Scaling darks or not

[Emanuele]

Hi guys,

Was wondering what would be better: building a dark library with set of the most commonly used exposures or just shoot 1200sec darks (30 of them) and then let PI do the scaling?
Thanks,
E.

[Niall Saunders]

Hi Emanuele,

OK - here is my 'thought experiment for the day'

Ensure that your CCD is temperature stabilised
Shoot five darks at X seconds exposure time - however, from this point on you will treat these as 'Lights'
Shoot an appropriate number of Biases
Shoot an appropriate number (say ten, each) of Darks at X seconds, (2 x X) seconds, (3 x X) seconds (etc.) - you could even try with (0.5 x X) seconds as well

Now, use PI to calibrate the original set of 'Lights' - do this for each differing source of Darks. Then perform a simple ImageIntegration of the results (no pixel rejection, etc. - just 'Averaging')

In an 'ideal' world - when you calibrate your lights with the Darks taken at the same exposure, you should (in theory) end up with a master frame that is effectively TOTALLY DARK (or at least with all pixels having the same ADU value, where this value 'represents total dark' for your CCD)

When you calibrate your Lights with the other Darks, you 'should' end up with the same result. Providing, of course, that the Dark Optimisation routine was able to correctly scale your 'exposure-time-mismatched' Darks.

This is not a difficult experiment to carry out - and is ideally suited to 'down-time' when the weather is not playing fair.

Let us know how you get on (just remember that the 'Lights' accumulated 'in the dark' must obviously be totally different from any 'real Darks' acquired for the same exposure time - this is a requirement for a valid statistical analysis)

Cheers,

[jmtanous]

Niall,

Excellent suggestion... I'll give it a try!

Thanks,

Jose

[NKV]

Hi Emanuele,
IMHO: If your CCD capable to see cosmic ray, you need to increase quantity of dark according exposure time.
In other word: too much cosmic ray require more dark frames, otherwise clipping algorithm will allow to pass the cosmic ray.

[Emanuele]

Thank you to all of you guys

I should have specified my setup, and I will update my signature with it.
I have an FLI ML8300.

Yes, I do see cosmic rays hits NKV, actually many of them, in a 15m dark.

[vicent_peris]

HI,

Acquire a lot of biases, they are cheap.


V.

[Emanuele]

HI,

Acquire a lot of biases, they are cheap.


V.

Vincent:
By saying this you are saying that you just use biases and don't use darks?

[Jack Harvey]

He doesn't use darks - his camera is liquid nitrogen cooled to -180C<G>.

[mmirot]

Hi Emanuele,
IMHO: If your CCD capable to see cosmic ray, you need to increase quantity of dark according exposure time.
In other word: too much cosmic ray require more dark frames, otherwise clipping algorithm will allow to pass the cosmic ray.

Good Point. My FLI KAF 16803 gets hit a lot. I generally use 25  darks of the same exposures. Perhaps I should expand my set.
 If it were not for the extra cosmic rays I could just scale one very long exposure.

Max

[Emanuele]

He doesn't use darks - his camera is liquid nitrogen cooled to -180C<G>.

Oh !  alright, well, wish I could have the same camera!

[vicent_peris]

No. Use a lot of biases to have a master bias without read noise. If your camera has 13 e- read noise, and if read noise was poissonian, just acquire 13^2=169 bias to reduce read noise to 1 e-. The same with darks... Although they are not so cheap as bias...

By reducing read noise in bias, dark frame scaling will be more precise.

Our camera is not at -180C, only -120C. :-P

Regards,
V.

HI,

Acquire a lot of biases, they are cheap.


V.

Vincent:
By saying this you are saying that you just use biases and don't use darks?

[NKV]

just acquire 13^2=169 bias to reduce read noise to 1 e-.

Wow! Now I know how much need frames for good result.
Vicent, thanks for the formula.

[vicent_peris]

Our dark frame scaling algorithm los for a minimum noise in the light frame. As you subtract the bias frame to the dark frame, you are transferring noise from the bias to the dark, resulting in a noisier dark. So, when looking for a minimum noise in the light frame, our algorithm always undercorrects dark signal because formerly we habe added to it the readout noise.

The corajes bias frame, the better dark signal supression we'll have.


V.

just acquire 13^2=169 bias to reduce read noise to 1 e-.

Wow! Now I know how much need frames for good result.
Vicent, thanks for the formula.

[Niall Saunders]

Hi Vincent,

Our camera is not at -180C, only -120C

If it would help cool your camera down any further, I could send you some boxes of cold weather from up here in Scotland - we have plenty to spare. Just handle them carefully, they are also full of cloud, so you don't want to have any leaks

Cheers,

[Emanuele]

Vincent, thank you for the formula!
My ML8300 has 8 -e read noise so I shouod be good with 64 biases. I have took 100 biases just to round up
The thing is that biases are only good for scaling darks, because otherwise darks already contain bias.


E.