• Mapping wavelength\grating\position on camera
  • add term to fit function so
  • plot histogram of residuals instead of 3d plot, where
  • histogram should follow Gaussian with , , is if all point lie above fit plane etc
  • with , write down as well so can get per degrees of freedom.
  • do a linear fit first, then add terms and see how varies.
  • could also plot data points for each grating position with fit function
  • put all lamps onto same plot
• Scan of peak across CCD
  • Have the scan plot as parametric function
  • then can see intrumental response
  • similtaneous fit with r of circle fit
  • fit with parabola
• Errors
  • Take lots of 0.5s frames and look at the spread of the data
  • normalise to the same area under curves, then see how individual point varies
  • to normalise, divide each point by bin width and the total number of points in the distribution
  • see how the varies with exposure time
  • derive some error bar from this and compare it to the poisson error found before
  • may follow students-t distribution not Gaussian
• Fit to line
  • for background fit quadratic that is not to large so affects Gaussian
  • add some skew to the Gaussian
  • see if the skew is the same on both sides of the chip
  • look up how to skew the Gaussian
  • odd/even functions in the
  • could use Gaussian as a function of where:
    • if then its a normal Gaussian
    • if or other function then can add skew
• Look at fourier transform of the ccd pixel readout

-- JosephBayley - 19 Nov 2015