CCD reality check

[ollypenrice]

Apologies for posting an image previously seen in a different context. (It was in an effort to show how fast mono and filters are.) However, as interest in the ASI 1600 understandably grows, I think that the 'slowness' of the old school CCD is becoming overstated. Our supposedly insensitive and out-dated Atik 11000 CCD needed one hour (2X30) to obtain this Ha result.

I wouldn't, in all honesty, call that slow...

Olly

[Marky1973]

I quite like being slow...I find rushing things just means I make mistakes more quickly.... :-)

[Filroden]

I think slow has many meanings.

Your example is 2x 30 minutes so each exposure could not be called quick! What I think it shows is the difference in QE. In the same given time your CCD is probably more sensitive and therefore detects more light than the CMOS sensor. In that regard the CCD is quicker to reach a certain level of exposure.

On the other hand, do the same test but run 120 x 30s and add some light pollution or even 3600 x 1s. The CMOS, with its lower read noise, will probably show fainter detail. In that (limited) regard, the CMOS could be considered quicker. It's quicker to exceed read noise which has its uses.

I don't think it's comparing apples with apples. The two chips work in different ways and are better suited to different techniques. I think it's important people understand the differences and choose the chip type best suited to their desired outcome with their equipment.

Give me your results any day!

[ollypenrice]

19 minutes ago, Filroden said:

I think slow has many meanings.

Your example is 2x 30 minutes so each exposure could not be called quick! What I think it shows is the difference in QE. In the same given time your CCD is probably more sensitive and therefore detects more light than the CMOS sensor. In that regard the CCD is quicker to reach a certain level of exposure.

On the other hand, do the same test but run 120 x 30s and add some light pollution or even 3600 x 1s. The CMOS, with its lower read noise, will probably show fainter detail. In that (limited) regard, the CMOS could be considered quicker. It's quicker to exceed read noise which has its uses.

I don't think it's comparing apples with apples. The two chips work in different ways and are better suited to different techniques. I think it's important people understand the differences and choose the chip type best suited to their desired outcome with their equipment.

Give me your results any day!

I think that what matters in the comparison is the total exposure time. (TBC. Time Between Clouds. )  You run CCD and CMOS differently so there is no point in running one as if it were the other. My result would have been very similar at 3x20 or 4x15.

I also think that if we compare two images of the Heart Nebula in Ha we are comparing apples with apples even if the apples came to market in different vehicles... Flirting with mixed metaphors here!

Olly

Edit: there is an advantage, in on-off weather, to a short sub camera. That's certainly true.

[Vox45]

I am by no mean an experienced imager but I am on the fence now regarding which way I should go in the future.

The price difference and the fact that I can get rid of guiding altogether is a strong argument for an ASI1600 and, to be honest , I doubt I will ever be able to attain the level of Olly's work as a part time imager so would I really be able to tell the subtle difference ? I think both methods are good (maybe an advantage for the CCD team) but, unless there is something inherently wrong with the new CMOS chips, wouldn't newcomer best be oriented to the new technology and methods ?

Like I said, I am on the fence now.

[Filroden]

9 minutes ago, ollypenrice said:

I also think that if we compare two images of the Heart Nebula in Ha we are comparing apples with apples even if the apples came to market in different vehicles... Flirting with mixed metaphors here!

You're right. It just becomes a straight shootout in terms of QE and I don't think CMOS is there yet. However, once you bring in environmental factors I'd rather apples grown in Somerset than the Sahara  If the metaphor is squeezed much more we will have cider. 

[ollypenrice]

2 minutes ago, Filroden said:

You're right. It just becomes a straight shootout in terms of QE and I don't think CMOS is there yet. However, once you bring in environmental factors I'd rather apples grown in Somerset than the Sahara  If the metaphor is squeezed much more we will have cider. 

 Or, worse, Calvados, and then we'll never be in focus again!

6 minutes ago, Vox45 said:

I am by no mean an experienced imager but I am on the fence now regarding which way I should go in the future.

The price difference and the fact that I can get rid of guiding altogether is a strong argument for an ASI1600 and, to be honest , I doubt I will ever be able to attain the level of Olly's work as a part time imager so would I really be able to tell the subtle difference ? I think both methods are good (maybe an advantage for the CCD team) but, unless there is something inherently wrong with the new CMOS chips, wouldn't newcomer best be oriented to the new technology and methods ?

Like I said, I am on the fence now.

The ASI is most certainly tempting on price. No doubt about it.

Olly

[LightBucket]

Surely that's all well and good if you have a mount that can track perfectly for 30 mins at a time, but to get that same or similar result above in say 3 min exposures, how many would it need...? my logic is with shorter subs you can get them between patches of cloud for instance, whereas one 30 min sub would require an excellent mount and perfect skies, no planes or satellites to ruin the longer sub, which can be pretty hard here in the U.K....so I think there is an argument for both CCD and the newer CMOS cameras, horses for courses as they say....

...just my twopeneth..

But the image is above is incredible Olly, there is no denying that...

William

[ollypenrice]

20 minutes ago, LightBucket said:

Surely that's all well and good if you have a mount that can track perfectly for 30 mins at a time, but to get that same or similar result above in say 3 min exposures, how many would it need...? my logic is with shorter subs you can get them between patches of cloud for instance, whereas one 30 min sub would require an excellent mount and perfect skies, no planes or satellites to ruin the longer sub, which can be pretty hard here in the U.K....so I think there is an argument for both CCD and the newer CMOS cameras, horses for courses as they say....

i also have seen a few thread like this, and they are nearly always started by a hardened expert CCD imager....Hmmmm very strange....getting pulled over to the dark side me thinks.....just my twopeneth..

But the image is above is incredible Olly, there is no denying that...

William

If you can track for 8 minutes (one turn of the average wheel) you should be able to track for 30. You'd need good PA of course. It doesn't have to be thirty, either. 4x15 would be very similar.

Aircraft would a spoiler, I agree.

And, yes, this isn't an 'anti-CMOS' proposition from me, just a 'keep things in proportion' one.

Olly

[Marky1973]

I am a bit confused.....so with the CMOS sensor, can you get the same quality of image with a similar total exposure length, just by using shorter subs, because the CMOS is, perhaps more sensitive/has lower read noise? This does make it an attractive prospect. I love my Atik but my last session was marred with losing every 3rd 600s sub due to cloud, at least with shorter subs, the damage would be lessened...if I am reading this right?

[LightBucket]

5 minutes ago, Marky1973 said:

I am a bit confused.....so with the CMOS sensor, can you get the same quality of image with a similar total exposure length, just by using shorter subs, because the CMOS is, perhaps more sensitive/has lower read noise? This does make it an attractive prospect. I love my Atik but my last session was marred with losing every 3rd 600s sub due to cloud, at least with shorter subs, the damage would be lessened...if I am reading this right?

Yes, that's how I see it exactly....!

[LightBucket]

10 minutes ago, ollypenrice said:

If you can track for 8 minutes (one turn of the average wheel) you should be able to track for 30. You'd need good PA of course. It doesn't have to be thirty, either. 4x15 would be very similar.

Aircraft would a spoiler, I agree.

And, yes, this isn't an 'anti-CMOS' proposition from me, just a 'keep things in proportion' one.

Olly

Sorry took the last bit of my post out, as thought it was a tad unfair....

[michaelmorris]

I looked into this a few months ago and a new CMOS-chipped camera certainly looked an attractive option.  My understanding (i.e. incorrect?) is that the CMOS chip gives you lower read noise, allowing for shorter exposures.  This would provide advantages during intermittent cloud, poor guiding, guideless imaging, and shooting at long focal lengths (due to the lesser need for accurate guiding).

On the minus side, the camera is far noisier than the Sony-chipped CCD cameras, leading to a need to take darks.  Cooled CMOS-chipped cameras are also a newer technology without the long track record of CCD astro cameras.  To be honest I have not (as yet) had that many problems with autoguiding.  However, I am planning to start shooting through a C9.25 with a f6.3 focal reducer and an off axis guider soon, so I may well end up eating my words!

In the end, I decided to plumb for the tried and tested and reliable Atik 460EX mono CCD camera.  However, a day may come in the not too distant future when CMOS technology could well become sufficiently advanced to overcome the noise issues and then I can foresee CCD cameras being usurped.

[LightBucket]

1 hour ago, michaelmorris said:

I looked into this a few months ago and a new CMOS-chipped camera certainly looked an attractive option.  My understanding (i.e. incorrect?) is that the CMOS chip gives you lower read noise, allowing for shorter exposures.  This would provide advantages during intermittent cloud, poor guiding, guideless imaging, and shooting at long focal lengths (due to the lesser need for accurate guiding).

On the minus side, the camera is far noisier than the Sony-chipped CCD cameras, leading to a need to take darks.  Cooled CMOS-chipped cameras are also a newer technology without the long track record of CCD astro cameras.  To be honest I have not (as yet) had that many problems with autoguiding.  However, I am planning to start shooting through a C9.25 with a f6.3 focal reducer and an off axis guider soon, so I may well end up eating my words!

In the end, I decided to plumb for the tried and tested and reliable Atik 460EX mono CCD camera.  However, a day may come in the not too distant future when CMOS technology could well become sufficiently advanced to overcome the noise issues and then I can foresee CCD cameras being usurped.

Can you explain something to me as I am finding it hard to get my head around....

when you say the camera has very low read noise, and then say that the same camera is much noisier....what does that mean, as it seem like a complete contradiction, obviously they are not the same but I do no understand.... 

thanks in advance

William

[michaelmorris]

There are various sources of noise introduced by the process of creating a digital image.  My understanding is that 'Read Noise' is extraneous signal introduced into the data by the process of reading off the values from each pixel at the end of an exposure.

The electronics in the chip and the associated electronics in the camera can also create a thermal signal (i.e infra-red photons), a.k.a thermal noise, which penetrate the back and sides of the imaging detector.  Cooling the camera will help reduce this thermal signal.  Some cameras and types of chip seem to handle such thermal noise much better than others.  Also some cameras have a greater variance in the sensitivity of individual pixels to signal.  This also creates a 'noisy' image.  Some of these imperfections are also temperature-dependent (I think).

It is possible to get rid of the most of this noise in the calibration process by taking dark frames.   However, personally, I just prefer to start off with a cleaner data to start off with -  it makes life that bit simpler.

I hope this helps.

[wxsatuser]

I had the choice of both but did'nt fancy taking loads of short subs, so got an Atik 383L.

10minute subs are easy with an HEQ5 pro and stand alone guiding, so I'm a happy bunny.

If those would just buzz off would be even happier.

[michaelmorris]

9 hours ago, michaelmorris said:

I looked into this a few months ago and a new CMOS-chipped camera certainly looked an attractive option.  My understanding (i.e. incorrect?) is that the CMOS chip gives you lower read noise, allowing for shorter exposures.  This would provide advantages during intermittent cloud, poor guiding, guideless imaging, and shooting at long focal lengths (due to the lesser need for accurate guiding).

On the minus side, the camera is far noisier than the Sony-chipped CCD cameras, leading to a need to take darks.  Cooled CMOS-chipped cameras are also a newer technology without the long track record of CCD astro cameras.  To be honest I have not (as yet) had that many problems with autoguiding.  However, I am planning to start shooting through a C9.25 with a f6.3 focal reducer and an off axis guider soon, so I may well end up eating my words!

In the end, I decided to plumb for the tried and tested and reliable Atik 460EX mono CCD camera.  However, a day may come in the not too distant future when CMOS technology could well become sufficiently advanced to overcome the noise issues and then I can foresee CCD cameras being usurped.

Sorry I forgot to add a major potential draw back of the 'multiple short exposure with a CMOS chip' imaging strategy - computer speed and capacity.  

120 x 1 minute subs are going to take up a shed load of hard disk space when compared to 12 x 10 minute subs.

12 subs with the Atik 460EX = 141Mb

180 subs with an ASI 1600MM = 1.2Gb (Admittedly, the ASI camera has a bit larger chip, but the difference is in storage needs is still vast.)

For an RGB + luminance image this will mean the FITS files for the CMOS camera would take take up a whopping 4.8GB.  Add in bias frames, flat and darks and we are talking serious amounts of storage and processing requirements and vast storage requirements if you want to back everything up.

Using a CMOS-based system would have meant a very costly upgrade from my £39, dual core Windows 7 Pro 64 bit with 2GB RAM and a 200 GB hard disk desktop I use in the observatory.

[wxsatuser]

Storage can be upgraded quite cheaply these days.

An external 2tb hdd should come in at £70 or less, I'm sure a UK user will find it hard to fill 2tb, with our clouds.

Failing that an Office 365 Home sub of £80, latest deal at the moment, will get 5tb of Cloud , 1tb for 5 users each plus they all get Office installed.

I use Onedrive to transfer my files from the obsy laptop to both my main processing laptop and it's backup processing laptop.

[xanthic]

4 hours ago, wxsatuser said:

Storage can be upgraded quite cheaply these days.

An external 2tb hdd should come in at £70 or less, I'm sure a UK user will find it hard to fill 2tb, with our clouds.

Failing that an Office 365 Home sub of £80, latest deal at the moment, will get 5tb of Cloud , 1tb for 5 users each plus they all get Office installed.

I use Onedrive to transfer my files from the obsy laptop to both my main processing laptop and it's backup processing laptop.

That's ok if you have a permanent setup. I'm running off battery, so another power hungry device would be unwelcome and I wouldn't 100% trust wifi to get my files synced with the cloud.

[vlaiv]

I think CCD still wins in this regard. Simply because of pixel size. CMOS sensors available as astro gear are order of 3~4 um. One can easily (well not taking cost into account) acquire 9, 12 and even larger pixel size in CCD astro camera.

This simply means that CCDs can be paired with larger aperture (and hence focal length) to obtain same resolution.

All of this can change really easy in favor or CMOS when large format CMOS is introduced to astro market. CMOS has the advantage of being software binnable with less read noise. Take for example ASI 1600 it has 3.8 um pixels with 1.6e read noise at unity gain. If you bin 2x2 you will get 7.6um pixel ( 14bit depth as a bonus) with 3.2e read noise - you will be hard pressed to find CCD with such features. But if you bin 3x3 or more you simply run out of chip surface - it can't compete with 35mm 9um CCD (or larger ones).

On matter of thermal noise, I guess there are quite a bit of difference from sensor to sensor in CMOS. On my ASI178 (back side illuminated) I found that dark noise is non issue when running at -20C for exposures up to 5mins (did no go longer than that with my setup). What I did find however is that Flats are paramount in getting good image. Not just because of vignetting and dust and all the usual reasons but because there is also variation from pixel to pixel in integrated amp circuits / qe - sensitivity.

[wxsatuser]

41 minutes ago, xanthic said:

That's ok if you have a permanent setup. I'm running off battery, so another power hungry device would be unwelcome and I wouldn't 100% trust wifi to get my files synced with the cloud.

I can understand that about wifi and the cloud but I have had no problem so far.
As long as the wifi link is fairly solid it does'nt take long to synch several hundred Mbs.

My obsy is at the bottom of the garden, it just gets the indoor router, BT hub6, but to make it solid
I have a BT extender in an outside power socket that makes a solid signal in the obsy.

[tomato]

I am hoping the CMOS route continues to gather momentum, then the classifieds should be filled with 'old tech' CCD cameras that I can take my pick of.

 

[Filroden]

25 minutes ago, tomato said:

I am hoping the CMOS route continues to gather momentum, then the classifieds should be filled with 'old tech' CCD cameras that I can take my pick of.

 

Nice thought but it almost certainly won't pan out that way. CMOS is a more likely first or second step into imaging after DSLR. And if CMOS technology continues to advance (and it will because of the large amounts of R&D being invested into CMOS compared to CCD) then it's likely that anyone taking the step from CCD to CMOS is doing it because it offers them an improvement. And then, why would you buy the more expensive, used, inferior product?  

Pure speculation as CCD still rules the roost for astronomy but once CMOS QE and well depths improve...

[frugal]

I have been following the development of the new CMOS cameras with interest. Mainly because either CMOS or CCD would be better than the DSLR noise generator I am currently using. I tend to take images at 5min or 10min at the moment, because that is where I am sky-glow limited without moving to narrowband, and either CCD or CMOS seem to play nicely in this space. I am not sure that my guiding is good enough for 30 minute images (even if the clouds permitted it). In a 4 hour imaging session, an error on a 30min sub will be a loss of 12% of my imaging time, but a loss of a 5min image will only be 2% of my session, but that is a different argument for a different day  

The "too many subs" argument seems a bit of a fallacy to me. If you are looking at a camera that is £300-700 cheaper than "equivalent" Atik CCDs (and a lot cheaper than some of the other CCD brands), you can buy a nice desktop for £700 that will stack all of your images quickly. And to be honest, stacking even 50 22Mb DSLR RAW files only takes up about 5% of the image processing time, the rest of the time is spent carefully teasing out all of the information.

To be honest my main interest in following the new cooled CMOS cameras is that the development of CCDs seems to have stagnated. It is hard to find any information on when specific CCD sensors were developed, but having a look at the Kodak KAF8300, the technical documents at V5 are dated 2010, and the earliest camera I can find that uses the chip is the Olympus E-300 which was released in 2004. Is a chip that is between 7 and 13 years old still considered to be industry leading? Finding information about the release dates of the Sony sensors is even harder.

[xanthic]

I can honestly say I've not yet seen a image from one of these CMOS cameras that holds up against a CCD image. Zoom in and the noise really starts to show, even on the longer integration images.  I'm also curious about the lack of the need for guiding. Even with good polar alignment, surely periodic error creeps in? Fine if you've got a permanent setup and can PEC train but not so much use for those of us on tripods? Happy to be corrected if I've got the wrong end of the stick.