Gain Settings - Zwo ASI 1600mm Cool

[Doug64]

Hi,

I've just bought the above camera to try some mono CMOS imaging having previously used a modified DSLR.

I've started reading about imaging / processing with this camera and it's going to be a big learning curve.

I'm already confused with regards to the gain settings.

I've read all sorts of settings people appear to use:- Unity Gain  - 139 offset 21, Gain 76 offset 21, Gain 76 offset 15, Gain 200 offset 30, Gain 200 offset 15, Gain 300 and also gain 0.

I will be using this camera with my ED80 with flatter/reducer and motorised filter wheel.  I live in a city with very bad light polution so was wondering if anyone could give me some advice about the gain and offset settings or is it just trial and error.

I have LRGB  and Ha OIII SII filters.

Thanks

Doug

[Adreneline]

Hi Doug,

I purchased an ASI-1600 Pro in late February and so far have had little time to use it - weather and all.

I was also confused by the various settings and found this article on CN site to be very helpful:

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

I have used the tables as a starting point and so far have been very pleased with the results I have achieved, however, there is clearly much in the way of experimentation to be done. I live in a Bortle 5 area (and with all the new build I think going on Bortle 6!) - 19.93 Magnitude.

Apologies if you have already seen this and I am telling you nothing new.

Adrian

Edited July 2, 2019 by Adreneline
Typo

[Doug64]

Hi,

Thanks for the reply.

I have seen the charts you refer to and was initially going to work from them but the more reading I did the more I found people using different gain / offset settings.

This us why I got confused.

I live in a Bortle 7 area.  18.79 Magnitude.

Regards

Doug

Edited July 2, 2019 by Doug64

[AbsolutelyN]

I think unity gain is a good starting point - 139. I vary gain depending on what I'm imaging and am still very much experimenting and figuring out what works. I found gain 0 really nice on M13 to hold highlights in the core, unity gain as a great all rounder and higher settings such as 200 great for narrowband.   

[eshy76]

Another vote for unity gain (139) to start off with. I've had the Pro version for nearly a year and it is only now that I am experimenting with other gain settings.

As you say, the move from DSLR has many different aspects to it without complicating it by fiddling with the default gain which will give you great all round results as mentioned above. 

That sub exposure table/thread is a good place to start to judge sub length. Good luck! It's a great cam!

[Doug64]

Thanks for all the replies.

I'll start off using unity gain and may try gain 200 with my narrowband filters, this way I can learn more about the software and filters etc.

Doug

[spillage]

Im in the position as @eshy76 and only last night did I go out and try different gain settings. I am not sure why but the tables on the CN forum suggest you can do over 20 mins broadband at 300 but my images were over exposed at 10 mins so I guess I either need to lower gain or decrease exposure times?. I am probably wrong but I have found that the longer you can expose for the smoother the image will be. 

I sure @vlaiv uses a 1600 and is a genius when it comes to the maths but I must admit I get a bit lost with it all.

[AbsolutelyN]

You need to take local conditions into account such as darkness of sky and light pollution. I've been doing narrowband last couple of nights and have gone with unity gain and 5 min subs as I don't want to blow highlights and sky doesn't really get properly dark. 

[spillage]

I do not want to hijack the thread but would you say high gain lower exposure times or lower gain and longer exposure times? I understand that being able to take more exposures at a shorter time allows for tracking and being able to be more ruthless deleting poor lights. If I can manage another late night and the sky is clear I will have another play around tonight.

[vlaiv]

Some good advice given on gain settings. I'll just expand a bit on how to choose it depending on your conditions, and I'll give advice on offset.

With gain you are balancing two things really - read noise and well depth.

Read noise bit is important in two ways. Firstly, it is of course source of noise, and should be kept as low as possible. It is however probably the smallest contributing noise in most cases. Only case when read noise is significant is if you have incredibly dark skies or you use narrow band filters (like really narrow 3nm) and imaging very faint target. In this case it will become dominant noise source and you might want to reduce it as much as possible.

Second way read noise impact things is single long vs multiple short exposures. Difference between the two is read noise. If read noise were 0, there would be absolutely no difference between two approaches, but as read noise becomes higher there will be more and more difference between many short subs and a few long ones (all of this depends on other factors/noise sources as well).

Depth of well is inverse of gain, so larger the gain you use - less pixel well depth you will have (and saturate in less time) - these two are fighting each other, because for narrow band, for example you want to raise gain (to have less of read noise) and you want longer exposures - but higher gain will lower full well capacity of pixels and longer exposure will saturate in some cases - you need to find a balance.

There are no exact settings that will work for everyone because all of the above depends on sky brightness but also on target brightness and scope used - how much aperture and what is working resolution (sampling rate - or arc seconds per pixel).

For most targets and LRGB, you can use unity gain and 1-4 minutes (if you use longer exposure, you star color will suffer on brighter stars, so you'll want to use a few short ones as filler for color - with luminance you don't need to worry about that - just make sure target is not clipping). For NB you'll want to use 4-5 minutes and higher gain - look at gain/read noise chart and select one that will not eat too much into full well capacity but will still provide you with low read noise.

Above is balancing act - so try with "recommended" settings, but tweak to your particular conditions.

Offset on the other hand is something that you want to do right, and there is right and wrong setting. With offset you aim to avoid clipping "to the left" - or trying to avoid 0 values in the image (in both darks and lights). Don't be afraid to go with high offset - it's impact on full well will be minimal, but it is safer bet that you won't get that clipping that will mess up your calibration. I personally use offset of 64, but anything higher than 50 is a good starting point.

There is a way to determine good offset for chosen exposure length (and it will probably be valid for range of exposure lengths) - take dozen or so darks, stack them with minimum method (not average, and no sigma clip and such, just straight minimum), do stats on resulting stack and if you find that you have pixels that are 16 in value (or 1 if you bit shift your images to 12 bits from 16bits) - or stats on that stack give you minimum of 16 (meaning there is at least one pixel with that value) - raise offset.

[vlaiv]

9 minutes ago, spillage said:

I do not want to hijack the thread but would you say high gain lower exposure times or lower gain and longer exposure times? I understand that being able to take more exposures at a shorter time allows for tracking and being able to be more ruthless deleting poor lights. If I can manage another late night and the sky is clear I will have another play around tonight.

My preference is to have shorter exposure times for precisely the things you mentioned - any sort of interference with capture process will ruin less data with short exposures - meaning plane / satellite trails, gusts of wind, spells of very poor seeing - what ever makes you discard your subs - you'll discard less data with shorter subs.

To determine what is good exposure length you really need to do some calculations - as each case is different. I'll briefly explain what happens - because of read noise and how "strong" it is compared to other noise sources (LP noise, target shot noise, and thermal noise - which is generally low with cooled camera so in most cases we don't even need to consider it). Each time you stack a sub you add one "dose" of read noise.

If we for example do one hour of exposure in 60 x 1 minute - your result will end up with 60 "doses" of read noise (all others will be the same between the two), but if you use 10 minute subs - you'll end up with only 6 "doses" of read noise in final stack.

Now depending on how high single "dose" of read noise is in comparison to other noise sources - this can create either significant or minimal impact. There is a point of diminishing returns when you increase exposure length and in turn decrease total number of subs (to keep total imaging time the same) - and difference in noise between the two is imperceptible to human eye.

There is no general rule for this, as it will depend on bunch of things - target brightness, LP levels, what sort of scope you use (aperture at resolution) and such. You can do some complex calculations to approximate where this point of diminishing returns is for your setup and base decision on that, or you can just do what most people does - take some guideline values and adjust for your conditions.

Just for comparison - CCD rule is something like 5-10 minutes for RGB. But CCD cameras have 5-10e read noise. CMOS has read noise about 5 times less than that. Although dependence is not linear, you can see that recommended values for CMOS sensors are in 1-2 minute range for RGB (so as read noise is 5 times less, so will be exposure length - but I again stress - relationship is not linear and simple as that). Same goes for NB - most people use 20+ minutes for NB on CCD sensors. With CMOS you can use 4+ minutes

[Doug64]

12 hours ago, vlaiv said:

Some good advice given on gain settings. I'll just expand a bit on how to choose it depending on your conditions, and I'll give advice on offset.

With gain you are balancing two things really - read noise and well depth.

Read noise bit is important in two ways. Firstly, it is of course source of noise, and should be kept as low as possible. It is however probably the smallest contributing noise in most cases. Only case when read noise is significant is if you have incredibly dark skies or you use narrow band filters (like really narrow 3nm) and imaging very faint target. In this case it will become dominant noise source and you might want to reduce it as much as possible.

Second way read noise impact things is single long vs multiple short exposures. Difference between the two is read noise. If read noise were 0, there would be absolutely no difference between two approaches, but as read noise becomes higher there will be more and more difference between many short subs and a few long ones (all of this depends on other factors/noise sources as well).

Depth of well is inverse of gain, so larger the gain you use - less pixel well depth you will have (and saturate in less time) - these two are fighting each other, because for narrow band, for example you want to raise gain (to have less of read noise) and you want longer exposures - but higher gain will lower full well capacity of pixels and longer exposure will saturate in some cases - you need to find a balance.

There are no exact settings that will work for everyone because all of the above depends on sky brightness but also on target brightness and scope used - how much aperture and what is working resolution (sampling rate - or arc seconds per pixel).

For most targets and LRGB, you can use unity gain and 1-4 minutes (if you use longer exposure, you star color will suffer on brighter stars, so you'll want to use a few short ones as filler for color - with luminance you don't need to worry about that - just make sure target is not clipping). For NB you'll want to use 4-5 minutes and higher gain - look at gain/read noise chart and select one that will not eat too much into full well capacity but will still provide you with low read noise.

Above is balancing act - so try with "recommended" settings, but tweak to your particular conditions.

Offset on the other hand is something that you want to do right, and there is right and wrong setting. With offset you aim to avoid clipping "to the left" - or trying to avoid 0 values in the image (in both darks and lights). Don't be afraid to go with high offset - it's impact on full well will be minimal, but it is safer bet that you won't get that clipping that will mess up your calibration. I personally use offset of 64, but anything higher than 50 is a good starting point.

There is a way to determine good offset for chosen exposure length (and it will probably be valid for range of exposure lengths) - take dozen or so darks, stack them with minimum method (not average, and no sigma clip and such, just straight minimum), do stats on resulting stack and if you find that you have pixels that are 16 in value (or 1 if you bit shift your images to 12 bits from 16bits) - or stats on that stack give you minimum of 16 (meaning there is at least one pixel with that value) - raise offset.

 

[Doug64]

Hi Vlaiv,

Thanks for the comprehensive reply, some of it was a bit over my head but I'm starting to get the idea about gain and offset.

I understand little of the math, I knew this was going go be a massive learning curve and I'm not wrong.

Thanks again 

Doug

Edited July 3, 2019 by Doug64

[eshy76]

All very sensible tips here. I would just add that I don't really use those exposure tables for exposure time per se - I used the underlying maths to work out what ADU I should, theoretically, be aiming for per sub.

So at unity gain and default offset, the maths chucks out a minimum ADU per sub for me of about 1400. You can then use the readout in your capture software to judge the right exposure times. It's a one-time thing unless you image in different places with different skies.

So I aim for this when judging exposure time, which is dependent on light pollution. With my bortle 7-8 skies this works out for me at about 30-60 seconds for RGB subs (much less for Lum) and 300 seconds for narrowband subs. Results seem good, which is, of course, the acid test!

[symmetal]

To avoid having a large collection of darks for different gains and exposures I just use unity gain all the time (as suggested by vlaiv ) and an offset of 56 to avoid black clipping.

For exposure length I use a similar method to eshy76 where you expose until the ADU level of the sky background in your image is 10 times the square of the read noise. This makes the read noise insignificant. CN forums have the maths to calculate this ADU. At unity gain offset 56 this works out at 1386 ADU (16 bit). Very close to eshy76's value. With my bortle 3 skies this works out at around 60s for L and 180s for RGB. Currently with no astro dark these exposure values are around half that. With narrowband I won't reach this swamping background ADU unless I expose for an hour or more so I settle for 360-480s normally and accept that the read noise will have some effect.

Exposing for longer than this swamping ADU value has no real benefits, and just causes more clipping of stars and a reduction in dynamic range, as the histogram just gets shifted to the right.

Alan

[Doug64]

Thanks for the replies,

Now I'm really confused  the more I read the more confused I get.

I think using unity gain will be a very good starting point for me for all filters as I then only have to work out exposure time.

I'm not use to all this ADU, well depth etc so I think I have a lot to learn, at least I can get imaging once I have clear skies using Unity gain.

Can anyone point me to the equation for working out exposure time and also where do you get the ADU level, is it from the histogram after taking the image.

Thanks

Doug

Edited July 4, 2019 by Doug64

[symmetal]

Hi Doug,

Here's a CN post describing the read noise swamping by sky background calculations. Look for the Jon Rista entries.

I made an Excel chart calculating these values for various gain/offset values for the ASI1600. Initially I used the varying offset values recommended by Zwo but changed to a fixed offset for all gain settings (for simplicity) which avoided black clipping as the Zwo offsets were too low. The Read Noise and Gain graphs are plotted from the column entries and mimic the Zwo graphs from the ASI1600 manual.

During astro dark, take an image of a star field which doesn't contain large areas of bright nebulosity. The image sky background ADU can be read as the median value of the image statistics panel which your capture program should display. The mean value should be pretty much the same value as the median.

There are different opinions as to which calculation method (I've shown four) should be used for the 'swamping the read noise ADU value' and I use 10 x RN^2 which is the last column. The calculation for the 1386 ADU value in box J12 is shown in the equation at the top. If you use offset 50 just subtract 96 from these ADU values, ie. (56-50) * 16

Adjust your exposure duration until the sky background ADU is around the value in the table and that is the maximum exposure you need to use for that filter. The RG and B exposures should be fairly similar while the L should be about 1/3 that value. During astro dark, use these same exposures for your targets and you should be close to optimum. If you move to darker or lighter site you need to re-evaluate these exposure times. As currently there is no astro dark in the UK, exposures don't need to be as long to reach these ADU values. If you're in a light polluted site the exposure duration values will be a lot lower than than at a dark site.

Hope this helps Doug and hasn't left you more confused. 

Alan

Edited July 5, 2019 by symmetal

[Doug64]

8 hours ago, symmetal said:

Hi Doug,

Here's a CN post describing the read noise swamping by sky background calculations. Look for the Jon Rista entries.

I made an Excel chart calculating these values for various gain/offset values for the ASI1600. Initially I used the varying offset values recommended by Zwo but changed to a fixed offset for all gain settings (for simplicity) which avoided black clipping as the Zwo offsets were too low. The Read Noise and Gain graphs are plotted from the column entries and mimic the Zwo graphs from the ASI1600 manual.

During astro dark, take an image of a star field which doesn't contain large areas of bright nebulosity. The image sky background ADU can be read as the median value of the image statistics panel which your capture program should display. The mean value should be pretty much the same value as the median.

There are different opinions as to which calculation method (I've shown four) should be used for the 'swamping the read noise ADU value' and I use 10 x RN^2 which is the last column. The calculation for the 1386 ADU value in box J12 is shown in the equation at the top. If you use offset 50 just subtract 96 from these ADU values, ie. (56-50) * 16

Adjust your exposure duration until the sky background ADU is around the value in the table and that is the maximum exposure you need to use for that filter. The RG and B exposures should be fairly similar while the L should be about 1/3 that value. During astro dark, use these same exposures for your targets and you should be close to optimum. If you move to darker or lighter site you need to re-evaluate these exposure times. As currently there is no astro dark in the UK, exposures don't need to be as long to reach these ADU values. If you're in a light polluted site the exposure duration values will be a lot lower than than at a dark site.

Hope this helps Doug and hasn't left you more confused. 

Alan

Hi,

Thanks very much for your explanation and chart, it's really helped me understand, I think .

If I've understood everything for Gain 139 with 56 offset my background target ADU is 1386.  In order to find exposure times I set up my telescope etc and at astro dark I'll need to take some test exposures at varying exposure times until I get one where the background level is 1386 and I do this for each filter and record times etc.

Just a few questions if you don't mind:-

I'm going to be using Astrophotography Tools, if you know this software does this allow me to check background level. (I'm just starting to learn this and haven't taken an image using it).

If I image from my garden once I have these timings will they basically remain the same for each session and then I only need to check them again if I move to dark sight etc or in winter with astro dark.

Why do you choose offset 56 is it to stop the clipping.

Thanks again

Douh

[spillage]

14 minutes ago, Doug64 said:

I'm going to be using Astrophotography Tools, if you know this software does this allow me to check background level. (I'm just starting to learn this and haven't taken an image using it).

Its been a while but I am sure when you load the image you click the tools tab and choose pixel aid.

[Doug64]

14 minutes ago, spillage said:

Its been a while but I am sure when you load the image you click the tools tab and choose pixel aid.

Thanks

[spillage]

@symmetal That spreadsheet is really really useful and I have been looking for something similar for ages as my experience with excel and maths is not the best. Would you be prepared to upload the spread sheet or at least this should maybe go into a new thread and be made a sticky by the mods.

[spillage]

@Doug64 If you are imaging from only one location you could always upload the .fits and have one of us open it in sgp to get the median value for you.

[Doug64]

45 minutes ago, spillage said:

@Doug64 If you are imaging from only one location you could always upload the .fits and have one of us open it in sgp to get the median value for you.

Thanks for the reply.

Once I have everything set up, just getting APT to link to all my equipment, I'll take an image and see if I can get the background median value using APT, if not I may have to do as you suggest.

90% of my imaging will be done from my back garden, this is why I've just bought the mono camera as I'll need to use Narrowband most of the time due to light pollution (Bortle 7).  I do take my telescope etc with me twice a year to Kelling, and to other places such as the flooded SGL star party and Galloway.

I'm really looking forward to using this camera once I've learnt how to use it.

Doug

[symmetal]

5 hours ago, Doug64 said:

If I image from my garden once I have these timings will they basically remain the same for each session and then I only need to check them again if I move to dark sight etc or in winter with astro dark.

Yes, that's right. It's more convenient to use the same exposure time for RGB so choose an exposure value which is close to the 'optimum' ADU value for all three colours. Astro dark isn't a fixed darkness but is a threshold value so images taken in the middle of the astro dark period will have a darker sky background than ones taken at the beginning or end of the period so wait until an hour or so after astro dark begins before doing the tests to get an 'average' astro dark sky background level and use this exposure setting for all your images.

These are really only useful for LRGB imaging as doing narrowband the required exposures to swamp the read noise would probably be longer than 30mins unless you're in a badly light polluted site. Take a sample narrowband image at your usual exposure duration and see how the sky background compares to the 'optimum' ADU value. If it does exceed it then you should use a shorter duration but at most sites you will probably be well below it.

A very useful tool for examining image statistics is Fits Liberator. This displays the mean and median values among others and gives a useful histogram of the image. Taking a very short dark frame (effectively a bias frame) the histogram will indicate whether the image is black clipped and is useful to find out your optimum offset value to avoid clipping at the gain used.  The higher the gain used the higher the offset needed but it's simpler to find the offset needed at gain 300 (or the highest gain you'll use) and use this for all gain settings. I found offset 56 just avoided clipping at gain 300. If I was only going to use unity gain I could use a lower offset and avoid clipping but the loss in dynamic range of using a higher offset than needed is very small.

4 hours ago, spillage said:

@symmetal That spreadsheet is really really useful and I have been looking for something similar for ages as my experience with excel and maths is not the best. Would you be prepared to upload the spread sheet or at least this should maybe go into a new thread and be made a sticky by the mods.

Yes, I can upload the spreadsheet file so you can fill in your own gain/offset values if you wish. I'll do that later today. I can also put it in a new thread if there is enough interest. 

Alan

Edited July 5, 2019 by symmetal

[symmetal]

I've uploaded the Excel spreadsheet to my MediaFire Cloud Storage account. I've uploaded two versions. The .xlsx is for the latest versions of Excel which I used, while the .xls version should work with older versions of Excel that don't recognise the newer format. The free Open Office suite of programs should read the .xls file without problems if you don't have Microsoft Office. I don't know if it recognises the newer format.

ASI1600 Sky background ADU chart (.xlsx)

ASI1600 Sky background ADU chart (.xls)

If you want to enter your own values for gain and offset in columns one and two, you also need to enter the Read Noise in electrons (e-) in column three, and Gain in electrons/ADU (e-/ADU) in column four as read from the ASI1600 graphs in the manual which correspond to the column one gain. The values you've entered in columns three and four will be plotted on the graphs in the spreadsheet so you can see if the values are good as the curves should still retain their overall shape. I've posted the Zwo graphs below if you've misplaced the manual. 

I included a graph for the offset as originally I used the Zwo gain and offset values where the offset varied with the gain and the graph shape indicated if the values you entered followed the Zwo curve. If you use a fixed offset value then the offset graph doesn't tell you anything useful.

Hope Mark and Doug and anyone else finds this useful. 

Alan