Correcting for Non-Linear Crosstalk - Update
Table of Contents
This page summarizes the current status of DECam crosstalk. Please see the links about previous results for some more information.
DECam crosstalk cannot be fully described by a linear model - non-linearities occuring are described by a polynomial. Intra CCD crosstalk typically is significantly larger than inter CCD crosstalk - but there are exceptions.
See below for details.
previous results
first results, negative crosstalk and first hint of non-linearities
crosstalk was determined by measuring the coefficients on single exposures and then averaging over those coefficients
closer look at non-linearities (with examples)
- crosstalk was determined by using multiple exposures and extracting coefficients from a fit to the binned average (of src-vic counts)
- checking for crosstalk on same DA, and for some inter DA (very low to no crosstalk), much cleaner results http://www.usm.uni-muenchen.de/people/paech/xtalk_mult.html
- more detailed investigation of non-linearities http://www.usm.uni-muenchen.de/people/paech/xtalk_nonlin.html, trying some parametrizations, but crosstalk coefficients provided still linear
current status
characterization
- for intra ccd crosstalk non-linearities occur above ~45 000 counts, but onset is different for different CCDs
- for inter ccd crosstalk (on the same DA board), no non-linearities have been detected
- no inter DA crosstalk has been detected so far
- crosstalk for super-saturated stars with edge bleed can not be predicted completely
parametrization
The parametrization of the crosstalk should sufficiently discribe the signal and still be quick to apply. Therefore the parametrization is split into a linear regime where the classic linear correction is applied and a non-linear regime where a higher order polynomial is used to describe the crosstalk (which is more compute intensive)
- estimate the onset of the non-linearities (src_nl)
- xtalk_lin(src_counts) = coeff * src_counts for src_counts <= src_nl
- xtalk_nonlin(src_counts) = polynomial(src_counts,degree=n)
- continuity is ensured by condition xtalk_lin(src_counts) = xtalk_nonlin(src_counts)
- a 3rd order polynomial seems to be sufficient
values
The most recent crosstalk coefficients can be found here It contains the following information:
- victim amplifier
- source amplifier
- linear crosstalk coefficient x
- uncertainty of linear crosstalk coefficient
- source count of onset of non-linearities src_nl
- p_1 to p_n coefficients for the polynomial
For src <= src_nl:
- crosstalk = src * x
For src > src_nl:
- crosstalk = src_nl * x + p1 * (src - src_nl) + p2 * (src - src_nl)**2 + p3 * (src - src_nl)**3
The following image shows the magnitude of the linear part of the crosstalk coefficients. Typically, inter-ccd
crosstalk is small, but there are a few exceptions.
