ZWO1600MM - Some novice questions

[teoria_del_big_bang]

I am at the start of what seems like a fairly steep learning curve after switching from DSLR, and I wish I had a quid for everytime I heard the phrase "Steep Learning Curve" since I started this ?  

I must admit after buying the new camera and then started reading up on all the settings and all the processing required for these cameras with LRGB or NB I felt I should maybe return it to FLO and stick to my DSLR. I was just bewildered with data overload and I was convinced it was too difficult for myself. One issue for me is that with the DSLR I did so much in the daytime just to get used to what needs doing and setting it up correctly as I could at least focus on something like a distant tree and test things. With this it seems to me it is so sensitive I must do all my work at night with at least some stars to focus on so it taking longer than anticipated to get to grips.

Last night with a couple of hours clearish sky I did seem to get on quite well and with the info I had absorbed so far from here using APT set up to try some LRGB at M42 (nice easy target although I only had this available for a maximum of 1 1/2 hour then out of sight but it was in a spot the clouds were less dense)

Gain 76, Offset 15 (I think - see below), CCD Temp -20, Exp 30Sec, Bin 1x1

Then took 30 exposures at this with Luminance filter. All seemed to go fairly well, not sure my exposure time or gain is optimal I was pretty pleased with the results. Data pretty poor due to some cloud cover but this was just a test and left no time for RGB data as it was full cloud cover by now.

All above was just to say how I was progressing so you can tell what early stage I am at. Anyway I will  stop rambling and get on with questions ?

1) Any thoughts on Exposure time and gain/ offset etc. I know this is possibly for me to experiment with to find the optimal.

2) What exactly does the "Offset" do? Also I cannot find anyway to set this in APT is that correct and if so will the camera just use what I set in another external program such as ASICAp that came with camera? Also cannot see this offset when I look at the FITS Header so really no idea if this offset of 15 was actually used or not.

3) When I get round to taking full LRGB or NB data do I need to refocus whenever changing filters? 

4) Is it better to take all the L data in one stint, then R, G etc or is it better to take one L, then R then G and B and back to L etc ?

Sorry if these seem pretty basic questions and I will probably have a lot more to come, this is just where i am at the moment but do not want to give up, I am determined to get some good images with this camera.

I do have many questions but will ask bit by bit If I cannot find the answers myself easily.

 

Steve

 

[vlaiv]

1) I use gain setting of 139 (unity gain) and offset of 64 (maybe a tad "overkill" but that is the way I like it). Mind you, changing gain and offset will require fresh set of darks.

As for exposure length, that really depends on many factors - like how well you guide, how much data you want to store and process, what is likelihood of lost subs, what is the dominant source of noise (like LP levels). For luminance don't worry about over exposing stars (try not to over expose DSO cores and such). Same goes for NB. For color data, you want to keep star cores from saturating, but it can be done with a mix of short subs at the end of the session. Go with at least 1 minute exposures for LRGB and 4-5 for NB.

2) Have look here:

3) Probably yes. There are "parfocal" filter sets out there, but it's best to tweak focus for optimal results when changing filters. You should also tweak focus every now and then during session if temperature changes. Best to monitor your FWHM/HFD values and star sizes in subs and if you see them get bigger - adjust focus. If you have motor focuser, do refocus every hour or so (with SGP for example).

4) Due to need to refocus, it's best to do it in "runs" - do a run of L, then other filters. If you do multiple nights on same target, do L when seeing is the best. I do one run each night. If I feel the need for more data, you can always do additional run next available night.

[cjdawson]

Hiya

as you can see from @vlaiv comment, I'm working on the same things as you are.   Here's the short version of my knowledge so far. (still working on it)

1. Gain for the ZWO ASI1-600MM Pro is 139 for unity.  That's the sweet spot.   As for the ofset value, I'm still working on that one.

2. See the post linked above for the indepth, basically, it's the black level, the higher the value the higher the minimum reading from each pixel will be, this will help to stop the darkest pixels clippinng to black and looking data.

3. You are best off taking a run of each filter, then the flats to go with before changing to the next filter.

4. It doesn't make any difference if you do it in one stint or switch.   That said, flats work best when taken with the lights and you do need to refocus for each filter.

 

Going back to the gain and offset values, I'm also working on an experiment using SharpCap pro's Sensor analysis and the smart histogram feature.  Apparently, these can be used to get SharpCap to suggest what it thinks are the optimal settings for both.  I know that for my gain, it got it about right, but wasn't accurate enough for me.  The offset, I have yet to measure.  Might be worth you checking that out. As well.    once I figure things out, I'm going to post about it so that everyone can see my results.

[Adreneline]

Hello Steve

I am also at the beginning of the learning curve with the ASI1600 and so far I've only managed two imaging sessions. Prior to starting I did find this CN thread very helpful but at the same time confusing; it is very long and jumps about quite a bit but the tables at the beginning helped me get going at least:

https://www.cloudynights.com/topic/573886-sub-exposure-tables-for-asi-1600-and-maybe-qhy163/

I'm in a Bortle 5 area (~20 mags/arcsec^2) and using an f5 system with Ha and OIII filters. I completed one session at gain setting 139, offset 21 and one session at gain setting 200, offset 50, with exposure times of 120s, 180s and 240s; I extrapolated between the two NB tables and tried a range of exposure values). Bearing in mind the target (IC1396) was only 30 degrees above the horizon in a heavily light polluted part of the sky I was very pleased with the results although I think the 200 gain setting worked better; I think I managed to keep the lid on the Garnet star but others may disagree. (I can post the image if you wish but I don't want to divert/derail your post).

I think I understand offset but on the basis I don't want to embarrass myself I will refrain from answering that part of your post; someone who really does understand offset is bound to pop up and answer your question (I will read with interest!).

I'm hoping to experiment in a similar way with broadband, again using the tables to provide me with a starting point.

HTH

Adrian

[RolandKol]

I would add to Mr Vlaiv about exposure times,

Your best friend in this field, is Hystogram, which every AP software has, it summarizes all in one nice graph.

You will be able to go for quite long exposures in NB, especially in HA, but keep in mind, some nights are brighter than others (Moon, little haze and etc).

Simply test your location and your options, - it does not take much time. With NB, shoot a long one 5min, check histogram, if it getting closer to the middle, reduce to 4 or 3 mins.

And again, Darks library, - do you want to have lots of libs per exposure? or or just 1 or 3?

Keep in mind, if you will keep swapping exposures, you will need dark library per each exposure.

I wanted to stick to as little additional files as possible, so tested my exposures in NB with OIII filter, which is most sensitive to the light pollution and ended up with 180sec per sub.

With LRGB 30sec (Red filter probably can go for longer, still learning this field).

I have 3 libs of darks at the moment. 30sec, 180sec and 300sec (for HA), for start

 

As per re-focusing per filter, - 100% yes... needed...

And if fact, you do need to refocus if temperature changes.... So 2 birds with one stone... You finish one filter, go to another and refocus as temp has changed and as you use another filter.

Keep in mind, HA and L filters are the main filters which will give most of the details in the final image, and these two filters have to be focused as good as possible, the rest - "not so essential", but if focus is way off.. you will have another problems...

Also,

Once Moon is full, avoid L and OIII filters... they are too sensitive..

 

 

[teoria_del_big_bang]

Thanks for all the replies so far. I have read but will read through several times as I always do and hopefully take in and understand all the advice. Once again thank you so much for the help I would find this so difficult without such advice.

Steve

[teoria_del_big_bang]

Right I have digested all that info pretty much and very useful it is too so thank-you. I now understand a lot better and think that when the clouds eventually disperse I am a lot better equipped to get some meaningful data.

• I am sticking with LRGB, for now at least, I think unless maybe the moon is too bright then maybe I will have to try NB.
• Looking at starting with Unity Gain (139) with maybe an offset of 60. Given time I may do some testing as suggested to check for clipping to see exactly what offset is needed.
• From my understanding I should probably do L at 1X1 binning and RGB at 2X2. Does that sound reasonable?. But until I become more competent I am thinking maybe just stick to 1X1 for now for all frames. Will that make it easier for me to begin with?
• Exposure I will need to test but thinking around 90S for now. I do not want to push this yet as I have had issues with guiding using PHD2 that I need to bottom - but thats for another thread.
• Camera temperature -20C. I could probably get it lower but thinking when our nights get warmer I might struggle to get lower and will need new dark frames etc so thought I would stick with -20 for now.

Does the above sound a plan to begin with anyway?

Next things to tackle are the calibration frames. I have read quite a few threads on this, with a great deal of info coming from @vlaiv 

Darks- I understand so long as I take these with camera at same temperature, gain and offset I can take these anytime and have a library of master darks of the exposures I normally use from the stacked darks taken.

If I do actually take some lights at 2X2 do I also need darks at 2X2 ?

Flats - Taken same time as lights with same settings using white T-Shirt method + Aurora Flatfield Panel.  short exposure but plenty of them and do it for each filter.

Dark flats - same settings (exposure, gain, offset, Temperature etc) as flats.

Bias - Not Required for 1600MM.

Does this sound correct?

 

Steve

 

 

[RolandKol]

as per Darks,

yes, you will need to have darks Binned in the same rate.

as per -20C, as far as I learned, it is more than enough, Read Noise does not change much at the lower temps, plus

it is very difficult to collect darks at Lower temps in house (your flat/house ambient temp is usually higher than 20C)...

In addition, in warmer nights camera may struggle to get to lower temps. -20C is OK for UK.

P.S.

I did my Darks for NB (300s and 180s - 100subs each - took quite a while.....) in the basement as I was not able to reach -20c in the room, - it was -17'ish at the lowest.....

 

[RolandKol]

as per Bias,

you can collect 200 subs or even more at 0.3sec exposures as ZWO forum suggests, note - not at 0.0003s .

In fact, I did the test with FlatDarks vs  Bias at 0.3sec Vs Bias 0.0003s, - stacking results with FlatDarks were better, but 0.3Sec Bias, gave quite good results with correctable field, Bias 0.0003 have not damaged the final image, but flats were not correcting properly and post processing would be much more difficult in comparison to the rest.

So Experiment and check what best fits you... Even if FlatDarks gives the best results,

 I am a bit too lazy to play with FlatDars for each filter (I use 7) while stacking as the procedure becomes very manual and time consuming, so I still use 0.33sec MasterBias and Pi Script for stacks.

 

 

[RolandKol]

Hrrrrrrrrrrrr

@vlaiv I have not tested without BIAS at all! :)))))))))))))))))

Why my head is always empty...

[vlaiv]

That pretty much sounds ok, here are few further tips:

Re binning - don't use capture time binning. Simply not necessary with CMOS sensors as software binning does the same and you have more flexibility. Capture your data and calibration frames at x1. You can bin your subs later, after calibration and before stacking if there is need for it.

Need for binning will depend on your working resolution / pixel scale, and also what sort of FWHM you are getting. Here is a simple rule that will give you best results. Do your stacking of x1 bin data - you don't need to do anything fancy, just simple average stack will do (for speed). Measure FWHM of your stack in arc seconds (convert from your imaging resolution). Divide resulting value with 1.6 and this should be your sampling resolution. If you are way off - bin your data to get closer match.

For example, you are imaging at 0.7"/px, and you find that your stack has 2.8" FWHM stars. From above, optimum sampling rate is 1.75"/px. You can now choose between bin x2 and bin x3. Bin x2 will give you 1.4"/px - slight over sampling but you get larger image. Bin x3 will get you 2.1"/px - slight under sampling, image will be smaller but resulting stack will have better SNR and your image will go deeper with less noise.

-20C for me is regular temperature, but do bear in mind that you might need "summer" set. I shoot at -15C at summer, because nights can be as warm as 25C and you can't get cooling down to -20C although delta T is quoted to be 45C. Often I can reach something like -19C but that makes mismatch between darks and lights, so better keep it at -15C for summer if you get warm nights.

For flats - these don't need to be taken at the same settings as lights, you can change gain, offset and you will certainly change exposure time. Flat darks need to match flats in all parameters (like regular darks match regular lights). However, there is no real need to change parameters between flats and lights, so keeping gain and offset the same for everything is simply easier thing to do and not have to worry about it.

For darks, yes, build your dark lib in basement, or in house (if you don't have permanent setup) - but do make sure you have your camera properly "sealed" for light leaks. Regular cap / sensor cover will handle visible light, but camera is also sensitive in IR - some people use tin/aluminum foil as another layer of protection - my method is to place camera "face down" (sensor down) on wooden table (of course with cap on) and do darks like that.

 

[teoria_del_big_bang]

3 minutes ago, RolandKol said:

Hrrrrrrrrrrrr

@vlaiv I have not tested without BIAS at all! :)))))))))))))))))

Why my head is always empty...

With my limited knowledge I obviously cannot comment but have read several times about not using Bias frames with the ZWO1600MM

Steve

[vlaiv]

4 minutes ago, RolandKol said:

Hrrrrrrrrrrrr

@vlaiv I have not tested without BIAS at all! :)))))))))))))))))

Why my head is always empty...

Here is simple answer why bias is not needed:

bias = bias_signal

darks = bias_signal + dark_signal

lights = bias_signal + dark_signal + light_signal

Using bias:

master_dark = stack(dark - bias) = stack(bias_signal + dark_signal - bias_signal) = stack(dark_signal)

calibrated_light = (bias_signal + dark_signal + light_signal) - bias_signal - stack(dark_signal) = dark_signal + light_signal - stack(dark_signal) = light_signal

Not using bias:

master_dark = stack(bias_signal + dark_signal)

calibrated_light = (bias_signal + dark_signal + light_signal) - master_dark = (bias_signal + dark_signal + light_signal) - stack(bias_signal + dark_signal) = bias_signal - bias_signal + dark_signal - dark_signal + light_signal = light_signal

Just mind that stack() above means just average, and for signal that aligns perfectly (as does dark and bias - no need to shift subs to align them) - it is just the same thing.

[vlaiv]

5 minutes ago, teoria_del_big_bang said:

With my limited knowledge I obviously cannot comment but have read several times about not using Bias frames with the ZWO1600MM

Steve

Yes, bias is 1) not necessary, 2) causes problems, as @RolandKol noted above:

16 minutes ago, RolandKol said:

In fact, I did the test with FlatDarks vs  Bias at 0.3sec Vs Bias 0.0003s, - stacking results with FlatDarks were better, but 0.3Sec Bias, gave quite good results with correctable field, Bias 0.0003 have not damaged the final image, but flats were not correcting properly and post processing would be much more difficult in comparison to the rest.

For example - flats not correcting properly and such. This is because bias in "pure bias" files and bias signal in darks, lights and flats is different in intensity. Hence recommendation not to go with pure bias, but instead use 0.3s "Bias" - which is not Bias at all, but rather very short dark (that causes problems again because it contains a bit of dark current signal that should not be there).

[RolandKol]

8 minutes ago, teoria_del_big_bang said:

With my limited knowledge I obviously cannot comment but have read several times about not using Bias frames with the ZWO1600MM

Steve

I Do agree, and I mentioned that, FlatDarks are better way to go.
 

3 minutes ago, vlaiv said:

Here is simple answer why bias is not needed:

bias = bias_signal

darks = bias_signal + dark_signal

lights = bias_signal + dark_signal + light_signal

Using bias:

master_dark = stack(dark - bias) = stack(bias_signal + dark_signal - bias_signal) = stack(dark_signal)

calibrated_light = (bias_signal + dark_signal + light_signal) - bias_signal - stack(dark_signal) = dark_signal + light_signal - stack(dark_signal) = light_signal

Not using bias:

master_dark = stack(bias_signal + dark_signal)

calibrated_light = (bias_signal + dark_signal + light_signal) - master_dark = (bias_signal + dark_signal + light_signal) - stack(bias_signal + dark_signal) = bias_signal - bias_signal + dark_signal - dark_signal + light_signal = light_signal

Just mind that stack() above means just average, and for signal that aligns perfectly (as does dark and bias - no need to shift subs to align them) - it is just the same thing.

I will do some experiments anyway.

If my stacks will be more or less the same as with Bias 0.3s and without Bias at all...

I will not use Bias and Flatdarks all together
 

[teoria_del_big_bang]

5 minutes ago, vlaiv said:

Re binning - don't use capture time binning. Simply not necessary with CMOS sensors as software binning does the same and you have more flexibility. Capture your data and calibration frames at x1. You can bin your subs later, after calibration and before stacking if there is need for it. Makes good sense - great tip ? 

Need for binning will depend on your working resolution / pixel scale, and also what sort of FWHM you are getting. Here is a simple rule that will give you best results. Do your stacking of x1 bin data - you don't need to do anything fancy, just simple average stack will do (for speed). Measure FWHM of your stack in arc seconds (convert from your imaging resolution). Divide resulting value with 1.6 and this should be your sampling resolution. If you are way off - bin your data to get closer match. 

For example, you are imaging at 0.7"/px, and you find that your stack has 2.8" FWHM stars. From above, optimum sampling rate is 1.75"/px. You can now choose between bin x2 and bin x3. Bin x2 will give you 1.4"/px - slight over sampling but you get larger image. Bin x3 will get you 2.1"/px - slight under sampling, image will be smaller but resulting stack will have better SNR and your image will go deeper with less noise. Sounds simple but so new to all this I need to get some data and try all this. I take it I can do this in most stacking software. Up to press I have mostly used Nebulosity4 / DSS but moving to use PI now. 

-20C for me is regular temperature, but do bear in mind that you might need "summer" set. I shoot at -15C at summer, because nights can be as warm as 25C and you can't get cooling down to -20C although delta T is quoted to be 45C. Often I can reach something like -19C but that makes mismatch between darks and lights, so better keep it at -15C for summer if you get warm nights. Yes I did think I might need this although we had a good hot summer in UK last year so I do not expect another one for at least 20 years ?

For flats - these don't need to be taken at the same settings as lights, you can change gain, offset and you will certainly change exposure time. Flat darks need to match flats in all parameters (like regular darks match regular lights). However, there is no real need to change parameters between flats and lights, so keeping gain and offset the same for everything is simply easier thing to do and not have to worry about it.

For darks, yes, build your dark lib in basement, or in house (if you don't have permanent setup) - but do make sure you have your camera properly "sealed" for light leaks. Regular cap / sensor cover will handle visible light, but camera is also sensitive in IR - some people use tin/aluminum foil as another layer of protection - my method is to place camera "face down" (sensor down) on wooden table (of course with cap on) and do darks like that. My scope had a metal cap (Pretty sure it is aluminium) I thought that would suffice, can IR still get to the sensor , should I do as your suggestion just to make sure?

 

 

[vlaiv]

2 minutes ago, RolandKol said:

I Do agree, and I mentioned that, FlatDarks are better way to go.
 

I will do some experiments anyway.

If my stacks will be more or less the same as with Bias 0.3s and without Bias at all...

I will not use Bias and Flatdarks all together
 

Do use flatdarks - these are necessary for proper flat correction.

With flat correction you need only light signal in flats to get relative sensitivity between pixels. Any other signal will skew this and you will get either over or under correction.

Take for example using 0.3s "Bias" instead of proper flat darks.

Let's say your flats are 1s long.

Flats will consist of: bias_signal + 1s x dark_signal + light_signal

now you calibrate those with 0.3s bias which is: bias_signal + 0.3s x dark_signal

what you have left is this: bias_signal + 1s x dark_signal + light_signal - bias_signal + 0.3s dark_signal = 0.7s dark_signal + light_signal

So you don't have pure light signal in your flats. What does it do? Imagine you have regular part of image where 100% of light falls, and you have shadow where only 80% falls.

This will be "recorded" in flat as: 1.0 and 0.8 respectively.

In light you might have something like 100e and 80e respectively for signal 100e strong. If you divide the two, you will get what you expect - flat corrected image 100e/1 = 100e, 80e / 0.8 = 100e (no variation in brightness)

But look what happens if you left small amount of dark_signal in there

flat will no longer be 1 and 0.8, but for example 1.01 and 0.81, and let's scale that back to 0-1 range 1.01 / 1.01, 0.81/1.01 = 1, ~0.80198

Correcting 100e and 80e will now produce 100e / 1.0, 80e / ~0.80198 = 100e, ~99.735e

We have slight under correction in this case! Signal is no longer fully corrected and uniform, there is small shading from the shadow left on it as described by 99.735e signal.

[spillage]

I have found that using a light panel for my flats makes things much easier and I can control the brightness as needed. I think in APT I just choose an adu of 20k but I have seem other number floating around both higher and lower.

[vlaiv]

1 minute ago, teoria_del_big_bang said:

My scope had a metal cap (Pretty sure it is aluminium) I thought that would suffice, can IR still get to the sensor , should I do as your suggestion just to make sure?

If you are going to leave it on scope, and your scope is all metal - I think you will be just fine. In odd case you get under or over correcting flats - we can do analysis why it's happening and if likely source is light leak - then you will need to deal with it, but I doubt it will be the case.

[teoria_del_big_bang]

Great responses many thanks. 

A lot of the info seems very daunting and complicated to me at this stage. I hope as I put it into practice it all becomes clearer. So long as each session I understand a bit more of the puzzle I am happy and do not expect to master any of this overnight but am grateful for all the help and look forward to little improvements as time goes by.

Well fingers crossed anyway ?

Steve

[spillage]

You could try doing what I do...I am sure people will be shouting at me, but just run of 30-50 darks and 30-50 flats (just using your luminace filter) and the same number of dark flats.

 

[teoria_del_big_bang]

8 minutes ago, spillage said:

You could try doing what I do...I am sure people will be shouting at me, but just run of 30-50 darks and 30-50 flats (just using your luminace filter) and the same number of dark flats.

 

I think you will be safe, I have not witnessed much in the way of shouting on this forum - generally all well mannered even when a difference of opinion ? 

I guess it all comes down to what works for you and often with flats anyway is what you have time to do on the night, or the morning after when you finish the session.

Steve

[teoria_del_big_bang]

Had a practice at taking some flats today using the "CCD Flats Aid" in APT.

With Gain 139, offset 60 it calculated an exposure time of 2 seconds for all the filters and obtained ADU values from 20000 to 25000 depending on filter.

Unstretched looks like this. Does this seem about right exposure for that gain?

 

Steve

[RolandKol]

You cannot be sure from one Sub, but looks like will be OK.

The final test will show,

Just take around 40 subs per filter 

The more, the better (like always  )

[teoria_del_big_bang]

Cheers Roland,

I sort of thought it would not be possible to say that is okay or not but more wanted some thoughts to the exposure time and ADu were around what they should be rather than plodding on with something that was obviously incorrect.

I did have some issues as just using my light panel at these gains the exposures were around 0.1 S or less but when I stretched the FITs obtained they has a lot of banding on them which after a bit of a search other people had issues with and I believe is down to the short exposure time and the LED panel flickering (albeit very fast flickering). So ended up with about 4 layers of white paper in between the panel and scope to get the results above.

I will give all this a go when the clouds let me. Looks like they are due to roll in now so tonight may be a no no ? 

Steve