[KIDS] Crosstalk on CCD96 exceeds bias level

[Massimo Dall'Ora]

Yes, you are absolutely right, the bias level has an intrinsic noise, due to 
the shot noise variations of the electronics, which does not follow Poisson, 
but it depends on both read out noise and gain.

max


On Thursday 19 April 2012 11:15:46 Konrad Kuijken wrote:
> Indeed, no option now but to mask the pixels where the information is lost.
> I did not see this behaviour in the commissioning data I looked at last
> year, the signal remained above zero.
> 
> Regarding the effect of changing the bias level: indeed, a higher bias level
> reduces the level of saturation. But I do not believe bias level gives
> Poisson noise, it is purely a dial in the video chain I think, designed to
> hit the most linear part of the amplifier response. It is not clear to me
> that we have much freedom in setting the bias level - all CCDs talk to each
> other and I do not think Olaf can control them individually.
> 
> How are you doing the correction? Previously I showed that also unsaturated
> pixels can cause electronic ghosts (see commissioning reports).
> 
> K
> 
> On Apr 19, 2012, at 8:42, Massimo Dall'Ora wrote:
> > Hi Ewout,
> > 
> > interesting plot....
> > 
> >> From the stellar photometry point of view (that is my point of view),
> >> to
> > 
> > increase the bias level has no scientific effects, apart of to slightly
> > reduce the dynamical range, and to increase the noise of the
> > measurements at high magnitudes.
> > 
> > Probably some issue could arise for the extended sources, since an
> > increase of the bias level of ~600 counts introduces an additional
> > noise of  ~25 counts, that could mask faint features.
> > 
> > Regards
> > massimo
> > 
> > On Wednesday 18 April 2012 12:22:30 Ewout M. Helmich wrote:
> >> Hi everyone,
> >> 
> >> We're investigating correcting the crosstalk of OmegaCAM. One issue
> >> that
> >> may require action on Paranal is that the crosstalk on CCD#96 (upto
> >> -600
> >> ADU from CCD#95) exceeds the bias level of ~400 ADU in CCD#96. See the
> >> attached plot. Pixels exceeding ~50000ADU in CCD#95 will put that
> >> pixel
> >> to 0 in CCD#96, destroying the information there (this in the absense
> >> of
> >> sky background, the background in the attached example is about 100
> >> ADU,
> >> in g'). I see two ways to address this:
> >> 
> >> 1) We can mask pixels with value 0 in CCD #96 (will have to for
> >> existing
> >> data).
> >> 2) The bias level of CCD #96 can be increased by e.g. 300-600 ADU.
> >> 
> >> The question is if there are objections from a scientific point of
> >> view
> >> to this bias level increase.
> >> 
> >> Regards,
> >> Ewout
> > 
> > _______________________________________________
> > KiDS mailing list
> > KiDS at astro-wise.org
> > http://listman.astro-wise.org/mailman/listinfo/kids