Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Interesting phenomenon with asi 1600mm and OIII flat field

  • Please log in to reply
15 replies to this topic

#1 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 14 April 2019 - 06:34 PM

Hello everyone,

 

I have noticed something odd with my narrowband flat frames with my asi 1600mm.  My HA flats seem perfectly normal.  There is a small differential between pixels around 2-3%, which seems pretty normal.  This is a screenshot of my HA flat.

 

HA Flat.jpg

 

The OIII flat is another matter, however.  It shows a very deep dropoff in light to one side.  This dropoff is in the range of 7-9%, which seems very high for differences in pixel responses in the chip.  I am figuring that it is chip related because I rotated the camera in between shots and the right side continues to show the effect.  If it was optical or filter related, I think it would move if I rotated the camera.  Here is the OIII flat.

 

OIII Flat.jpg

 

Is it possible that a CMOS chip could have a large response variation in one wavelength of light and a small variation in another?

 

I have included the original fits files here for those who want a bit better quality: https://www.dropbox....yIEV8Psyqa?dl=0

 

Some more information:

 

Telescope is WO71GT with reducer

The filters are 2" round filters in Atik EFW3 (preparing for full frame camera one day).

Flats are sky flats taken at midday with a pair of white shirts over the lens.

I have taken many flats with varying exposure times to confirm that it is not an exposure problem.

I have taken exposures through a range of focuses to make sure focus didn't slip when things were idle recently.

 

Anyone have any ideas?

 

Thanks much,

 

-John



#2 OldManSky

OldManSky

    Skylab

  • *****
  • Posts: 4,418
  • Joined: 03 Jan 2019
  • Loc: Valley Center, CA USA

Posted 14 April 2019 - 07:38 PM

Do the OIII lights show a similar light drop-off on the right side?

Or is this only occurring on flats?



#3 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 14 April 2019 - 09:41 PM

I checked back with some of my old data, and it's hard to tell because ADU numbers are in the low 1000s instead of 20,000, but it does not seem to be there.  The median adu value are within about 1-2% of each other, with samples all taken from the same region.

 

 

-John



#4 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,824
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 14 April 2019 - 09:46 PM

The only question that really matters is: Are the OIII flats properly calibrating OIII lights? 


  • Stelios and rockstarbill like this

#5 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 14 April 2019 - 10:25 PM

That is the pertinent question, indeed, Jon.  So, far, I have very few light frames to test.  I will explain.  I just came back from a 6 month hiatus after my son was born so I have been getting everything back up and running.  Things have been idle in the observatory for that time.  I decided to grab some data on some planetaries and pulled out my VC200l that I rarely use.  The light data was good, and the flats looked fine, but they overcorrected significantly.  This caught my attention because it had never happened to me before.  So, I went to debugging things, and have attached my most used imaging train, my WO71gt, and went through each of my filters.  First, my camera straight through showed a flat image, then each of the filters looked fine.  But the OIII looked not so great.  This is where my question really comes in.  I am trying to figure out whether this could be causing the overcorrection, but I can't really do that until I know why this one filter is causing these problems.  Right now my best guess is that my chip just happens to be less sensitive to the blue/green wavelength in that part of the chip.  I have no real knowledge of CMOS architecture to confirm whether that is a realistic possibility, though.  I know that you are very knowledgeable about these kinds of things.  Do you think that it is reasonable to believe that there could be an 8% differential between the pixels in one part of the frame compared to the rest of the frame?

Hopefully I have not deluged you with too much information.

 

Thanks for all of your help,

 

-John



#6 TelescopeGreg

TelescopeGreg

    Mercury-Atlas

  • -----
  • Posts: 2,961
  • Joined: 16 Jul 2018
  • Loc: Auburn, California, USA

Posted 15 April 2019 - 02:01 AM

Can you rotate the OIII filter with respect to the camera, to determine if the filter or camera has the issue?



#7 terry59

terry59

    Voyager 1

  • *****
  • Posts: 10,348
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 15 April 2019 - 06:57 AM

Hello everyone,

 

I have noticed something odd with my narrowband flat frames with my asi 1600mm.  My HA flats seem perfectly normal.  There is a small differential between pixels around 2-3%, which seems pretty normal.  This is a screenshot of my HA flat.

 

attachicon.gif HA Flat.jpg

 

The OIII flat is another matter, however.  It shows a very deep dropoff in light to one side.  This dropoff is in the range of 7-9%, which seems very high for differences in pixel responses in the chip.  I am figuring that it is chip related because I rotated the camera in between shots and the right side continues to show the effect.  If it was optical or filter related, I think it would move if I rotated the camera.  Here is the OIII flat.

 

attachicon.gif OIII Flat.jpg

 

Is it possible that a CMOS chip could have a large response variation in one wavelength of light and a small variation in another?

 

I have included the original fits files here for those who want a bit better quality: https://www.dropbox....yIEV8Psyqa?dl=0

 

Some more information:

 

Telescope is WO71GT with reducer

The filters are 2" round filters in Atik EFW3 (preparing for full frame camera one day).

Flats are sky flats taken at midday with a pair of white shirts over the lens.

I have taken many flats with varying exposure times to confirm that it is not an exposure problem.

I have taken exposures through a range of focuses to make sure focus didn't slip when things were idle recently.

 

Anyone have any ideas?

 

Thanks much,

 

-John

Hi John,

 

  Exposure time is only one variable here. What is the spread between the mean and median ADU values? Have you tried varying the light source intensity and duration of the flat?

 

  Edit: If you haven't used them there is no way to know what issues there really are


Edited by terry59, 15 April 2019 - 07:09 AM.


#8 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 15 April 2019 - 08:08 AM

Can you rotate the OIII filter with respect to the camera, to determine if the filter or camera has the issue?

I rotated the camera 90ยบ with respect to the filter, and the phenomenon did not change its position or orientation.

 

 

 

Hi John,

 

  Exposure time is only one variable here. What is the spread between the mean and median ADU values? Have you tried varying the light source intensity and duration of the flat?

 

  Edit: If you haven't used them there is no way to know what issues there really are

I took a range of exposure times to see if it was exposure based, and it was present in all of them.  The shots I included in the post has a center brightness of 27680 median and a side brightness of 25300 median, so a difference of 9.4%.

 

As for not having used them, that is correct, not with this scope.  I noticed this effect when trying to debug my flats from a different scope; one I do not use very often.  I figured that I would rule out as many things as I could with an imaging train that I knew well, before jumping into trying to debug the imaging train that I was not familiar with.  As soon as the weather clears (hopefully tomorrow), I will get some light frames with my 71gt, and see exactly what is going on calibration wise.  In the end, I probably made this post a bit prematurely, as I don't have enough data to really help you guys make a diagnosis.  A part of me thought that someone would come back quickly and say, "yeah, chips have this kind of response sometime, this is normal, your flats problem lies elsewhere." :)

 

Thank you again, everyone, for all of your help; you guys are all awesome!

 

-John



#9 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 20 April 2019 - 10:24 AM

So, as a quick update.  I took some more light frames, and confirmed that it is visible in them as well.  The reason I did not see them in my previous ones (it was there) is that I was looking at the uncalibrated light frames.  This darker area is located right where the dark current shows up, so the added dark current was compensating for it and thus made it pretty much invisible.  When I only calibrated with dark frames, it became evident.

Thus, it looks like in the OIII band, the far right side of the chip is about 8% less sensitive.  Is that a viable and realistic conclusion?  I figure that someone has to know more about CMOS tolerances than I do.  Is this difference in sensitivity (if it even is that) a normal thing?

 

Thanks much,

 

-John



#10 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,824
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 20 April 2019 - 01:50 PM

So, as a quick update.  I took some more light frames, and confirmed that it is visible in them as well.  The reason I did not see them in my previous ones (it was there) is that I was looking at the uncalibrated light frames.  This darker area is located right where the dark current shows up, so the added dark current was compensating for it and thus made it pretty much invisible.  When I only calibrated with dark frames, it became evident.

Thus, it looks like in the OIII band, the far right side of the chip is about 8% less sensitive.  Is that a viable and realistic conclusion?  I figure that someone has to know more about CMOS tolerances than I do.  Is this difference in sensitivity (if it even is that) a normal thing?

 

Thanks much,

 

-John

It is probably less that one side of the chip is less sensitive, and more that the OIII filter (or perhaps even something else...something protruding into the light path) is not getting completely centered over the chip. If the edge of the filter starts to block light (shade) that side of the sensor, then you will see what you are seeing. If it is something other than the filter, it may just be protruding into the light path. It may even be something more translucent to Ha and more Opaque to OIII. We call this Vignetting, which in photographic terms is "decoration of the periphery of the sensor frame by shadows", for lack of a better description.

 

Now...I would not fiddle with the centering of your OIII filter yet. I wouldn't fiddle with your image train at all yet. For flats to properly correct lights, everything must remain in teh same orientation. If the OIII filter is a bit decentered and thus shading one side of the sensor more, then that should be evident in both the lights and the flats. If it is something protruding into the light path, it may even be inside the camera...which may explain why it is rotating with the camera. If the flats are reproducing the full "field shape and structure" identically to the lights, then dividing the master flat out of dark calibrated lights should correct any shading issues. On the other hand, if for some reason the flats are not accurately reproducing the field shape and structure as in the lights, then the flats may not perfectly correct the lights...they may leave remnants or shadows of dust motes, may invert shadows, etc.

 

Sometimes the nature and dispersion of the rays of photons with flat frames isn't exactly the same as with light frames. This may result in changes to shading, or perhaps additional reflection, etc. in the flat frames. This is usually indicative of some kind of issue with the light source used to create the flats...so trying different sources may help. Try with and without diffusion (i.e. with an without a stretched T-shirt). Try against computer screens, white evenly illuminated walls, or even an empty flat blue sky during the daytime, if you can. See which light source reproduces the field structure of your light frames best.


Edited by Jon Rista, 20 April 2019 - 01:54 PM.


#11 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,824
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 20 April 2019 - 02:02 PM

I also wanted to make sure you have seen the flats without any stretch. Yes, the relative difference between the median of the peak and the median of the shadow is ~7-8%, this is what the flat ACTUALLY looks like:

 

mAdwKdc.jpg

 

The effect is actually quite subtle. Since it is identical between the rotated and non-rotated versions, it probably is not OIII filter centering. Have you taken a look at the camera itself to see if there is anything that might be shading one side of the sensor? If you can rule that out...I am then curious if you have tried taking a flat with a broadband blue filter. Does that also have similar shading?



#12 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,824
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 20 April 2019 - 02:35 PM

After looking at your Ha flats...I am starting to wonder about your flats in general. The Ha flats don't really look any better...they look different, but not really better. They actually look like they have structures that may not belong...perhaps a light leak, or perhaps the diffuser is not doing a very good job... Have you actually tried calibrating any of your lights with any of your flats? Can you share some of all of your data...a few bias, dark, flat, and light frames from each channel?



#13 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 20 April 2019 - 04:50 PM

Hi Jon,

 

Thank you so much for your response.  I have looked in the camera, and I don't see any obstructions, but I will look again and make sure that I did not miss anything.  I began just taking sky flats, and have moved on to the white t-shirt over the front of the scope.  There could definitely be something wrong with the way that I have been doing things.  I have taken off and re-attached the diffuser a number of times, so if it was a mis-positioning, I figure that it would have been solved after a couple of repositionings.  I have taken a flat with a broadband blue filter, and it did not show the darkening on the right side, however, I compared it to the OIII, where the darkening is quite pronounced, so I might have missed something more subtle.  I will take some more this evening, time permitting.

 

When I comes down to it, flats have always been my weakness.  My main problem is a lack of time for this hobby.  Flats have taken a lot more time to get good at, so I have focused mainly on keeping a clean image train, and just relying upon darks to calibrate things.  I figured that after my recent small hiatus that it was a good time to finally get good at taking flats and using flats.

 

I will try and take some lights, calibration frames, and flats tonight, time permitting.  Hopefully, with your help, I can figure out what, if anything, is going on.

 

Thank you again for all of your help.  You have been great, and very patient!

 

-John



#14 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,824
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 20 April 2019 - 07:38 PM

Again, in the end...if the flats properly calibrate the lights, then that is what matters. This would only really be an issue if the flats were not calibrating the lights properly for any reason. 



#15 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 22 April 2019 - 10:02 PM

Ok, since baby decided not to sleep last night, we made some time in the observatory.  Need to get them loving the stars early smile.gif.  I took some light frames, some dark frames, and flat frames as well as some flat-darks.  I have included the master dark, the master flat, and the master dark-flat in the dropbox link along with the integration of the lights with no flats and with flats.  I have only included a few uncalibrated lights since I have used up my dropbox space, heh heh.  Here is the executive summary:

 

The flats are overcorrecting.  I was only able to get data for OIII, but that's where most of the problems seem to be located, so I started there.  The stack of lights calibrated only with the dark frames show a definite darkening on the right side of the frame.  The flats look to show that as well.  But, when I calibrate with both darks and flats, the areas that are dark in the flats are significantly too bright in the calibrated lights, thus showing overcorrection.  I included the dark-flats in case you can see any sign of a light leak.  The stack look pretty clean to me.  I made the dark-flats in the daytime with the observatory closed and only a lens cap over the telescope.  I figured if there was a light leak that it would make itself known pretty quickly as the ADU values with go up significantly as it was daytime.  This did not happen.

 

As a note, I took the flats with a light panel placed over a what t-shirt that was used as a diffuser.  The shirt was flat across the lens opening.  I also tried raw sky flats, but I have some high clouds right now, so I can only confirm that raw sky flats show the same darkening on the right side, but they are not high quality enough to actually use in calibration since the clouds cause some problems.

 

Here is the stack of lights calibrated with darks.  The stack was 30 OIII lights taken at 120s.  I integrated without registering them so as to keep the edges as clean as possible.  They have been stretched to show the problems.  You can see a region to the right and left that are darker than the rest.

 

integration wo flats.jpg

 

 

Here is the stack of lights calibrated with both darks and flats.  The flats have been calibrated with flat-darks as well, although they were only .75 seconds each, so there was not much dark current to take care of.  The regions that were dark are now far too light.

 

integration with flat.jpg

 

 

I have included a link to my dropbox which contains the masters and a few raw light frames.  I uploaded until I ran out of space.  Here is the link to the dropbox: https://www.dropbox....kzaddRyXUa?dl=0

 

 

What are your thoughts on this?  Does this look like a light leak on the flats?  There is clearly some issue with the green-blue region of the spectrum when it comes to these particular areas of this cmos chip, however, I have trouble understanding why they show up in both the flats and the lights, but the flats just overcorrect for them.  If anyone has any ideas, I am open to trying them.

 

Thank you so much for all of the time you are spending looking at my problems.

 

-John

 

 



#16 ceteris_paribus

ceteris_paribus

    Vostok 1

  • *****
  • topic starter
  • Posts: 109
  • Joined: 21 Sep 2014
  • Loc: California

Posted 23 April 2019 - 07:46 AM

Well, I think I might have found the problem.  I went back and tried to manually do the calibration in pixel math, and found that my master dark was only registering at just over 100 ADU instead of 750 ADU.  I am not sure why the ADU value for a 120s dark was so much less than for a .75s dark, but likely when I integrated the individual darks, I must have left either a scaling box checked or a normalization box checked.  I re-stacked the darks, and once they had the correct ADU value, the lights calibrated just fine like they did before.  So, it looks like no damage from being idle for a few months, and I have learned that just because they looked weird doesn't mean that the flats were to blame.  Thank you Jon, and everyone for all of your help.  I love that you guys are here for me and others to ask questions of; it makes the debugging process for things like this a lot less intimidating.

 

Thanks again!

 

-John




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics