View Single Post
  #4  
Old 27-09-2016, 09:59 AM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
 
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,918
Hi Ken. A few comments that may be of some use - or possibly not.

the ripple shown in the white paper looks ghastly, but is due solely to the use of a single angle of illumination. With real optics and a cone of illumination, there will still be some ripple, but it will be the average of that at many different angles and far less pronounced than that shown in the article. In any event, your optical geometry is fixed, so the effect can be flat-calibrated out in principle. I am not clear on how you process the data, but if nothing moves between subframes, you should be able to apply a flat to the final stack, rather than the individual AVI subframes.

However, I doubt that is what you are seeing. CMOS chips have a split conversion/amplification mechanism that operates both at pixel level and chip level. The pixel-level converters can all have slightly different bias levels, which gives the banding typical of older CMOS chips and similar to what you show. The chip makers are not forthcoming on details, but apparently some modern CMOS chips have a mechanism that adjusts the pixel level biases to remove banding - but that never settles to exactly the same place on each power up, so bias data is not fixed from session to session. Suggest that you try applying a bias subtraction to some data and see if that tidies up the banding. If the 174 is self-calibrating, you may find that the bias data will have to be taken on the same power-up as the lights (actually only a minor inconvenience).

Overall though, because CCD chips have only one amplifier and stable bias, they are probably a little easier to use than CMOS for spectrometry. However, the extremely low read noise of CMOS (which is due to the split amplification) must make it worth persevering with CMOS.

Last edited by Shiraz; 27-09-2016 at 10:16 AM.
Reply With Quote