Over the last couple of years I have conducted some experiments to determine what my imaging system could capture compared to a fast imaging system such as Mike Sidonio's. I have found that whilst my system performs well on distant objects such as galaxies where I get quite good detail and good contrast, it really struggles when it comes to imaging outer halos on planetary nebula such as IC5148 and NGC3132 and things like the optical jets on NGC1097. I put this down to two things. Firstly, the slower focal ratio would and does have a significant impact on gathering those photons. Secondly, the sensor tech I have is aging and impacts also.
This image is composed of 45 hours of integration with 30 hours in Ha. The outer chevrons are barely visible in the colour image and even viewing the Ha data itself shows only the barest of signal. This is despite doing subs at 20 minutes, which should be near optimum for my system. I had considered doing 30 minutes subs but was concerned about the core detail being totally blown out.
What this shows is that to get similar detail to a 12" f3.8 system I need to image about 6x or more the amount of hours for any channel to get similar detail on these very faint outer halos.
That is some really nice detail Paul! It is an interesting comparison Re-Fainter halos.
Both scopes are 12". F/3.8 vs F/8. Pixel size is 9 vs 4.54. QE is ~ .6 vs maybe .45 at a guess.
Assuming both systems are read noise limited, you would need 2x the exposure time to get the same SNR on fainter areas. I imagine that with a 11002 sensor it would need far more than 20 minutes to reach that point at F/8 though. At a guess I would say that it is the sensor read noise that is the limiting factor.
just ran both systems through the spreadsheet. It agrees with your observation - with 20 minute subs on both, your system requires ~6x the time to get to the same depth in Ha.
As Colin pointed out, the f8/9micron and f3.8/4.5micron optics configurations are essentially equivalent, but your 11002 has about half the Ha QE of the 694 so you need 2x the exposure to compensate. The biggest killer though is that the 11002 has about double the read noise of the 694 and, since neither system is sky limited at 20 minutes under dark skies, the read noise makes a real difference - you need an extra ~3x exposure to compensate. You don't have many options to do anything about this, short of using much longer subs so that you introduce less read noise. However, you would need maybe 10 hour subs to really help a lot, so that has practical and dynamic range limitations. A camera with lower read noise would help, but it must have ~9 micron pixels to match the rest of your system. The only solutions I can think of would be one of the Sony 694/814 chipped cameras used in 2x2 bin to get large pixels (but only 1.5/2mp) - or maybe even a ZWO/QHY 16mp CMOS camera with software binning for 4mp. Neither would be much good for wider fields, but good for smaller objects such as PNs. If you want to try this approach, let me know - you can experiment with my 694 if that would help. With the 694 in bin2, you would get back the missing 6x factor using your existing scope.
[Ray beat me to it with a more comprehensive answer, but we said similar things ]
Hi Paul,
Your system is approx one third the speed of Mike's based on the small difference in image scale, a little more obstruction and the larger difference in QE (for Ha the QE of the KAI-11002m is down around 30%.)
I'd agree with Colin that with 20 min NB subs you're going to be taking a big hit from read noise, which is up around 12e- for that sensor.
Longer subs would help, but you're at a big disadvantage based on QE and read noise. It is definitely the sensor that is the weak link in the chain.
That is some really nice detail Paul! It is an interesting comparison Re-Fainter halos.
Both scopes are 12". F/3.8 vs F/8. Pixel size is 9 vs 4.54. QE is ~ .6 vs maybe .45 at a guess.
Assuming both systems are read noise limited, you would need 2x the exposure time to get the same SNR on fainter areas. I imagine that with a 11002 sensor it would need far more than 20 minutes to reach that point at F/8 though. At a guess I would say that it is the sensor read noise that is the limiting factor.
Quote:
Originally Posted by Shiraz
very high res image Paul - nice result.
just ran both systems through the spreadsheet. It agrees with your observation - with 20 minute subs on both, your system requires ~6x the time to get to the same depth in Ha.
As Colin pointed out, the f8/9micron and f3.8/4.5micron optics configurations are essentially equivalent, but your 11002 has about half the Ha QE of the 694 so you need 2x the exposure to compensate. The biggest killer though is that the 11002 has about double the read noise of the 694 and, since neither system is sky limited at 20 minutes under dark skies, the read noise makes a real difference - you need an extra ~3x exposure to compensate. You don't have many options to do anything about this, short of using much longer subs so that you introduce less read noise. However, you would need maybe 10 hour subs to really help a lot, so that has practical and dynamic range limitations. A camera with lower read noise would help, but it must have ~9 micron pixels to match the rest of your system. The only solutions I can think of would be one of the Sony 694/814 chipped cameras used in 2x2 bin to get large pixels (but only 1.5/2mp) - or maybe even a ZWO/QHY 16mp CMOS camera with software binning for 4mp. Neither would be much good for wider fields, but good for smaller objects such as PNs. If you want to try this approach, let me know - you can experiment with my 694 if that would help. With the 694 in bin2, you would get back the missing 6x factor using you existing scope.
Quote:
Originally Posted by RickS
[Ray beat me to it with a more comprehensive answer, but we said similar things ]
Hi Paul,
Your system is approx one third the speed of Mike's based on the small difference in image scale, a little more obstruction and the larger difference in QE (for Ha the QE of the KAI-11002m is down around 30%.)
I'd agree with Colin that with 20 min NB subs you're going to be taking a big hit from read noise, which is up around 12e- for that sensor.
Longer subs would help, but you're at a big disadvantage based on QE and read noise. It is definitely the sensor that is the weak link in the chain.
Cheers,
Rick.
Thanks guys for the responses. I agree the read noise is the biggest problem here, with QE coming in a close second, then the scope itself.
I have been evaluating whether I will continue down this path for much longer and have considered other systems for quite some time. Imaging faint obscure objects is currently a criteria for this analysis. I am involved in some research imaging and really need a faster system for that purpose.
Agree that using a 20 minute sub was probably a mistake in the equation this time, though I thought the integration would have made up the difference and it has slowly over time but not enough to really be worthwhile, and there are diminishing returns here. Perhaps I should have used say 30- 60 minute subs (would welcome alternative comments here).
if you can get up to 60 minute subs, you will make a signifcant difference Paul - the read noise component will drop to about 0.6 of the level at 20 minutes.
from a sensitivity perspective, your scope is functionally equivalent to Mike's, with the same aperture and roughly the same sampling.
Depending on your filter passband (3,5,7nm) you'd be looking at somewhere between 1-5 hour subs to be read noise limited if you have 12e- RMS read noise; also depends on the exact numbers I use in the iPhone calculator
I'd say that it is the camera that is the biggest link in the chain. Focal ratio doesn't mean anything as long as you have pixel sizes to match. Mike is using 4.54 micron pixels at F/3.8, you'd have the same pixel scale with 9.55 micron pixels at F/8 (~0.8"/pixel).
Of anything however, a lower read noise than the KAF-11002 would make the biggest difference.
if you can get up to 60 minute subs, you will make a signifcant difference Paul - the read noise component will drop to about 0.6 of the level at 20 minutes.
from a sensitivity perspective, your scope is functionally equivalent to Mike's, with the same aperture and roughly the same sampling.
Ah ok, 60 minutes ought not be a problem but I would need to do this on nights when I can be assured of no cloud. Worth a try though to see what results could be obtained.
Quote:
Originally Posted by Atmos
Depending on your filter passband (3,5,7nm) you'd be looking at somewhere between 1-5 hour subs to be read noise limited if you have 12e- RMS read noise; also depends on the exact numbers I use in the iPhone calculator
I'd say that it is the camera that is the biggest link in the chain. Focal ratio doesn't mean anything as long as you have pixel sizes to match. Mike is using 4.54 micron pixels at F/3.8, you'd have the same pixel scale with 9.55 micron pixels at F/8 (~0.8"/pixel).
Of anything however, a lower read noise than the KAF-11002 would make the biggest difference.
Yes our image scale is very similar. And, I am contemplating a new camera as an new solution to the problem.
Band pass on the Ha is 5 nm , SII 5nm and 3nm on OIII, if that helps with your calculations. OIII on this object extends a little way around the object from the main ring. Deeper subs would be required to show it further out I suppose.
Quote:
Originally Posted by DaveNZ
Have you looked at focal reducers ? Im not sure what will work well with the GSO RC. So perhaps a new camera and a focal reducer.
Cheers Dave
Thanks Dave, actually this has been the line of development I have looked at the most, it is the most cost effective. Reducing the focal length down and using a new sensor to match the existing scale or thereabouts is going to work better for me at present and it will eliminate issues with my current camera, which is unreliable. I am making changes at the observatory to an existing pier this weekend to allow me to put the big scope in the roll off roof. This will give me some flexibility to house a large Newtonian or a large RC. If I use a reducer and newer sensor with smaller pixels at first that will help determine whether I need to go further and buy a fast Newtonian. A large Newtonian would not fit in the dome, but it will fit in the ROR.
Interesting workouts about comparable sensitivities of systems.
It makes sense as to why so many manufacturers are offering 12 inch plus compound scopes at F3.8 or faster. There are several.
694/300mm F3.8 is a powerful narrowband setup but limited in FOV.
The KAF16200 is probably the closest to ideal on the market at this time. Low read noise with some brands coming close to the 694 read noise 6micron pixels suit many systems (although not 3 metres unless you regularly get very good seeing).
Its hard to get one scope to do all things. My conclusion for what its worth is 300/F3.8 as in my Honders is great for widefield but not so great for galaxies etc. Long focal length is still best for galaxies. Small pixels/shorter focal length/fast F ratio versus larger pixels/longer focal length still seems to have the larger pixel longer focal length ahead for some reason although one could expect to get the same result. You don't from my experience but of course you can still get great results from small pixel/short focal length/fast F ratio just not the very best.
Why the 2 approaches don't get the same result is not entirely clear to me. I suspect its the drop in efficiency of the smaller pixels or cross talk between neighbouring pixels in the smaller pixels.
Also the 694 micro lenses can show up in 2x2 binning if you are not very careful with biases. They sometimes still show up. So the pattern noise of the Sony sensor which is not measured in the specs is much worse than with the dinosaur Kodak CCD sensors.
A new crop of backit CMOS sensors with low read noise appear to be coming to market soon and perhaps they have the potential to blow all these older CCDs out of the water. The ASI1600 results are showing that with care awesome results are attainable with CMOS now.
Daylight DSLR sensors have for some time had much lower read noise than the best Kodak CCDs. My Sony A7r2 sensor for example at higher ISO has read noise much lower than 1 electron. There are no CCDs that match that that I am aware of.
Tiny...screaming for more focal length....but still very nice central detail!be great If you can manage the outer shells
Agreed and thanks Louie. A longer focal length would be good, but maybe more to the point is a smaller image scale might be better for this object. If only I could get a 20" F6 at my disposal.
Quote:
Originally Posted by SimmoW
Wow a heap of work for that precious little gem Paul. Makes it even more valuable.
Ah sensor shopping, love it! Great time to be a consumer.
Thanks Simon. There has been some great improvements in sensor tech in recent years. I will considering for a while longer yet.
Quote:
Originally Posted by Bassnut
Excellent colour and detail Paul, could stand a fair bit of croping given it didnt suffer too much zoomed in a bit.
What about a NABG sensor?. Its rare I have blooming trouble with NB.
Thanks Fred. True it could be cropped a bit more, neglected to do this better.
As to NABG sensor, I doubt this would be helpful. The read noise on a 6303E will still be high and I will never go back to a SBIG camera again. I'll be looking at the newer sensors this time around.
Quote:
Originally Posted by gregbradley
A beautiful image.
Interesting workouts about comparable sensitivities of systems.
It makes sense as to why so many manufacturers are offering 12 inch plus compound scopes at F3.8 or faster. There are several.
694/300mm F3.8 is a powerful narrowband setup but limited in FOV.
The KAF16200 is probably the closest to ideal on the market at this time. Low read noise with some brands coming close to the 694 read noise 6micron pixels suit many systems (although not 3 metres unless you regularly get very good seeing).
Its hard to get one scope to do all things. My conclusion for what its worth is 300/F3.8 as in my Honders is great for widefield but not so great for galaxies etc. Long focal length is still best for galaxies. Small pixels/shorter focal length/fast F ratio versus larger pixels/longer focal length still seems to have the larger pixel longer focal length ahead for some reason although one could expect to get the same result. You don't from my experience but of course you can still get great results from small pixel/short focal length/fast F ratio just not the very best.
Why the 2 approaches don't get the same result is not entirely clear to me. I suspect its the drop in efficiency of the smaller pixels or cross talk between neighbouring pixels in the smaller pixels.
Also the 694 micro lenses can show up in 2x2 binning if you are not very careful with biases. They sometimes still show up. So the pattern noise of the Sony sensor which is not measured in the specs is much worse than with the dinosaur Kodak CCD sensors.
A new crop of backit CMOS sensors with low read noise appear to be coming to market soon and perhaps they have the potential to blow all these older CCDs out of the water. The ASI1600 results are showing that with care awesome results are attainable with CMOS now.
Daylight DSLR sensors have for some time had much lower read noise than the best Kodak CCDs. My Sony A7r2 sensor for example at higher ISO has read noise much lower than 1 electron. There are no CCDs that match that that I am aware of.
Greg.
Thanks Greg for the detailed response. I have been looking at the 16200 sensor with low read noise and using a reducer. The ASI1600 is another option but I am not keen on the image scale even with a reducer. I might wait a few months to see what happens with the new CMOS sensors coming on line. In the mean time if the STXL dies I'll change over to the QSI with a reducer and 2x2 binning.
Quote:
Originally Posted by Shiraz
if you can get up to 60 minute subs, you will make a signifcant difference Paul - the read noise component will drop to about 0.6 of the level at 20 minutes.
from a sensitivity perspective, your scope is functionally equivalent to Mike's, with the same aperture and roughly the same sampling.
Been doing some 60 minute subs. At 10 Subs it does looks similar to what I have achieved with 30 hours worth of data. I am pushing along to capture a bit more to see where I can end up. The attached image shows what it looks like.
Ultimately I doubt that I can get the same data as Mike's image.
Indeed a very beautiful and detailed image of a teensy tiny beast.
We achieved considerably less detail with the 20" F6.7. (Perhaps the thing we were missing was good seeing or a high mountain to perch it on). We could probably have a crack at the outer shells though.
Indeed a very beautiful and detailed image of a teensy tiny beast.
We achieved considerably less detail with the 20" F6.7. (Perhaps the thing we were missing was good seeing or a high mountain to perch it on). We could probably have a crack at the outer shells though.
I'd be interested in seeing how bright the outer shells could be achieved with your setup. I would expect an image resolution boost too, but seeing might not be as good on your site as on mine.
Seeing could be the main issue you had but I think also the use of an AO in good seeing does help.
If you do go ahead with a test I would like to see a 10 hour image just for comparison.
]
