ADU and flats confusion

[Anthonyexmouth]

Been reading a few threads on my asi294mc pro and it seems people are saying very short flats are not great for this camera and 2-4sec flats are an advantage but i cant find any chat on how to accomplish this. 

i was using the t-shirt method and the flats wizard in APT and an ADU of 25000 which gives an exposure of .00596s according to APT

anyone know what my ideal APT would be for this camera or how to work it out? also how do i get my exposure length longer? do i need a flat panel? Is anything i've been reading true. 

thanks

[Laurin Dave]

I use a tracing panel and enough sheets of A4 writing paper to get to the right exposure..  5-10 seconds for my camera.  An iPad set to a white screen would also work. An ADU of 25-30000 based on 16 bit works well

Dave

Edited September 21, 2019 by Laurin Dave

[Anthonyexmouth]

3 minutes ago, Laurin Dave said:

I use a tracing panel and enough sheets of A4 writing paper to get to the right exposure..  5-10 seconds for my camera.  An iPad set to a white screen would also work. An ADU of 25-30000 based on 16 bit works well

Dave

ah ok, cool

does the fact the 294 is 14 bit make much difference to the target ADU?

[Laurin Dave]

So long as APT is set to 16bit ie 0-65000 then no,  what you are trying to do is to get the ADU to just less than half the max reading when the pixels are saturated (which is 65000 or so at 16bit). To test just your setup over expose and see what ADU reading you get... then set the exposure to give just under half of that..   You may also need to take matching flat darks at the same exposure/temperature ( I do with the ASI1600) to calibrate the flats, others with this camera will hopefully chip in on that

Dave

[alan potts]

2 minutes ago, Laurin Dave said:

So long as APT is set to 16bit ie 0-65000 then no,  what you are trying to do is to get the ADU to just less than half the max reading when the pixels are saturated (which is 65000 or so at 16bit). To test just your setup over expose and see what ADU reading you get... then set the exposure to give just under half of that..   You may also need to take matching flat darks at the same exposure/temperature ( I do with the ASI1600) to calibrate the flats, others with this camera will hopefully chip in on that

Dave

Of course I was deeply confused by this, I looked at the ADU which meant nothing to me, saw some exposure info, set it and banged them off, seems to work alright but then I am hardly an expert. Took the data for an ADU of 20,000 which oddly was the same exposures as for the Canon.

Alan

[newbie alert]

I keep reading about this 50% peak for flats and can't get my head around it.. 30% yes I agree with..is this a thing with cmos?

[andrew s]

Reading this thread made me think it might be worth revisiting what flats are for and if that sheds any light on what level they should be at.

We use flats for two very different reasons.

1) To correct for non-uniform illumination of the detector by the optics (vignetting) and to remove local non-uniformity due to dust etc. on the optics (dust bunnies).

2) To correct for the non-uniform response of the individual pixels to light (PRNU) and in the case of CMOS also the individual pixel amps and A/D converters.

An ideal detector would be linear in its response to light across its whole range without adding any noise so we would just have the noise due to photon statistics.

A real detector however, has 4 principle "noise" regions as the incident light increases from zero to the maximum it can record. These are read noise, photon statistics, PRNU and saturation non-linearity (due to blooming or ABG operation).

In addition to these a mechanical shutter can cause non-linearity for short exposures and with CMOS (and CCDs to a lesser extent) amp glow does the same at longer exposures. I have not seen any reports that electronic shutters cause any problems.

To remove the effects of PRNU the flat must be exposed long enough to be in that region i.e. beyond the read and photon noise levels but below saturation. The PRNU region kicks in at about 1/(PRNU)^2 so if the  PRNU is about a typical 1% (0.01) then 10,000 electrons.

How this translates to ADU is via the inverse gain e/ADU and for a CMOS camera can normally be changed by the user.

Why does this matter. Most people focus on the first use of flats and as long as the noise is low that it will work.

However, if you what to correct for PRNU you need the flats to be in the PRNU range. 

Incidentally, if key elements of your your images are in the PRNU range then adding more data (reducing photon noise) will not have any effect as the PRNU will remain at a proportionally constant level. Using your flats to remove it will allow you to gain the benefit of more exposure.

Regards Andrew

[carastro]

I always try to get around 22,000 ADU which translates to approx 1/3 of 65,000.  If I am using a DSLR I try to get the histogram between 1/3 and under halfway across.  

I am not into all the technicalities, but this seems to work for me.

If the reading shows it is too bright, then dim the light source, if it's not bright enough increase the length of the exposure.

I tend to do sky flats, but I also have one of those tracing boards mentioned above enclosed in a box and put typing paper over it to diffuse the light.

If your flats are too bright they won't do their job.

With a mechanical shutter, you need to expose for long enough to avoid shutter shadow, so this could mean having to dim the light down to compensate.

All this will vary from one filter to another if you're using a mono camera.

HTH

Carole 

Edited September 22, 2019 by carastro

[Deeko]

I have the ZWO 294MC Pro and use the t-shirt method with a LED tracing panel (see attached photo) for creating flats. Like the OP I have very recently read more technical discussions on the 294 and issues with creating flats and bias frames with exposures that are too short, I have been using 0.03 sec flats and 0.001 sec bias files (via APT's CCD Flats Aid feature) which in all likely have not been correct with flats in particular too bright, though I haven't had a hard time with dust etc. either. This weekend I tried the t-shirt + 2 sheets of paper with the LED panel I am able to get 4 second flats to reach 25000 ADU, I am also going to ditch bias frames and create dark flats of 4 sec exposure (same as flats). This is a recent development for me and not yet stacked any lights yet with the longer flats + dark flats. 

[Deeko]

Following on from my last post, does anyone have any advice/info on the APT histogram for flat files shown in my attached screenshot? The flat frame in image was taken with the 294 with the IDAS D2 filter (designed for LED street light pollution) for a target ADU of 25000.