Which Colour Camera?

[PRejto (Peter)]

I'm imaging with two scopes on one mount. A TEC180 for luminance and a TEC140 for color. Automating is proving very vexing and I'm ready to try simplifying. Eliminating a FW and lots of re-focusing will certainly help. Thus, the idea of replacing my mono camera (Trius) on the TEC140 is gaining some traction.

Currently my setup yields the following:

TEC180/KAF8300 CCD: .88 arc-sec resolution FOV: 37 x 49.4

TEC140/Trius .95 arc-sec FOV: 35 x 43.7

The resolutions and FOV are similar so combining goes well. I do, however, not especially like having a smaller FOV for colour.

Here is what I'm considering, though open to suggestions (please):

ICX815 (ATIK490 or Trius 814c) .78 arc-sec FOV: 35 x 43.7

ICX613 (QHY12) 1.08 arc-sec FOV: 56.7 x 83.2

The QHY appeals with larger pix and well depth (5.12 microns, 32,000 well) vs the ICX815 (3.69 microns, 15,000 well). But the ICX613 is noisier at 8-10e- vs the ICX613 5 e-.

Then there is the QE to consider where the ICX815 seems lots better. The one factor possibly in the ICX613's favor might be the bit of blue issue and the TEC140. My Trius sometimes reveals a problem where the ICX613 might act a bit more like a KAF type chip.

Anyway, I'm not ready to spring on either camera and would appreciate any that I might have missed. I have looked at the QHY10 as a possibility (1.27 arc-sec resolution and a large FOC: 57.2 x 83.6). I Think I would prefer to have the higher resolution of the QHY12.

Any feedback much appreciated!

Peter

EDIT: I just heard about the SBIG STF8050SC. It would give me 1.16 arc-sec resolution and FOV 48.1 x 64. I can't find out very much about this camera but seems to perform a lot like a KAF8300.

[gregbradley]

I was going to suggest the Truesense Sparse 8050. I think there is another one similar the 16200. QHY has a 16200 6 micron 60% QE built in filter wheel and OAG coming out in a short while. I am sure there will be a OSC version.

The KAI6050 is an APS sized 5.5 micron sensor. Although I am not sure about the performance of the 5.5 micron Kodak chips. The same pixel size is in the KAI29050. If you look at the Cedic images from Chile the ones using an AP175 and Proline 29050 are the weakest images whilst the KAI16070 (full frame 16mp) and the Proline 16803 are the best.

The Starlight Express MX26C is one to look at and the venerable SX 25C has a long history of good images.

http://www.sxccd.com/trius-sx26c

Or the QHY version of the same sensor:

http://qhyccd.com/QHY10.html

The Trius has the 3 usb ports and you are already setup with all the drivers etc. The 3 usb ports are handy. I don't know about QHY versus Starlight Express for quality and reliability. But QHY is probably cheaper.

Greg.

[LewisM]

I am an OSC junkie...

I have used the Starlight Xpress SXVR-M25C and the SBIG ST-8300C. Hands down, the SX wins, BUT, the pixels are pretty large.

The SBIG works very well, but the SX's Sony chip eats the Kodak for breakfast.

[Slawomir (Suavi)]

Hi Peter,

Since you have been considering Atik490, what about stepping into unknown territory with Atik1120ex?

[rally]

Peter,

Not sure if the KAI-08050 is a good candidate for Astro

Specs here
http://www.ccd.com/pdf/ccd_8050.pdf

Its only got 20,000e well depth and the spec sheet says 12e- Read Noise although the Camera maker is quoting 9e- Read Noise
Maybe they are using a lower transfer speed to get the lower noise ?

But 20ke and 12e only gives you 10.7 bits - so not a big dynamic range and even less than the KAF-8300.

The TrueSense Filter Pattern - an additional option to Bayer and Mono would make it more sensitive in Qe than a Bayer one shot - but at a cost of colour space.
Not sure if all the astro processing platforms handle this Bayer alternative pattern or not - would be worth checking.

The KAI-16050 is the same in the dynamic range department - only 20ke- well depth.
http://www.ccd.com/pdf/ccd_16050.pdf

Compare that to
The KAI-11000 series have 50ke- well depth
The KAF-16803 well depth is 100ke-

I think the intended market is industrial, traffic and scientific.

Originally Posted by PRejto

EDIT: I just heard about the SBIG STF8050SC. It would give me 1.16 arc-sec resolution and FOV 48.1 x 64. I can't find out very much about this camera but seems to perform a lot like a KAF8300.

[PRejto (Peter)]

Guys,

I really appreciate your suggestions! I'm going to over San Jose for AIC next week and I plan to ask a lot of people about this subject.

Slawomir, I have certainly looked at the ATIK 4120. It would give a fairly over sampled image and a FOV actually smaller than my Trius currently. For those reasons alone I think it's not the "answer." (.65 arcsec, 30.7 x 46.1)

Lewis, when you say the the Sony eats the Kodak chop (er I mean chip!) I assume you mean in terms of QE, darks, the usual?? The Kodak does win in terms of the FOV (except when considering the QHY12 which nobody commented on). I did receive a not too favorable comment about the QHY12 in terms of build quality. Too bad because on paper it seems to fit the bill pretty well, though without the high QE of the newer Sony CCDs.

Greg, I think Rally points out some important things to consider re the KAI8050. Probably for my purposes the SC version wouldn't be of much use given I only want the colour data to combine with high quality luminance from my TEC180. The SX25 & 26 are interesting cameras. The SX would give me a resolution of 1.27 arcsec and FOV 55.4 x 82.7. The FOV is great but I'm not sure the resolution is what I need given the TEC180 is at .88 arcsec. The SX25 is even worse in that respect giving 1.64 arcsec.

Rally, You have given me lots of good info to ponder. I really appreciate it and I'm thinking more and more that the KAI8050 isn't the right chip for my needs.

It seems a given that if I move from a mono camera (for colour) I will be giving up resolution somewhat of a similar nature to binning a mono camera 2x2 but without getting any gain in sensitivity. So, compared to a mono camera R & B would be 25% of the pix active, Green 50% of an array of 4 pix. Now, since it "was" common practice to bin colour data when doing LRGB it would seem prudent to not make matters worse by using a camera that under samples compared to my L data. Is that correct? If so, I think I ought to stay in a range fairly close to .88 arc-sec. I don't expect to save any time getting colour data, but I do think it all might prove easier. Maybe I'll have a better feel for all this after my trip.

Thanks all,

Peter

[LewisM]

Yes to what you replied re Sony vs KAF. With the SXVR-M25C, darks are NOT required (in fact, makes the image WORSE). QE is high. And I feel all round the SX's are MUCH better made than SBIG.

I had some issues with the SX in terms of header sizes not reading right, but their software guy sorted that within a day!!! Like most SX's, it doesn't get particularly cold, but with the very clean Sony chip it is not much of an issue. The newer and rebuilt M25C's now have double stage cooling, so go colder. SX did that for me for $130 by memory.

I want to go back to the 25C... just funds are tight right now. I do like the ST-8300C, but I feel the 25C is a better option for me in the end, despite the largish pixels.,

[Bassnut (Fred)]

Originally Posted by LewisM

And I feel all round the SX's are MUCH better made than SBIG.

Thats interesting, whys that?.

[PRejto (Peter)]

I'm beginning to think that if I go down this path I ought to stick with a Starlight Xpress camera if only to be able to use the same fittings I use with my current Trius. That way I can swap out cameras when I want to do NB imaging with minimal fuss. That leads to two questions.

1. Would it be incorrect to use a luminance filter (already in place if I leave the SX filter wheel) and just swap over to a one shot colour camera? On a reflector I imagine one might remove the L filter, but on a refractor?

2. Since I would be giving up resolution going to a one shot camera (currently .95 in the mono camera Trius) would it better to try to gain back some resolution going with the Trius 8214C (.78 arc-sec), or stick with the 694 chip at .95 and larger pix (4.5 microns vs 3.69 microns)?

Thanks,

Peter

[Slawomir (Suavi)]

Hi Peter,

To my knowledge, in OSC camera you already have filters permanently attached to each pixel, so adding an additional filter in the optical path will lead to extended exposures and may affect the final colour of your image. But I cannot see any problems with that really.

I think the theory suggests to use smaller pixels with OSC if you want to maintain resolution. De-bayering involves estimating missing values based on surrounding pixels, thus with the same pixel size mono camera should normally yield higher resolution (detail) than OSC.

[LewisM]

Originally Posted by Bassnut

Thats interesting, whys that?.

I have the ST-8300C. The build quality is fine, it just seems they made a few too many shortcuts to save some $. Fibre dessicant that the user must fit, in rather a delicate spot and with no real method, and the thing bricks if you turn it off the wrong way (no switch on the power supply). I guess they improved it all with the more expensive STF and STT etc.

I have had a few SBIG before, and they were hassle free, but the SX to me is just a better package - smaller, lighter, comes as standard with a tip-tilt faceplate to get orthogonality, simple power and USB setup, and actually pretty decent software with it (though I use MaxIM, so it doesn't matter much).

Now having said that, I'd drop the lot to get another FLI - just a masterpiece of perfection! The ML8300C I had blew my socks off - should NEVER have sold that one!

[Bassnut (Fred)]

Thanks Lewis, your experience is good to keep in mind, ive only owned one brand.

[PRejto (Peter)]

Originally Posted by Slawomir

Hi Peter,

To my knowledge, in OSC camera you already have filters permanently attached to each pixel, so adding an additional filter in the optical path will lead to extended exposures and may affect the final colour of your image. But I cannot see any problems with that really.

I think the theory suggests to use smaller pixels with OSC if you want to maintain resolution. De-bayering involves estimating missing values based on surrounding pixels, thus with the same pixel size mono camera should normally yield higher resolution (detail) than OSC.

Thanks Slawomir. I appreciate the feedback. I've acually been using a Hutech IDAS for a luminance filter, and I used to use it in front of my RGB to pretty good effect given the amount of LP I face. I suppose it would work about the same. A straight luminance filter probably wouldn't greatly increase exposure (I hope). I guess time will tell. Big pain to need to change out the wheel or filter with a camera change which otherwise would be rather easy,

I understand the logic of using smaller pix but I wonder if given the usual seeing there would be any real benefit. The price for a possible benefit would be lower sensitivity and even smaller wells. Worth it?

Peter

[PRejto (Peter)]

I wrote to Terry Platt at SX. His reply was that in his view the resolution of a colour camera is about .7 of the mono camera. So, not as bad as 2x2 binning. He felt that unless my seeing was really good the better camera would be the SX694. He felt the loss in sensitivity with the SX815 would not be worth the increased resolution (unless my seeing really warranted it). So, the choice narrows.

Peter

[gregbradley]

I agree. The 814 is more for short focal lengths like Slawomir is using. Its a bit limited to that and the 1120 well is really only for Hyperstars.

The STL11000XCM I used to have used a luminance filter. I doubt the Bayer filters are UV and IR corrected. They certainly aren't in DSLRs and have a separate UV/IR blocking filter.

So I would continue to use your Hutech IDAS or a clear UV/IR block filter luminance.

By the way how well does the IDAS filter work? I was considering putting one in the imaging train for all the filters to see if it improved contrast. Not sure how, I suppose it would screw into the front of the Trius or Proline.

If you want to use this for colour then resolution is not as critical hence a lot using 2x2 binning. Colour gradations in a colour image tend to be gradual not with sharp sudden borders like Luminance has.

I am finding blending CDK images taken at 2 metres with 9 micron pixels are working well with LRGB data from the Honders with a Trius 694 with 4.5 micron pixels. The fact the 694 pixels are evenly half the size may help - don't know, but it can't be a bad thing.

Thomas Davis uses an STL11 Mono and an STL11 OSC to good effect.

I think the key is the FOV of the colour needs to be quite a bit larger than the luminance so if they are not exactlyrotated you can still use them as you have a margin for error. My CDK data for example from a few years ago was quite rotated from the Honders data but fiortunately the Honders image is about half the size so I can rotate it a lot and still not have to crop. Lucky but it made me realise that is something to design into your choice of cameras on the 2 scopes. So with that I think you need APS sized colour and not worry so much about exact pixel sizing. That's probably more of a secondary concern.

The upcoming QHY ic16200 is APSH and 6 micron pixels with QE matching or exceeding KAF8300 so that could be ideal. I am keeping an eye on that one for the Honders as 6 microns is the perfect pixel size. Also 35K full well would be a blessing as I personally am a bit tired of the tiny wells of the Sony sensor. The KAF16803 is bullet proof for image processing the Sony a bit touchy. I noticed lately I have to be very careful with my bias as they can introduce fixed pattern noise - a grid which must be a manufacturing pattern. If you get deep into processing and notice it later then that's not good! The image is trashed.

Greg.

[PRejto (Peter)]

Hi Greg,


Thanks for the feedback.

Yes, I do agree that it would be much nicer to have a larger FOV for colour vs mono luminance. However, the only cameras I've considered so far to have a larger FOV, but a not terribly similar resolution to the luminance (.88 arc-sec) are the SBIG 8050SC and the QHY12. When I have binned my colour 2x2 in the past I thought it somewhat degraded the sharpness of my luminance data...but, perhaps that was/is a processing mistake (of which I make many!). So, I have thought that since I'm going to lose resolution using a one-shot camera I would be better served by not getting too far from .88. One other option I could consider is just another KAF8300 camera. The resolution on my TEC140 would be 1.14 arc-sec and the FOV 48.1 x 64 (vs my TEC180 37 x 49.4) so, considerably larger FOV. If Terry is correct about the .7 figure then my "real" resolution in colour would be ca. 1.62 arc-sec (1.14/.7). Interesting because that is pretty close to 2x2 binning of the camera on the TEC180 (1.76 arc-sec). I think the bottom line would be that I couldn't expect to do any better than if I used a KAF8300 on my TEC180 and always binned 2x2.

All this exercise does is drive my thinking around in a circle back to the Trius. Yes, smaller FOV, but better resolution and more sensitive. I'm not rushing to decide. Later this week I will ask some experts very direct questions at AIC.

Peter

PS: Another "crazy" idea floating in the background concerns the new AP QUAD4 reducer/flattener that has been tested on the TEC refractors by Roland. It would make the TEC140 f5. Together with the ATIK 490 (3.69 micron pix) resolution would be 1.06 Arc-sec and FOV 48 x 60. Huge drawbacks, I think; the QUAD4 is $1500 before fittings, and what's the point to do that and use a small CCD and small pix? I can't see it as a serious option.

[gregbradley]

Will that quad tcc work on the 180 as well? I tried a reducer and I got reasonable results on my 180.

A Trius 814 on an F5 TEC140 - sounds pretty good. I am finding the 694 on the Honders quite good.

Greg.

[PRejto (Peter)]

Yes, Roland tested the QUAD4 on the TEC160 and TEC180 as well. No doubt it's a great product. There will be one at AIC that I plan to inspect.

Still, I'm not exactly seeing the point of using a reducer ony to couple it together with a small pix camera. Why not just use a larger pix camera at f7? What, besides a flat field, would be the advantage? I don't think the scope would be any faster. My guess is the larger pix camera would win if the resolution is about the same.

Peter

[gregbradley]

Originally Posted by PRejto

Yes, Roland tested the QUAD4 on the TEC160 and TEC180 as well. No doubt it's a great product. There will be one at AIC that I plan to inspect.

Still, I'm not exactly seeing the point of using a reducer ony to couple it together with a small pix camera. Why not just use a larger pix camera at f7? What, besides a flat field, would be the advantage? I don't think the scope would be any faster. My guess is the larger pix camera would win if the resolution is about the same.

Peter

The 4.5 micron 694 pixels gives .98 arc sec/pixel on the 180 with the reducer which is the magic 1 arc sec/pixel formula. But yes per Ray's sensitivity thread the extra QE of the 694 may be secondary to the pixel size which was a squared in his formula. Larger pixels don't seem to have a lot of downside with those 2 scopes or any smaller refractor for that matter.

Certainly more robust for the stars.

Greg.