Choice of CCD chips for imaging

[gregbradley]

I thought this was an excellent post about one of the aspects when it comes to choice of chip for imaging. This is from Tim Khan of Florida who is very knowledgeable about CCD chips and the technology behind them:

Matt,

Regarding you question on camera match-up:

The rule of thumb is 1.5 arcsec/pixel; however I do not agree to this rule.

Having a ML8300 and ML16803 I get 0.84 and 1.4 arcsec/pixel respectively with my 7f7#7. I really like the small/fine pixel scales, but you have to compensate with increases # of exposures.

The 8300 sensor camera would be a great first camera; however, this would more demanding of you mount during guiding as small errors can become evident.

You also have choice of 6.8u pixel 3200 sensor and also 9u & 7.4u interline sensors.

The full frame sensors all suffer from Residual Bulk Image so it is important that you get a camera that can mitigate this problem; FLI is best at mitigating RBI. The interline does not suffer from RBI although it is physically present; there is a buried channel that is voltage biased during exposures and keeps the electrons in the epitaxial layer from migrating into the well regions. But due to this extra layer in the sensor, you loose 50% of the exposed well area and red sensitivity. The red sensitivity is lost since the longer wavelenght penetrate deeper into the sensor before converting into a charge, and some of that deep red goes past the buried channel before converting into a charge and is therefore lost.

So now the question is: clusters & galaxies, large diffuse nebula, tricolor narrowband?

With narrowband, SII and HA are deep reds (longer wavelengths), so you would want to go with a full frame sensor such as 6303 and 3200 (excellent sensitivity)

If your main goal is to do broadband images nebulas / large fov; you might want to consider 16803 or 11002 although it is interline. This is not to say you cannot do this with the 8300, there are many fine examples out there using the 8300.

You certainly have a scope that performs and will perform with a wide selection of cameras, but it boils down to what you want to image, and the time of year that you can image, and how much you want to spend.

Tim


So there is a bit to chosing your camera rather than just a brand or cost.

The actual type of chip (Kodak makes 2 basic types) will influence your results.

I have noticed what he says there - the 16803 being a KAF full frame chip seemed to be more red sensitive than my previous KAI11000 chipped camera.

There's a lot of discussion about this RBI. I would like to take a range of sample photos with and without its control to see for myself the effect on my images and if it is worth the extra setup to control it or when is it affecting my images and when is it minimal.

I imagine it would be poor practice to do dusk flats and then take a bunch of darks. You'd most likely see it then.

Or focus your KAF series chipped camera on a really bright star or the moon and then image a dim little galaxy or the Helix for example.

By the way there is now a 10 megapixel Sony chip out on the market. Sony has the lowest noise. Starlight Express offers a camera with one. I think its only one shot colour though.

Greg.

[Terry B]

Very interesting.
My only other comment about the different sensors is whether you want to use the camera for science (photometry and spectroscopy) as well as pretty pics. The linearity of the sensor then becomes much more important. NABG sensors are much more linear than their ABG versions but some of the newer large ABG sensors seem to be better than the older ones.
It would be interesting to see data on the linearity up to saturation for these chips.

[Paul Haese]

Agreed, sensor selection and RBI control is important. More so than any manufacturer. The QSI has RBI control which pops up in maxim under the advanced settings. I have always had it set to medium for the flush and not yet seen any echos of the previous image. Ideally if I could combine several camera's together with a 6303E sensor with ABG then I would be happy. This sensor is around the optimum in my opinion. Sure the larger sensors are great but that means having a larger focusor and that is not always possible. Plus that sort of real estate requires some very nice field correction and that means even more money. Finding the right camera for your gear, budget and uses is pretty hard and takes an aweful lot of research. Even then there are compromises.

Most of the time it looks to me like one also needs a variety of sensors as well as telescopes. Some are good for this and some are good for that. As Terry points out; a choice has to be made about what type of imaging you are going to do from the start.

[Bassnut]

Very interesting. Despite their age 3200 and 6303 chips are still the go for NF/NB imaging IMO, I just cant see anything coming close yet, unless shear MP and FOV are of interest, with seriously less sensitivity and as Paul says, huge money in field correction/other hardware.

[bert]

Im playing with a 6303e at the moment, setting it up with a moag/aol etc.

It is unbelievably sensitive. Even short exposures have a lots of detail, and it picks up lots of cosmic ray hits. This is my first time using a nabg camera and it is quite an experience.

Another thing that has not been mentioned with this chip is the well depth, up near 100000e from memory. It does like to be cool tho.... From what I have read it is especially sensitive in the red/near ir. Cant wait to get some images out of it!

One trouble is tho, although it red sensitive, most red nebulas are excited by bright star/s which are usually in the same field of view, and make it bloom heavily in LRGB. As Fred said its a narrow band weapon.

Brett

[higginsdj]

I do not put much stock in RBI. I do not say that it does not exist but I can say that I have been unable to see an impact in 10 years of precision photometry with KAF chips (ST-8, ST-9e, STL1001e).

The 1 occassion I was able to detect it, the 'ghost' image varied in brightness by less than 2% from the image background (next image inside 5 minutes) and was not apparent at all on the next image after that (another 5 minutes after that) and that was for a previously grossly over-exposed image where I had a mag 4 and 6 star imaged for 240 seconds on the STL at f/9.4 on a 14" scope.

Cheers

[Terry B]

I have been finding significant RBI on my spectroscopy images using both a ST8XME and a ST9E. If I take a spectrum of the neon reference and then take a data spectrum, the neon lines are clearly visible in the spectrum of the star. this is despite the neon lines only reaching levels of about 2000. To get around this I take my star images first, then the neon lines then take flats with a halogen light. The flats will have levels around 20000 and this swamps the neon lines. If I then tahe more data images the raised background form the flat is very obvious for about 15 mins but this shouldn't affect the spectrum.

[gregbradley]

The KAF3200 is 80% QE 55,000 electron well capacity 6.6um pixels ABG
The KAF6303 is 65% peak QE 100,000 well capacity 9um pixels non-ABG
The KAF8300 is 61% QE in a FLI with no cover slip 5.4um pixels ABG
The KAF16803 is 61% QE and 100,000 well capacity 9um pixels ABG
The KAI11002 is 50%QE and 60,000 well capacity ABG


ABG = Antiblooming.

Only KAF chips experience RBI not KAI interline chips. So not a problem with KAI11002.

I personally have not seen its effects in images. Not saying its not doing something but I haven't noticed it.

On paper they are all somewhat of a compromise except for KAF16803 which is a good allround performer. QE of 6303 is no higher overall than 8300 or 16803 but 3200ME is but it comes at a cost of slight microlens spikes (minor). 6303 is not antiblooming which means better for narrowband or limited to shorter exposures in LRGB. 8300 has small wells so again star bloating on bright stars could occur in some setups - again shorter exposure time is the solution.

If RBI is wrecking your images then select a KAI chip for your camera.
Perhaps one of these new Sony 10mp super exHAD chips.

Of the current sensors available I think KAF3200, 6303, 8300 and 16803 are the go and in Interline KAI 11002 and perhaps 29050. There is a 10100 and 8050 one shot colour with the new LRGB matrix for those who want to be pioneers. They could be excellent for imaging and easy to process.

Fred if QE and Narrowband with narrowfield is really your thing you'd be hard pressed to go past the Proline 4240 with back illuminated chip 13um pixels and 95% QE. I am thinking of getting one. Large pixels would be a good match for my CDK17. High QE would be a bonus boosted by the reducer giving F4.45 to the CDK so it would be CDK17 at F4.45 95% QE with large pixels for optimum sampling with my seeing and narrowband would be a breeze.

Greg.

[gregbradley]

One trouble is tho, although it red sensitive, most red nebulas are excited by bright star/s which are usually in the same field of view, and make it bloom heavily in LRGB. As Fred said its a narrow band weapon.

Brett[/QUOTE]

I was considering one of these cameras as well. Steve Crouch makes good use of his so the blooming can be controlled. Shorter exposure times in LRGB imaging would be the go to control the blooming and high cooling to reduce noise so stacking does not add too much noise.
Highish QE and high well capacity with largish pixels is a good formula there. Too bad the antiblooming version dropped so much in QE and is no longer made.

Greg.

[renormalised]

Greg, e2v (http://www.e2v.com/home/) also offer a much larger chip than the CCD42-40 for astroimaging...the CCD230-84, which is a back illuminated, 16.8MP chip with 15 micron pixels. Mind you, looking at the price of the Proline4240, if it had the larger chip you'd have to mortgage your house to buy one!!!.

[Bassnut]

Those figures are a bit random Greg, peak QE.
QE at 656nm (Ha) is what counts for me, its where I spend 80% of imaging time.

Things look a bit different then

The KAF3200 is 84%
The KAF6303 is 60%
The KAF8300 is 45%
The KAF16803 is 45%
The KAI11002 is 32%

I dont know the Proline 4240, ill check it out

[renormalised]

The 4240 has about 95% QE at 656nm with a 100000 full well capacity. Peak QE is 96%.

[Bassnut]

Well, if you get sick of it, Im here ready and waiting :D

[Bassnut]

Thats about as good as it gets :D, but GAUD, just checked the price :eyepop::eyepop:.

But then, that would be the last cam you ever buy, period.

[renormalised]

It's the sort of camera you'd be buying SWMBO flowers and diamond rings, every few weeks, for the next 10-15 years, as compensation and appeasement:):P

Either that, or mortgage the house:):P

[renormalised]

I'm guessing if the 42-40 is nigh on $38500, then a 230-84 equipped camera is probably going to be around $80000-$100000 to buy. It's a much larger piece of silicon real estate. That would require not only all the flowers and diamonds, but a villa on the Italian Lakes and free lifetime personal consultation by Yves Saint Laurent for SWMBO's fashion needs.

[bert]

Its a friends... if it was mine I'd lend it to you.:P

It does suffer from rbi. When I was aligning a scope I took an exposure of the moon, and then an exposure of a starfield, it was very pronounced. Kinda extreme situation though.

Brett

[gregbradley]

Good point Fred. It is a bit hard to read the Kodak specs smetimes as they don't stick to the same format from one chip to the next.
The 3200 is the best performer of the commonly used chips.

Greg

[gregbradley]

There are a few versions. The one I was referring to is about $14,000.

Its hard to find. I think Optcorp had it listed at one point or perhaps the FLI website.

Greg.