darkframe problem still QHY9

[Peteram2003]

  Hello group,

I am still having problems with my darkframes. I am using Astroart for capture with the Ascom Drivers. I use slow download settings with or without "no usb during readout".
I can capture signal just fine, however darkframes and bias frames is where the problem starts. Often my darkframes have higher average backgrounds then my lightframes or so close that I end up with many pixel values below zere ( with exact same settings and darks taken in absolute darkness). My bias frames have higher average values then my darks? and a 60 second dark has higher average values then my 600 sec darks.  I was wondering if any of you are seeing this behavior because it makes absolutely no sense to me at all. It happens at all binning modes as well.
I have included a screenshot of astro art. In it you see the statistics of a bias frame, a 60 second dark frame and a 600 second darkframe. Take notice of the average background values. If I subtract the 60 second darkframe from the 600 second darkframe I end up with a negative background value!!
I really need some help on this one.


Best regards,


Peter

[QiuHY]

Hello,

        We are just solving this problem. Now you can set the offset value(increase a little)  if you plan to capture bias frame. Later our new version software may handle this automaticly.
For darm frame subtract. Please use the same exposure time of the light frame:  For example:  600sec light frame should use 600sec dark frame.


Best regards,

[Peteram2003]

Hello Q,

Are you saying that you can replicate the problem now and that you know what causes it?


Regards,

Peter

[Ivan7(rus)]

I don't see this problem on my camera in 1x binning (I didn't test it in 2x2 binning).

The software solution could be to calculate average value on dummy pixels (not affected by light) and substract this value from image (+ add a constant value, for example 1000). But you need to capture all pixels, not only visible.

Dummy pixels areas are described in KAF-8300LongSpec.pdf (pdf can be easily found by google)

[QiuHY]

Hello,

        I have test some QHY9 here.And will post a curve of it.  I am also analyze the reason cause it. One thinking is the thermal various(Before capture there is a high speed pixel clean operation) cause the offset shift. We found the output level of 8300 sensor is sensitivity with the temperature.(From 10degree to -20degree it will shift more than 300mV).  The temperature sensor can not get the intrenal CCD chip temperature so it will not show this tempertuare various.

Best regards,

[Peteram2003]

Hello Q,


Quote: "Now you can set the offset value(increase a little)  if you plan to capture bias frame" unquote.
Can you explain to me what exactly you mean here?
Also I have used darkframes with the same exposure time as my light frames, but ended up with a lot of darkspots in my lightframe and even some negative values. I just don't understand how that can even happen. It must have something to do with readout values I would think even though that should be exactly the same in a light frame.
What I have done so far is to add a pedastal (offset) in the lightframes of i.e. 1000adu before I subtract the dark frame to prevent it from going into negative territory.
It then seems to at least get rid of any pattern noise in the image.
Another thing I have wanted to try is to use a somewhat higher gain in my lightframes as opposed to the gain in my darkframes. But those are ofcourse nothing more than "tricks" and that doesn't really solve the problem ofcourse.
The temperature variations that Q talks about, I don't really understand. If I look at 2 hours worth of 10min darkframes, they all have about the same average pixel values with only minor variations, so I doubt that that is the problem.
I will experiment some more with different readout settings and other software to see if that makes any difference.

Regards,

Peter

[Peteram2003]

Hi Ivan,

I am glad you don't experience the same problem, which makes me think that the issue might be a hardware issue. I appreciate your idea for the workaround, but in the end I just want to be able to properly calibrate my images with dark frames (I can live without the bias if necessary). I will read up on the dummypixels however.

Regards,

Peter

[QiuHY]

Hi Ivan,

I am glad you don't experience the same problem, which makes me think that the issue might be a hardware issue. I appreciate your idea for the workaround, but in the end I just want to be able to properly calibrate my images with dark frames (I can live without the bias if necessary). I will read up on the dummypixels however.

Regards,

Peter

For example: Offset =120 . If you set OFFSET=125.You will get a frame with higher average value, This is a constant, it will not effect otherthing change.

Light frame offset=120.   Dark frame=125. BIAS frame =130. This will solve you problem.

Remeber for a correctly darm frame calibration. The Dark frame exposure time must the same with light frame.    So only the BIAS frame will meet this problem. You can adjust the offset when capture bias frame.

Don't use difference gain.That's will not make correct calibration.







Best regards,

[Peteram2003]

Hello Q,

I your example I use a higher offset for my darkframe and an even higher offset for my bias frame. When I use the same offset, the bias is allready a higher value than the darkframe. If I use an even higher offset, will that not make things even worse?


Peter

[QiuHY]

Hello,

         Sorry a mistake.  Should be lower value in dark and bias frame.


Light frame offset=130.   Dark frame=125. BIAS frame =120

         You can use this setting for first. We are reseraching where is the reason. If just a offset level problem. This will not effect the calibration. 

And I think add some offset(constant) to the lightframe is not bad.Even you get the same average value on light frame and dark frame. After subtract, half of the background histogram curve will below zero. So in any condition you should
add the constant to the lightframe to keep all the pixel not below zero.





Best regards,

[Peteram2003]

Hi Q,

Ok I understand what you mean then. I am curious though if you are able to replicate the problem. I am testing right now with the drivers from your website and EZCAP V1.6.
So far I am getting the exact same results: average pixel value on a 60 second darkframe is lower than on a 10 sec darkframe. The only thing is that now I have a bias frame with a slightly lower background than all the others. I'll keep testing and will report here.


Peter

[Peteram2003]

Allright, I have tested with EZCAP,  Tom's CCDcap and Astroart with Ascom driver old and beta. Settings binned 1x1 2x2 3x3 and 4x4. High readout, normal and slow readout. In all cases I get the exact same result. A 60 sec darkframe will have a LOWER average pixel value than a 10sec darkframe at the same settings. The 10 is lower than a 5 sec and so on with the bias having the highest average values.
The dark frames do show hot pixels as opposed to the bias frame which does not. This is the exact opposite of what can be expected.
Q I did try to change the offset value of a bias (lower) compared to that of the dark frame and ofcourse than I end up with lower pixelvalues in my biasframes, but that would be a cheat so to say.
I cannot wait for you to tell me how this is possible. Please keep me updated as much as possible.

Peter

[QiuHY]

I just test two camera in lab condition and one has a little negative level change in 0, 60,120,240sec exposure. The longer exposure image has  lower level than short exposure. In another camera it has a positive level change. I am compare the hardware to see if it is exist in CCD sensor itself or in the PCB board.

The CCD output level may change in some conditions.   There is a process way called "Optic black offset calibration" technology in most consumer camera like the video camera , DLSR etc. It is use the average value of the optic black area or overscan area to correct the image area on each line. So that the ccd is become offset free and you can not see the offset settings. But it may cause some over correction if there is hot pixel exist in the optic black area in long exposure.

In astronomy camera the target is output the original RAW date without any software correction in camera. So you have to do these correction in future data process. This is to get the real RAW data and keep the physical value.

Of cause we can add the level correction in software or hardware of the camera to reduce some more process work . First we are trying find out where is the level change happen.

Best regards,

[Peteram2003]

Hello Q,

I am happy that you are working on the problem and that you are able to replicate the problem in another camera. I suppose you will let me know here or via my technical request form what you find out or how I can fix mine.

Thanks very much,

Peter

[QiuHY]

Hello,

          Can you do a simple test . I will see if the result  is the same:

You must set the temperature in auto look control mode. After it arrived the stable statu, you make a 0sec at first. Then make a 120sec. And then make a 0sec again. And please tell me the result of the average level.You can use the overscan area's average.


Best regards,