Disappointing Ha result on Pacman

[AKB]

There's so many things wrong with this image, that I really hesitate to post it, but need to improve so would appreciate any kindly advice.

I don't really do Ha (perhaps that should be "really don't do Ha") but it was a full moon last night, and I had just put everything back into the observatory after being away on holiday (to France, yes, very nice, thanks.)  So nothing was tweaked in any way, but I wanted to see if everything basically worked.  Setup is:

• Quattro 200mm reflector
• Avalon M-Uno, guided , but not dithered
• Atik 460EX with Baader 7nm Ha filter in a filter slider body
• 12 x 5min

Although the guiding was done with a separate guide scope, the main optical train did include an OAG but the prism was well out of the way.

Things wrong in the image include:

• Eggy stars – it was very windy
• Noisy – should 60 minutes with 200mm and F4 really be so bad?
• Focus – I setup with a clear filter and and did my best before switching to Ha, but didn't refocus during the hour
• Star spikes – I don't mind the ones from the stars within the frame, but there is a huge vertical one top-centre from out of frame.

So the main issue are the noise level and the spike.  I'm trying to see where the spike may have come from and my best candidate is Shedir.  It's in the right direction, I think, but way out of frame, perhaps four Pacman diameters away?

Am I expecting too much from this setup and the (relatively) short exposure? Thanks for any guidance.

 

 

[tooth_dr]

I would not be too annoyed about that. The stars aren’t round but noise seems good for an hour. I have 4.5 hours with my Atik, and it seems noisier lol 

[AKB]

Ah, OK, thanks.  That’s a useful sanity check.  I shall revise my Ha noise expectations.  TBH, this is one reason why I prefer RGB, or just plain L!

[carastro]

I would have done much longer subs for Ha, but since you were already getting eggy stars maybe that wasn't an option.  That's a good camera (I have the same) and you have a reasonably fast scope, so should have shown a bit more detail than that, maybe it was the sub length that was the problem.

I personally find narrowband easier to do than LRGB.  

Carole 

[Skipper Billy]

The Achilles heel of the Avalon mounts is the performance in windy conditions - the belts stretch and rebound - that's probably the cause of your eggy shaped stars.

5 mins is very short for Ha subs and probably the cause of your noise - although its not that bad.

Can't help with suggestions to get rid of the intrusive spike !

There is a significant halo around brighter objects - are you sure you weren't suffering from dew ??

[vlaiv]

Level of noise vs total exposure surely depends on target surface brightness, so what might be enough total exposure for one target will look quite noisy on another (or if one restrains themselves in processing - same amount of visible noise, but target will indeed look much dimmer).

Atik 460ex is "medium" read noise camera, so I don't think you need to go crazy long exposures (20-30mins like you would need with 9e read noise camera), I would say 5-10 mins is quite enough. Total imaging time on the other hand can be improved, relatively easy (well it does depend on weather and available time under stars).

As for offending star spike, there is very simple solution to apply before hand (can't help now, some post processing magic needs to happen to remove it from subs, and I'm afraid I can't help much there) - rotate ota one way and camera the other for the same angle - target will remain framed the same way, but secondary spider will change orientation and so will offending spikes - since they are far away, you can "rotate them out of the FOV".

[AKB]

58 minutes ago, carastro said:

I would have done much longer subs for Ha, but since you were already getting eggy stars maybe that wasn't an option. 

That's true, Carole, and indeed, I'm otherwise pleased with the camera ("Gina-cam" to denote its source!)

8 minutes ago, Skipper Billy said:

The Achilles heel of the Avalon mounts is the performance in windy conditions - the belts stretch and rebound - that's probably the cause of your eggy shaped stars.

Thanks Skipper, yes I'm assuming that's the case, so not too bothered about it.

9 minutes ago, Skipper Billy said:

There is a significant halo around brighter objects - are you sure you weren't suffering from dew ??

Ah yes, I hadn't mentioned that defect.  I thought, perhaps, some reflection from the filter / coma corrector combination.  Pretty sure it's not dew.  Here's the first sub of the night, just stretched...

 

[Stub Mandrel]

3 hours ago, AKB said:

So the main issue are the noise level and the spike. 

I've had exactly the same spike on the Pacman, more than once. I'm sure it's Schedir.

[AKB]

39 minutes ago, Stub Mandrel said:

I've had exactly the same spike on the Pacman, more than once. I'm sure it's Schedir.

Ah!  Music to my ears.  Thanks so much.  I am encouraged to try again (probably on a different target!)

[tooth_dr]

1 hour ago, vlaiv said:

I don't think you need to go crazy long exposures (20-30mins like you would need with 9e read noise camera).

I wonder is this where I am going wrong in term of noise. I'm using an Atik 383L+ which i guess is quite a noisy camera, so longer exposures would be better?

[vlaiv]

10 minutes ago, tooth_dr said:

I wonder is this where I am going wrong in term of noise. I'm using an Atik 383L+ which i guess is quite a noisy camera, so longer exposures would be better?

Yes indeed, specs say that 7e is typical for this camera, but you can take your bias files and check this, I would not be surprised that you measure your read noise in 8-9e range.

Narrow band will certainly require longer exposures. I mean, you can certainly go with shorter exposures, but for same total imaging time, impact on SNR can be significant. LRGB will benefit from longer exposures provided you are in dark skies (20mag+).

Here is a quick comparison of SNR with different exposure lengths and 7e read noise camera, under relatively "common" imaging parameters (1.1"/pixel, 8" F/5 scope, 21mag skies, target mag 24, Lum, 4h total imaging time):

1 minute exposures (240 of them): 3.453

2 minute exposures (120 of them): 4.11 (19% increase in SNR over 1 minute)

5 minute exposures (x48): 4.743 (37% increase in SNR over 1 minute)

10 minute exposures (x24): 5.03 (46% increase in SNR)

15 minute exposures (x16): 5.137 (almost 50% increase over 1 minute exposures, for same total imaging time, so nothing changed except using longer subs)

With NB imaging, this difference is even greater, due to minimal contribution of LP noise. In above case LP noise is comparable with read noise (8e vs 7e) in two minute exposures, and both are x4 larger than target shot noise (for single exposure of course).

[AKB]

3 hours ago, vlaiv said:

rotate ota one way and camera the other for the same angle - target will remain framed the same way, but secondary spider will change orientation and so will offending spikes - since they are far away, you can "rotate them out of the FOV".

Thanks for that.  Good idea, but it would be hard for me since this would affect the OTA balance considerably (the focuser is aligned to the bottom in the home position.)  But I'll bear it in mind.

___________

Edit:  Actually, I may try this – it should not be necessary to rotate much since the offending star is, I believe, far away from the normal FOV.

[tooth_dr]

52 minutes ago, vlaiv said:

Yes indeed, specs say that 7e is typical for this camera, but you can take your bias files and check this, I would not be surprised that you measure your read noise in 8-9e range.

Narrow band will certainly require longer exposures. I mean, you can certainly go with shorter exposures, but for same total imaging time, impact on SNR can be significant. LRGB will benefit from longer exposures provided you are in dark skies (20mag+).

Here is a quick comparison of SNR with different exposure lengths and 7e read noise camera, under relatively "common" imaging parameters (1.1"/pixel, 8" F/5 scope, 21mag skies, target mag 24, Lum, 4h total imaging time):

1 minute exposures (240 of them): 3.453

2 minute exposures (120 of them): 4.11 (19% increase in SNR over 1 minute)

5 minute exposures (x48): 4.743 (37% increase in SNR over 1 minute)

10 minute exposures (x24): 5.03 (46% increase in SNR)

15 minute exposures (x16): 5.137 (almost 50% increase over 1 minute exposures, for same total imaging time, so nothing changed except using longer subs)

With NB imaging, this difference is even greater, due to minimal contribution of LP noise. In above case LP noise is comparable with read noise (8e vs 7e) in two minute exposures, and both are x4 larger than target shot noise (for single exposure of course).

Thanks again for your answer

How do I check read out noise from my bias frames

 

[vlaiv]

5 minutes ago, tooth_dr said:

Thanks again for your answer

How do I check read out noise from my bias frames 

 

Easiest thing to do is to take set of bias files, convert each bias sub to electron count - Atik website lists gain as: 0.41e-/ ADU, so convert all subs to 32bit float point precision, multiply their values with 0.41 and split them into two equal groups (take even number of subs so they can be split into two equal groups). Stack each group (average) and subtract two stacks. Measure standard deviation of resulting sub, and multiply with square root of number of subs to get read noise.

[tooth_dr]

28 minutes ago, vlaiv said:

Easiest thing to do is to take set of bias files, convert each bias sub to electron count - Atik website lists gain as: 0.41e-/ ADU, so convert all subs to 32bit float point precision, multiply their values with 0.41 and split them into two equal groups (take even number of subs so they can be split into two equal groups). Stack each group (average) and subtract two stacks. Measure standard deviation of resulting sub, and multiply with square root of number of subs to get read noise.

Yep, I’m not sure I understood much of that ?

[Stub Mandrel]

2 hours ago, tooth_dr said:

Yep, I’m not sure I understood much of that ?

A bit like listening to a distant radio station it fades in and you understand a bit, then it fades out again!

I have no idea how you do all the maths with image files, convert to 32-bit I get, and stack each group but...

Sorry Vlaiv!

[vlaiv]

ImageJ to the rescue!

1. File / Import image sequence - select appropriate files, and parameters (for example for first stack tell it to start at one and end at mid sub, for second stack to start at midsub+1 and load the rest) ...

It should open up image stack ..

2. Image / Type / 32 bit ...

This one is self explanatory, it should convert image stack to 32bit format ...

3. Process / math / multiply, enter e/ADU value (0.41 in case of Atik 383l+), select all slices in stack when asked (yes)

This will multiply each pixel in image with entered value

4. Image / Stacks / Z-project, and then select average - this will create new image that is average stack of all images in previous stack

Repeat above for second stack, you can close stacks after this and you will be left with two images (AVG_something and AVG_something_else, depending on file names, you can also change then name by right click and rename ...)

After that, you do:

5. Process / Image calculator

Select first image, second image, and subtract as operation, have 32bit precision and create new image checked off ...

This will produce new image that is difference of two images

6. Analyze / Measure on new image (difference of two stacked images)

It should open a window named "Results" with table - read off StdDev column, punch that number into calculator and divide with square root of number of subs used (total number of subs, so you had 16 subs in two groups, first was 1-8 stacked, second was 9-16 stacked, and you divide StdDev result with 4).

This is your read noise value.

After that you might find it interesting to play around with images and discover wonderful things that you can do with them using ImageJ and some math (like photometry, astrometry, ... even deconvolution, alighment and stacking using suitable plugins, mosaic stitching, all sorts of things, but it is kind of "manual" labor compared to ready made solutions.

 

[AKB]

Tried again tonight – moon, of course, and some thin clouds most everywhere.  I've switched targets to the Pelican because it's almost directly above.  So now:

• eggy stars gone (wasn't windy)
• noise better (stronger target?)
• focus better
• star halos ... still TERRIBLE!

This is 3x5m and I've removed the (unused) OAG, and also moved the Ha filter to before the coma corrector, rather than after, thinking that it was a reflection between them.  Processing was a bit of noise reduction (Mure) and some stretching.

Is this a problem with the filter?  Just don't know where to start looking.  Suggestions welcomed.

Aside from that, this seems rather encouraging for 15 minutes given the conditions.

 

 

[carastro]

That looks more promising, but the subs really need to be at least 600secs long to get the best out of your set up.  I sometimes do 900secs on a dimmer target.  

Carole 

[AKB]

Thanks Carole.  Yes, longer subs and lots more of them in due course, but at the moment, it's only going to make the halos worse...?  I assume so, anyway.

Tony

[AKB]

Just for comparison, here's 5x2m of L data (TBH, I'm surprised I could see anything in these conditions, but the point of the comparison is simply to look at the bright star and the lack of a strong halo.)  I'm really suspecting the filter now...

[ollypenrice]

I wouldn't underestimate the effect of a full moon, especially with a 7nm Ha filter. I find that the 3nm Astrodon (expensive!) is vastly more moon-proof but I still skip the days around full moon.

Spikes from stars off shot: what I've found is that the spikes and flares often disappear if the progenitor star is near the centre of the frame, so you might be able to shoot a 'patch' panel by putting the star in frame but with the flared part of the original still in the frame.

I find my 460 works best with 15 minute subs from our dark site. I didn't find 30 minute subs very effective, though on our Kodak cameras they work well.

Olly

[tomato]

13 minutes ago, ollypenrice said:

 I still skip the days around full moon.

Such luxury Olly, still, I guess you have to sleep sometime. I wonder if I will ever be in a position where I can forego a clear night just because there's a full moon in it. I live in hope....

[RayD]

On 24/09/2018 at 19:31, tooth_dr said:

Yep, I’m not sure I understood much of that ?

The worrying thing is he said that's the easiest thing to do!! 

[AKB]

5 hours ago, ollypenrice said:

wouldn't underestimate the effect of a full moon, especially with a 7nm Ha filter.

Thanks for that, Olly.  I simply hadn’t contemplated that it was to do with anything other than the star itself.  Will try again later, but this was three nights IN A ROW, so had to be doing something.  Great point about the ‘patch’ panel and the longer exposures too.   Thanks again.