STT-3200 characterization

Discussion in 'Legacy Models - Community Support' started by Jason, Apr 6, 2019.

Tags:
  1. Jason

    Jason Cyanogen Customer

    Joined:
    Oct 11, 2016
    Messages:
    30
    Location:
    Long Beach, CA
    I recently set up a permanent setup, which means I've also found time to finally characterize my camera. Qualitatively, my raw frames look very good; so much so that dark subtraction and flat fielding can have little effect.

    So, I was pleasantly surprised to discover just how quiet this particular camera is. The read noise is 6.2 e- rms and the dark current rate at -30°C is 2.6 e-/pix/hour(!) with an excellent doubling rate of 4.80°C (meaning, a smaller temperature drop than the usual 6°C rule of thumb will result in a halving of the dark current).

    I've posted a 1-hour image of M 101 in a Baader B filter over in the astrophotos thread.
     

    Attached Files:

  2. Colin Haig

    Colin Haig Staff Member

    Joined:
    Oct 27, 2014
    Messages:
    7,378
    Location:
    Earth
    Jason - These are great! Thanks for sharing this with us.

    Am curious -what tool(s) did you use to produce the charts?

    The 3200 is one of the best chips they make, and they do a great job in our cameras - we try to do our best to get the most out of them.
     
    Jason likes this.
  3. Jason

    Jason Cyanogen Customer

    Joined:
    Oct 11, 2016
    Messages:
    30
    Location:
    Long Beach, CA
    I followed the prescription in Photon Transfer: DN -> lambda by Jim Janesick and explained online by Richard Crisp and others. My only addition to the process is to take n frames at each illumination level, yielding n (n -1) / 2 frame pairs, in order to improve statistical confidence. Bias overscan columns are important for separating photon noise from camera noise. I wrote a small C program using the CFITSIO library to go through the images and perform the analysis and write out a huge table of numbers. The table is loaded into Excel and the PTC is then plotted.

    The dark current data was done manually with five frames of 120 sec at the five temperatures, with plotting and fitting in Mathematica. I was surprised enough that I took five 1-hour darks at -30°C and confirmed 6 DN/pix/hour in each. There are a few isolated pixels that saturate due to dark current alone, but meet spec and calibrate out.

    I’ve just finally had a night of good seeing and the results with this camera are pretty incredible, due to high QE and good uniformity. This is 9x480 sec at -30°C with with a 5” f/7 refractor and only dark subtraction, stacking, and stretching. There’s dust lane detail down into the core.

    I’m usually critically sampled at 1.53”/pix, but that is intentional given my usual seeing conditions (in this image, the seeing was good enough to make me a touch undersampled). I am considering putting this camera behind a telescope with ~50% more f.l. and almost twice the aperture later this year (still debating about going to a large format camera for that telescope, but 1”/pixel would be an ideal scale for pretty pictures; the pixel etendue would only increase by 1.2x but with almost 2x better resolution).
     

    Attached Files:

    Last edited: Apr 9, 2019
    mike m benjamin likes this.

Share This Page