single function which will represents the
> entire range. The position of the break appears to vary slightly (with
> temperature?). 
> 
> After correcting for the non-linearity, it remains stable to better than
> 0.1% from night-to-night, and about 0.2% over several years.  I measured
> the linearity by using precision-timed exposures of a uniformly
> illuminated screen, checking carefully for changes in illumination with
> time. I guess there is the possibility of non-linearity in the remaining
> analog signal chain, but I doubt it (this is with a Wright Instrument's
> camera, which has given excellent performance).
>  
> Has anyone observed similar effects?  I am wondering if a knowledge of FET
> characteristics would predict the behaviour I see, and perhaps suggest a
> more suitable functional form for the correction, and reasons for the
> variation in the break point. 
> 
> For those interested in the polynomials I found, here is some C code
> to correct the raw ADU number "a". Full-well corresponds to 40000 counts
> (we use a very low gain so that we can sample the full range of the CCD).
> 
>     if (a < 12000) {
>       a =  a*(1-6.8e-7*a);
>     } else {
>       a = 806+a*(0.8586+a*(5.90e-6-a*3.12e-11));
>     }
> 
> Regards,
> Michael
> --
> Michael Ashley; Department of Astrophysics, University of NSW;
> For further information:   "finger mcba at newt.phys.unsw.edu.au"
> 
> -+-+-+-
> For information about CCD-world, send email to
> owner-CCD-world at cfht.hawaii.edu.
> 
-+-+-+-
For information about CCD-world, send email to owner-CCD-world at cfht.hawaii.edu.