-200e/ADU Gain

[Zeranoe]

I'm trying to get the gain of my Canon 5D but the Basic CCD Parameters script is giving some strange results.

I took the bias, flat, and dark frames following this guide.

I've attached the results for the script.

[mschuster]

Check out FlatSNREstimator script. You need two flats and one bias (or one flat dark instead of a bias). Raw, undebayered frames. This should give an average gain across the color channels.

You may also use SplitCFA script to split all frames into separate channels, and run the script on each channel separately. You will typically see differing signal, noise, and SNR's, but nearly equal gains. See the FlatSNREstimator script documentation for an example, section 3.2.

Regards,
Mike

PS: Double check that the flats are well exposed. The script won't work for extremely underexposed or overexposed/saturated/clipped flats.

[Zeranoe]

I tried the FlatSNREstimator and got a gain of 0:

[mschuster]

Flats may be exposed incorrectly.

Please zip two flats and bias into DropBox if possible. I will check them out.

Thanks,
Mike

[Zeranoe]

The same two flat and the bias frame I used in the FlatSNREstimator can be downloaded here.

Edit: First link only contained flats, new archive includes flats and bias.

[mschuster]

The flats appear overexposed/saturated, the script can't handle this.

The FITS header says exposure was 1 second.  I think at ISO 1600 something a lot shorter is needed to avoid saturation.

Regards,
Mike

EDIT: Bias looks ok

[Zeranoe]

I'll retake some that are less saturated. Is there a way to know if they are over/under saturated, like perhaps a visual indicator on the histogram?

[mschuster]

A histo like this one. A clear "hill" with no clipping on the tails.

[Zeranoe]

The gain looks much better with a new flat:




The new flat looks better, but I still don't know if it's over or under exposed.

[mschuster]

The average signal across all channels is ~700DN, which seems low. Typically I like to see 50% of saturation/maximum, which might be 50% of 16k or 64k or something else on your camera, I am not sure.

For purpose of master flat correction, a master flat SNR of several hundred or more is good to have. Your single flat has SNR 42. You'd need 25 combined in a master to get SNR ~200, eg 42 * sqrt(25) = 210. Having a flat with more exposure but nowhere near saturation would have higher SNR, so fewer would be needed to be combined in a master.

EDIT: because of the different exposures in the CFA channels, it is better to be conservative and underexpose a little just so that all channels are not near saturation.