Mosaicing vs a gegacam

[Bassnut]

I’ve come to considering a wider FOV via a bigger cam, or mosaics with what I have (ST10).Mosaics are a severe time consuming exercise, at least I thought so until I looked into the options.
Assuming in my case, that the most exposure time is with Ha (both in narrowband and Ha as Lum in LRGB), it was most interesting to see how a larger chip would influence exposure time for a given approximate image scale and FOV.
Here’s an analysis of 2 options available with a 10" scope, using a 0.67f FR to try and keep the images scales roughly equivalent.
On a 10" scope
Chip QE at 700nm, pixel, image scale, FOV arc min
KAF 3200 (ST10) 80% ,7um, F6 0.93, 23 * 34
KAF 11002 (ST11k) 25%, 9um, f9 0.82, 37 * 55
KAF 16803 40% ,9um, f9 0.82, 56 * 56
To get roughly the same FOV as a KAF 11002 (ST11k) would require 4 panels from an ST10, a KAF16803 would require 6 panels. Doubling the f ratio requires 4 times the exposure for the same flux, so the FR in this case from f6 to f9 would double exposure time.
To get the same flux and FOV then from an ST10 mosaic with 1hr Ha requires these exposure times with multiple panels.
KAF 11002 (ST11k) 4 panels of 1hr =4hrs on an ST10 at f6 to equal a single panel of 6.2 hrs at f9 .
KAF 16803 6 panels of 1 hr = 6 hrs on an ST10 at f6 to equal a single panel of 4 hrs at f9.
The ST11k then requires more exposure time for the same result, and the 16803 less, but im not convinced the diff is worth the money. Certainly mosaicing is not the horror one would expect.

[bert]

What about the stl6303? Its qe is still quite high compared to the abg cams you have compared the st10 to.

[Bassnut]

well, thats true, the 6303 (54%QE at 700nm) at f6 offers a 42* 63 FOV , more than a 16803 at f9 (37*37) for a lot less money and comparable exposure times. Still pricey though. Mind you, if 16803s were free, id take it ;-).

[bert]

When you have the sensitivity of an st10.... its kinda hard choice where to go next..

[CometGuy]

Hi Fred,

Here are my numbers:

The qe values I get for H-Alpha (656nm) for those sensors is:

KAF-3200ME = 0.80
KAF-6303E = 0.65
KAI-11002 = 0.30
KAF-16803 = 0.45

Also because narrowband darkens the sky so much, sensor noise is also going become a factor and the values for those (I use values I have seen measured - not those spec'd by Kodak).

KAF-3200 = 11 e-
KAF-6303 = 15 e-
KAI-11002 = 16 e-
KAF-16803 = 7 e-

The areas of the sensors are:

KAF-3200 = 152 mm2 (equivalent to 267 mm2 when focal reducer used to match pixel scale to other chips)
KAF-6303 = 511 mm2
KAI-11002 = 878 mm2
KAF-16803 = 1380 mm2

Assumming minimal vignetting problems with the larger sensors then I would calculate equivalent exposures as follows for the same area of sky:

KAF-3200 = 4.6 Hours
KAF-6303 = 2.9 Hours
KAI-11002 = 3.7 Hours
KAF-16803 = 1 Hour

I'd still be looking at the 16803 chip but whether the scope is going to cover it well is something you would need to research. Anyway good luck with your decision.

Terry

[Terry B]

The ST10 has a chip that is 2184 pixels wide. As a 4x4 mosaic you will end up with an image that is 4368 pixels wide.
What screen will show this?
What size print do you realistically get printed?
I know the big chips seem great for wide fields but I wonder what you can do with the resulting frame.
To make a reasonable size image with the 4x4 ST10 mosaic you could bin the pixels 2x2 significantly increasing the sensitivity and still coming out with an image that is over 2000 pixels wide.

[AlexN]

Tough one Fred... Real tough..

the 16803 seems the go... but as you say.. is it worth it? its not going to be a cheap move, its not just the camera as you well know, 50~65mm square filters, a CFW, I don't know if your AO unit will support a 16803? I doubt it will though, as SBIG are releasing a new AO unit as a direct result of the STX-16803's larger chip...

I'd assume that the RCOS would illuminate the 16803 well enough to make it viable...

The 6303 as has been mentioned here (and in a previous convo we had a few days back) is good option, cheaper by a long shot too when you consider the STL6303E includes a CFW and your AO unit will still be of use...

As you mentioned in your first post, sensitivity vs mosaic time, the KAF16803 is up there. 4hrs at F/9 will take you 6 hours to do with the ST10 @ F/6.. And the question is, how often do you stop at 4hrs for your Ha? Judging by your gallery, its more like 10~20hrs.. Given that, you're looking at a LOT more time for the ST10 to cover the same amount of sky...


As for Terry's comments re the final image resolution... I really don't think it matters how many pixels the image is.. sure a 16mp camera will chew up HDD space with raw files etc... It all comes down to sensor size or your eagerness to do mosaics. In any case, you're looking at having 4000px wide images minimum, whether you mosaic with the ST10 or shoot single panels with the 16803, sure, your screen can't display the whole image at once without it being scaled to say, 30%.. And yeah, I doubt you're printing A1 or A0 images very often... but there is no other choice, unless you really feel like using an STL1001E, which has a rather large sensor with 24um pixels, heavily sacrificing your resolution, but your 1024x1024 px image will fit on the screen all at once :)

[Bassnut]

Brett

There are other more obscure options for sensitivity, but not FOV unforch, unless the price gets really silly......

Terry

Yes, Ha is 656nm, not 700nm, that was a dumb approximate.

I did the numbers again, cant see how you get yours, for example, you quote 1 hr for the 16803, do mean 1 panel on a ST10 at 1hr at f6 equals 1 hr for 1 panel on the 16803 at f9, how can that be when it has half the Qe and double the exposure time (at f9)= 4hrs ?.

EDIT. OK,on a re read, I see what youve done, a bit confusing, ignore the above rubbish, still dont see how you get your numbers though ;-).

Actually 6 panels on the ST10 is larger FOV and 4 panels is smaller, so its sort of inbetween 6 or 4 hrs on the ST10.

I worked out the ST11k would actually take quite a bit longer than 4 panels on an ST10 for the same result 1hr * 2.7 (30% QE) * 2 (f9, over f6)= 5.4hrs. Actually the 4 ST10 panels at f6 (45*67.6 a/s) are a larger FOV than the 1 ST11k at f9 (36*55) too .

Terry

Its extra FOV im after, not a humungous number of pixels, as Alex says.
Binning is quite a good idea, at f6 the image scale with bin 2 is 1.87 a nd at f9 1.25. Given you should have it about half of seeing, which in my case is usually 3 to 6, then 1.87 would scrape in and 1.25 would be fine, although deconvolute works better sub 1.0 in my experience.

Alex

I cant consider a 16803, it was just for comparison, nothing else I have would take it (apart from the rotator), OTAs, AO, external focuser (when I use it), I just couldnt aford it, and I dont think you can get a wheel that would take a full set of LRGB and NB (with 2 Ha), which would be a pain, I use them all.

No, niether the RC or LX would take it.

The STL6303 comes with an 8 pos FW option, would just do nicely, and yes it appears the A08 will just do too. Its been around a while, so I may be able to get one S/H.

The new STXs have a 6303 version ;-).

You know, ive been thinking of the importance of image scale for a while, and I think things have changed lately that make the old formula not so reliable. Some ppl think sub 0.6 arcsec/pix does make a difference even in 2 arc/sec seeing, yet, and not much is mentioned of this explicitly yet, the new huge cams like the 16803 seem to do very well at 3 arcsec/pix and up simply due to the huge pic size viewed full frame.

Look at Martin Pughs recent demo pic with a tiny refractor, massively undersampled, the stars are blocks, but you have to zoom in a long long way to tell. Viewed full frame, you wouldnt have a clue, and it looks fantastic !.

So, as Terry says, binned ST10 mosaics at a huge final image size changes things significantly (taking into account the above,and anyway not that much undersampled really), that would halve (or more) the exposure time, and start making mosaics very attractive indeed ;-).

[gregbradley]

You can get a 7 position filter wheel for a 16803 from FLI and Apogee. The Apogee one in particular is state of the art.

I take it you were looking solely at Ha performance as QE in the 16803 peaks at about 62% with no cover slip.

Also no microlens diffractions spikes.

I don't think the comparison is quite right though Fred. Noise of the electronics and chip, ability to cool to reduce noise even further come into it as well.

16803 chip with FLI or Apogee electronics are very clean. Especially FLI with its best in market cooling performance.

My PL16803 is clean as a wolfs tooth at -35C a temp it can reach very often throughout the year (perhaps not on a warm summers night but -30C is probably realistic all year round.

I imagine the 16803 is more efficient after all is done in your comparison with a 3200 although its QE is extremely high and galaxy images abound that show that.

A downside is harder to process in that the image files are 32.4mb each and you need a fast computer and lots of RAM.

It seems to me a 16803 on a long focal length scope will get you awesome image scale with incredible resolution more than any other camera.

I noticed every time I put a bigger chip on my RCOS 12.5 inch the images improved substantially.

Does the 6303 suffer from star blooming? Its not anti-blooming and Lars told me he had a lot of trouble with blooms with a 6303.

Greg.

[Bassnut]

Im not suggesting mosiacs on a 3200 or 6303 are actually prefered over a 16803 for large FOV, just that they dont seem to be as painfull as I thought.

OK, well, a 7 pos wheel is a handy size, would take the full LRGB Ha SII OIII. Since 80% or more imaging time is at Ha, thats where the Qe counts (for me anyway), unless you image more Lum for galaxies etc, but then exposures are much shorter anyway.

The value in going below -20deg has been argued before, lower is better of course, why not if you can, but Im still not sure its THAT important.

The 12.5" RC will take the 16803, but not the 10" ive been told, so along with changing everything else in the optical path, thats a serious amount of money, hence a long list of excuses to avoid it ;-).

Yes, NABG can be difficult, but I havent found it so much of a bother actually.

That file size is frightning, certainly my PC couldnt process them, so thats another PC, although thats not so expensive now adays.

Anyway, ill give a mosiac a try and see how it goes.

[rally]

Fred,

I may have misinterpreted the information.
But in your initial post you listed the KAF1102 FOV as 37 x 55 but then the 16803 as only 37 x 37 - should that not be 55 x 55 ?
Assuming the same image scale.

If so - you have more than halved its area !!!

That might make the difference you need to justify its purchase :)

Rally

[Bassnut]

Rally. Sorry, I just typed in the wrong size in the post, the calculations are with the correct size.

[AlexN]

Give the mosaics a go Fred... Cant hurt and if it all goes well, it should save you a lot of money...

Given that you're mounted permanently on a PME, I would assume your repeatability in framing a target night to night would be pretty spot on.. and if you wanted to move the target "vertically" say 90% of the field of view, I would assume the PME can make this adjustment without a worry at all? Giving yourself a 10% frame overlap for stitching...

Am I overestimating the PME or is it everything I expect it to be?

[multiweb]

The larger the CCD chip the more demanding on the imaging circle and correction. THen you need all the "big" filters to go with it too. I agree that it's less work though. You capture everything in one go but is the extra degrees/minutes in your field worth the dow compared to mosaicing? Just offset shoot four frames and you're done. There are plenty of software solutions now to stitch/align.

[Doomsayer]

Hi Fred
As you know, I ended up with Lars' STL6303. I would recommend it highly. It is NABG, but so far it has not been that much of a problem and is an ideal match for my 12.5" RC which has a field flattener and a native FR of 6.7. I also intend to get a FW8 for it. There have been several STL6303s going on Amart in the last few months. I would still prefer an FLI or an STX for its cooling however. A 16803 is on my longer term shopping list, but I think I'd end up putting the bigger chip behind a widefield scope like an FSQ. Weather is not allowing much use of the 6303 so far.

guy

[Bassnut]

Maxim DL and the sky both have Mosiacing features (I havent tried them), and of course automation software will do it, with plate solves.

But its no extra work or software really, even though the PME repeatability is good, I always tweak position at the start of a night (its very easy, I plate solve an exposure to calibrate the nudge feature in DL, and just click "point here", its always right 1st go, whilst looking at the previous nights pic with crosshairs).

I almost never change objects in a night, and individual panels would all take more than one night anyway (with multiple filters), so theres no difference in setup time or extra work at all for mosaics apart from more total exposure time. I would just tweak the position for the required panel once each night and go, exactly the same as I do now.

Yes, although I havent tried stitching software yet, I understand some of it is amazing and just does align/scale/rotate automatically in one go.

[Bassnut]

Thanks Guy. And although id like an STX too (with all the extra features), thats expensive, although a tempting option, and I expect S/H ST6303s will drop even more in price when the STXs come out.

Fitting everything on the scope in different combinations including FR,AO and OAG is proving difficult, both mechanically and with BFD, so the new STX guiding is looking very attractive ;-).

[CometGuy]

Hi Fred,

Have you seen Richard Crisp's articles?

http://www.narrowbandimaging.com/home_page.htm

in particular this article regarding the 16803 chip (I'm sure he has tested the 6303 as well):

http://www.narrowbandimaging.com/images/proline16803_report_crisp.pdf

The graph on page 15 tells the story. Heavy going but worth understanding. Basically, you want to go past -25C for narrowband imaging if you can. As an example have a look at the pink curved line, which is a plot of noise versus temperature for a 10 minute exposure. At -25C it hasn't dipped below the bold dashed horizontal line (the 100% line) which means that the exposure is dark noise limited. It's not until you cool to -33C that the line drops below the bold dashed line and is no longer limited by dark noise. If you were planning for 10 minute subs I would be aiming for -35C with this chip.

In all cases he has assumed the sky background brightness is 0, problably a valid assumption at f9 and observing from a dark sky with narrowband filters.

Terry

[AlexN]

Fred - You may not have seen it, but there is a top of the line big chip camera setup in the for sale section of the forums.... Thought I'd throw that in here..

[Mighty_oz]

Oh that's cruel just cruel. Shame on u Alex :) Mighty tempting tho, but at least i can resist as i don't have a tely fit for that gear.

[AlexN]

How do you think it is for me :) I just bought a mono ccd, I have a scope that could make use of that camera, but alas, I need to curb the spending for a little while.. Had it come for sale 5 months ago.. :)

[Mighty_oz]

Well i guess i could help out 3 ! people or is that 4 including myself, if i suggested i could take that slightly used SBig 8300 off of your hands :) hehe
That was Fred has no worries trying to work out finances .

[Bassnut]

Terry, I havent seen that, thats very interesting. I notice the total noise for a 9e read noise cam (ST10) doesnt change much from -5 to -25, you have to go way past that to make a big diff (pg 25, with 20min exposures, common for me), and the 6303 is 13.5e !. He says somewhere very deep cooling doesnt help hi read noise cams as much as the 16803, as the read noise dominates.

I can see now why I get a fair bit of noise with 20min exposures and very low signal ADUs with the ST10 regardless of temp below -10, and the value of deep cooling with a low read noise cam, it all makes sense now ;-).

[Bassnut]

Yes, thats a nice cam kit there for sale, but for me (just me), the Qe is a concern. For anything other than dim, long FL very NB megdata though, its fine.

[gregbradley]

Hi Fred,

I'd go the 6303 for your setup. Kodak hasn't been able to better that chips performance in QE since. They have improved in other areas but the 6303 is still a really great chip with lots of great images on the net.

It also seems to be the narrowband imagers camera of choice.

STL6303's will no doubt be cheaper as the existing owners upgrade to an STX.

I think a 6303 also gives a nice image scale for a 10 inch RC. You should be able to get plenty of detail in galaxies etc.

Greg.