Page 1 of 1

ASI1600MM gain puzzel

Posted: Mon Jun 04, 2018 7:00 pm
by Andrew Smith
Prior to using my ASI1600MM for real I thought I would do some tests. I set the gain at 200 and offset at 50. I then did bias darks and flats at -15c and ran the detector noise test in ISIS V5.8.0 this gave the following:
Gain = 0.031 e/ADU
RON = 1.32 e
Offset 597.7 ADU
Dark 0.0086 e/s
The gain is an order of magnitude lower than the value in Christian's review 0.475 and on the FLO website.
I then repeated the calculation getting the values by hand with the same result.

I set the gain and offset via the ascom driver in The Sky X.

Any thought on what might be going on or how to investigate further?

Regards Andrew

Re: ASI1600MM gain puzzel

Posted: Tue Jun 05, 2018 5:29 pm
by Robin Leadbeater
Hi Andrew,

Can you check the ADU against that for another (eg CCD) camera of known gain using the same light source and exposure? An order of magnitude difference in gain should be easy to spot if it is real. If it is not real then that suggests to me there is something odd about the photon statistics ie they are not stochastic. There isn't a noise reduction feature switched on somewhere in the camera is there?

EDIT: Actually a noise reduction function would have the opposite effect. A false low calculated inverse gain would mean the noise in the difference between the pair of exposures is higher than it should be

Cheers
Robin

Re: ASI1600MM gain puzzel

Posted: Tue Jun 05, 2018 6:01 pm
by Robin Leadbeater
Also, what does the difference between the two flat images look like? (ie the histogram should be a normal distribution. There aren't any outliers like a cosmic ray hit or a change in amp glow or something similar biasing the statistics ?)

Robin

Re: ASI1600MM gain puzzel

Posted: Tue Jun 05, 2018 7:10 pm
by Andrew Smith
Thanks Robin, I could try a different CCD but I am reluctant to strip down my MRes spectrograph to try it on the HRes but as a last resort I might try it.

I have looked at the statistics of the various frames including the flat and dark difference but the latter look Gaussian on casual inspection. I tried with and without applying a bad-pixel map with the same results. There were no obvious cosmic ray hits or other issues. I cropped the image to a region well away from any amp glow.

I am not aware of any noise reduction features. I am going to try using the ZWO capture software ASICAP and compare it with The Sky X and see what happens.

Regards Andrew

Re: ASI1600MM gain puzzel

Posted: Tue Jun 05, 2018 11:29 pm
by Christian Buil
Andrew, I multiply all my SWO cameras raw images by the coefficient 1/16 = 0,0625 before gain analysis. The reason : the output of SWO camera are artificially 16-bits coded. The true dynamic is 12 bits. The change from 16 -> 12 correspond to the 0.0625 multiplication. Perhaps the origin of the problem (0.475 / 0.031 ratio is near 16) ?

Christian

Re: ASI1600MM gain puzzel

Posted: Wed Jun 06, 2018 10:11 am
by Andrew Smith
Puzzle solved - thanks Christian.

Re: ASI1600MM gain puzzel

Posted: Wed Jun 06, 2018 10:55 am
by Robin Leadbeater
Interesting. I wonder if photometrists using these cameras take this into account when calculating their uncertainties.

Robin

Re: ASI1600MM gain puzzel

Posted: Wed Jun 06, 2018 11:42 am
by Andrew Smith
Robin Leadbeater wrote:Interesting. I wonder if photometrists using these cameras take this into account when calculating their uncertainties.

Robin
Hi Robin, as long as you convert consistently to electrons I don't think it is an issue. If you leave the data in 16 bit and use the 16 bit gain you will get the same answer as converting back to 12 bits level by dividing by 16( 2^16/2^12) and using the 12 bit gain.

Regards Andrew
PS I see what you mean they could use the published 12 bit gain with the 16 bit ADU