Atik 383l Darks\Flats\Bias

[AlistairW]

Hello,

Got a few question on my process - hoping someone can help :-)

> Darks. These should be simple enough, - same duration as the lights and same temp. I can do these whenever.

> Flats. I use SGP capture software and I have an adjustable illuminated panel. I set the panel to some random illumination that then takes subs of 11 seconds to achieve a 23000 ADU. Is 11 seconds too long ? - Indeed does it mater how long ? I can make them shorter, simply by upping the brightness of the panel. (Flats need to be done with the original image train position I think).

> Bias. Is this simply setting a very short time on the image duration ? - No need to worry about temp, or image train. The Atik has a shutter so I think the limiting 'speed' would be 200ms to avoid shutter shadow. Again I think this can be done with the Atik lens cap on.

I am a bit new to CCD to thought it best to ask :-)

Thanks

Alistair

 

 

 

 

[Uplooker]

I can not find anything wrong with your logic, other than to reduce the length of your flats for the same ADU. It's the same as what I do with a QHY9 - same sensor.

[Uranium235]

Try going without darks, I've never used them with my 383. As long as you have enough subs, your stacking should take care of the hot pixels (sigma clipping).

Also, take your flats at the same temp as your lights - otherwise the ADU readings will be inaccurate.

[carastro]

Quote

Also, take your flats at the same temp as your lights - otherwise the ADU readings will be inaccurate.

Flats are not supposed to be temperature dependent, I have never heard that before.  Where did you get that info Rob? 

Quote

> Bias. Is this simply setting a very short time on the image duration ? - No need to worry about temp, or image train. The Atik has a shutter so I think the limiting 'speed' would be 200ms to avoid shutter shadow. Again I think this can be done with the Atik lens cap on.

Bias won't be affected by shutter shadow as the lens cap will be on.  Just take shortest time possible with lens cap on. 

Your flats are a bit long, but if it suits you to take them this long should be no problem so long as the ADU is right.  Personally I'd dim the light source down a bit and take shorter flats, but as you already know they need to be long enough to avoid shutter shadow.

I found my Atik383L very noisy with loads of hot pixels, so I found it necessary to take darks with that camera, but then I had not learnt about Sigma clipping then, so that might make a difference.

Carole 

[AlistairW]

Thanks for the replies.

I will certainly try without darks. Flats I did think were temp independent too - but no problem to do these at the same temp, as they are done after the imaging run anyway.

Rob, given your suggestion on using Sigma Clipping, do you bother to dither ? - And when you say Sigma Clipping is that simply the setting for Light\ Darks in DSS ?

Thanks

Alistair

[ollypenrice]

If your flats take too long it does matter. Flats must be dark frame subtracted. The strict textbook tells you to take dedicated darks for your flats (same exposure time, same temp.) but this is ridiculous. As long as your flats are of short exposure - a few seconds - the only dark you need for them is a master bias. It does need to be subtracted, though, or you will add noise and the bias pedestal will not be removed. You risk your flats over compensating as a result. So I would try to keep your flats to a couple or three seconds and calibrate them with a master bias as a dark for flats.

Olly

[AlistairW]

Thanks Olly - I will try that.

[Uranium235]

8 hours ago, carastro said:

Flats are not supposed to be temperature dependent, I have never heard that before.  Where did you get that info Rob? 

 

Its information based on first hand experience.

Example:

Cool the sensor to about 0deg, take a bunch of flats - then measure the maximum level of a few of the flats with Maxim DL.

Now, cool it to -15, do the same again.... you will notice a difference of up to 5000 - which is a lot. The reason being is that the hot pixels contribute towards the ADU reading, and therefore cooling the camera to match your lights will give a far more accurate reading.

[Uranium235]

I might add that you need to use the maximum reading, not the average.

[Uranium235]

2 hours ago, ollypenrice said:

If your flats take too long it does matter. Flats must be dark frame subtracted. The strict textbook tells you to take dedicated darks for your flats (same exposure time, same temp.) but this is ridiculous. As long as your flats are of short exposure - a few seconds - the only dark you need for them is a master bias. It does need to be subtracted, though, or you will add noise and the bias pedestal will not be removed. You risk your flats over compensating as a result. So I would try to keep your flats to a couple or three seconds and calibrate them with a master bias as a dark for flats.

Olly

Sorry Olly, for the first time I have to respectfully disagree with you regarding the flat length with a 383. Two or three seconds on a 383 is far too short as you will get a shutter shadow (especially when binning) leading to inaccurate correction, the manual states a minimum of 6 seconds and I found the ideal minimum to be 9 or 10 seconds.

[ollypenrice]

20 hours ago, Uranium235 said:

Sorry Olly, for the first time I have to respectfully disagree with you regarding the flat length with a 383. Two or three seconds on a 383 is far too short as you will get a shutter shadow (especially when binning) leading to inaccurate correction, the manual states a minimum of 6 seconds and I found the ideal minimum to be 9 or 10 seconds.

Fine by me, Rob. I thought that the shutter wipe problem was eliminated by subs of three seconds or more. I haven't used a 383 Atik, though I've processed QSI data from the same chip. If you find you need 10 second subs then I'm hardly going to disagree but it's longer than I've heard of previously. I haven't heard of binning needing longer subs to avoid the 'wipe effect' either but all that says is that I haven't heard of it, not that it isn't true!

As you know, with me theory goes straight in the bin when practice doesn't agree with it!!

Olly

[RayD]

I set up SGPro to run my flats on my Atik 460EX and to get the 22000 ADU I had really short subs of about 0.3s odd.  Do you think my flats generator is too bright?  I did have it turned quite low, but maybe I need to turn it down more?

[ollypenrice]

2 minutes ago, RayD said:

I set up SGPro to run my flats on my Atik 460EX and to get the 22000 ADU I had really short subs of about 0.3s odd.  Do you think my flats generator is too bright?  I did have it turned quite low, but maybe I need to turn it down more?

That should be fine. Your camera does not have a mechanical shutter. Rob and I were talking about cameras with a mechanical shutter which slides across the field of view quickly but, none the less, at a finite speed.

Olly

[RayD]

Ok that's great thanks.

[Uranium235]

27 minutes ago, ollypenrice said:

Fine by me, Rob. I thought that the shutter wipe problem was eliminated by subs of three seconds or more. I haven't used a 383 Atik, though I've processed QSI data from the same chip. If you find you need 10 second subs then I'm hardly going to disagree but it's longer than I've heard of previously. I haven't heard of binning needing longer subs to avoid the 'wipe effect' either but all that says is that I haven't heard of it, not that it isn't true!

As you know, with me theory goes straight in the bin when practice doesn't agree with it!!

Olly

I found the need for 10 seconds while binning becuase the camera is more sensitive to the absence of light in that particular corner with short flat lengths, even though the shutter takes a fraction of a second to open. You can get away with 8 seconds binned, but below that you are flirting with the wipe effect. I found the ideal time just by increasing the sub time until the wipe effect disappeared, and then settled on 10sec as a safe, conservative figure whether binned or not. I just adjust the brightness of the light panel to achieve that sub length for all my filters, which is useful since Im starting to experiment with dark flats (so I can use the same ones no matter what filter).

I've also recently gotten into the habit of checking all my calibration frames for oddities, as sometimes I find bias frames that are "half downloaded" or flats that have readings that are vastly different to the others... but thats usually about 3 or 4 out of every 50.... not sure why I get that, but a few min spent on QC is worth it to weed them out. Ive had the camera for over three years now, and despite its (very minor) quirks im still very pleased with it.... for a FOV that size im prepared to put up with a bit of wierdness