Atik 460 or Atik 4000?

[pete_l]

Sensitivity (for NB)...460 ... over size ...4000... - I think that's what it boils down to ............. And I can't decide!

Well, the sensitivity side is easy. The 4000 has a QE of 55% and the 460 has a QE of 77% according to the data sheets - though these figures are less at Ha wavelengths. Presuming the 460 is still more sensitive, that's like getting longer exposures for free: more signal electrons in the bucket for the same exposure time.

Regarding chip size. is it possible to compensate for that by buying a FR with some (or al l ) of the money saved. Then you'll have lower magnification, so a smaller image and therefore roughly the same image scale. As an added benefit, the lower focal ratio increases the brightness per pixel so giving you even more precious photons in each well.

[Spikey]

Well that's good news about the filters. I'd incorrectly assumed the chip was too big. I'll check up next time

[robbieince]

not quite as simple as that - pixel area makes a differemce

I think it works like this

4000, = 7.4 micron pixels so area is 7.4 x 7.4 = 54.76 *55 % QE = 30.118

460 4.54 micron pixels so area is 4.54 x 4.54 = 20.62 x 77%QE = 15.877

so the 4000 is about 1.9 times more sensitive at peak QE

Does that make sense?

[swag72]

Sorry Rob .............. No!!

So the 4000 is more sensitive? Even in narrowband?

[robbieince]

Doesn't feel right I agree. That assumes all the electronics ie gain etc are equal which I guess they aren't.

But I do know that large pixels are always going to be more sensitive than small - all else being equal.

Regards

Rob

[swag72]

But I had just today accepted that I would have to sacrifice the square chip for the NB sensitivity.

Olly - Do you have any NB data taken with an Atik 4000? Now I'm all confused again!!!

[robbieince]

Sorry - not my intention to confuse. Just commenting that all else being equal, bigger pixels are more sensitive.

I'll try and dig out some NB data I have for my 4000 too.

Regards

Rob

[swag72]

Thanks Rob - Would really appreciate seeing some 4000 NB data

[pete_l]

not quite as simple as that - pixel area makes a differemce

I think it works like this

4000, = 7.4 micron pixels so area is 7.4 x 7.4 = 54.76 *55 % QE = 30.118

460 4.54 micron pixels so area is 4.54 x 4.54 = 20.62 x 77%QE = 15.877

so the 4000 is about 1.9 times more sensitive at peak QE

Does that make sense?

I used to think that, too. However I read this document Quantum efficiency characterization of back-illuminated CCDs (I had to put my maths head on ). Specifically, equation (1). It appears the QE is measured for the CCD as a whole and then normalised to account for the CCD area compared to the area of the reference photodiode. So individual pixel size and overall CCD area are taken out of the equation, just leaving the ratio of photons that contribute to ADU and those that don't.

[Martin-Devon]

There is a calculation which Ian King gave me for determining the arc/second sampling rate per pixel for a CCD:- 206,215 x pixel size in mm/focal length in mm. Lower figures are best, since this translates into more sensitivity and sharper images. For a 530mm FL scope the figures are 1.77 arc seconds/pix for the 460, and 2.88 arc seconds/pix for the 4000.

Overall though I think with this discussion you can't really go wrong with either the 460 or the 4000, both have slight plusses and minuses over each other, but they are marginal. I wanted something with a bigger field of view than my current Atik 314L which has a field size of 58 x 45 arc minutes. By choosing the 460 I get a standard field size of 81 x 64, and with a focal reducer, this increases to 112 x 89. To put this in perspective the arc minute field view for the 4000 is 98 x 98.

So I think I can cover a reasonable number of field options with the 460 +/- the reducer, and it is far less noisy than the 4000, and has a better sampling rate figure/sensitivity - so for narrowband I think that the 460 with its latest generation Sony chip (ICX694) is my choice. Having said that - I'm sure the 4000 is also an excellent CCD likewise - I know lot of folks have this and are very happy with it. The 460 is the "new kid on the block" though, and I see several SGL members are now buying this, so we'll soon see a lot of images and what this can do.

[GlassWalker]

So individual pixel size and overall CCD area are taken out of the equation, just leaving the ratio of photons that contribute to ADU and those that don't.

Maybe I'm missing something but I don't see that as disagreeing with the earlier post trying to take into consideration pixel area. QE itself may be area independent, but you then do need to figure the area once you start looking at real devices. All else being equal, a bigger pixel will get more photons on it than a smaller one, similarly comparing total sensor sizes. Of course, this is complicated when things are not equal...

[pete_l]

Maybe I'm missing something but I don't see that as disagreeing with the earlier post trying to take into consideration pixel area. QE itself may be area independent, but you then do need to figure the area once you start looking at real devices. All else being equal, a bigger pixel will get more photons on it than a smaller one, similarly comparing total sensor sizes. Of course, this is complicated when things are not equal...

Oh yes, without a doubt. The problem is, all things are not equal .

I'm no expert and I'm making this up as I go along, so please chip in when I go off the road.

Suppose you're imaging a DSO (i.e. not a star / point source) of say: 20x10 arc-minutes (one of the larger Messiers for example) and you have a telescope with a fixed focal length - 1000mm for the sake of argument. Then with a CCD with 9 micron pixels, your image scale is 1.86 arcsec per pixel (32 pixels per arcminute) and your DSO will illuminate an area of 640 x 320 pixels; quite a decent size. If that DSO is squirting enough photons per second into the aperture of your telescope, such that 1,000,000 per second arrive at the focal plane of your CCD, then they will be spread across the 640*320 = 200,000-odd pixels.

Now, if your CCD has a QE of 55%, then 550,000 photons per second will be turned into electrons within the CCD - roughly 2.75 photons per pixel per second.

For another CCD with half that size of pixels: 4.5 microns, you'll be imaging at 0.95 arcsec/px (64 pixels/arc-min) and with the same telescope, the same DSO will illuminate 1280 x 640 ~ 800,000 pixels - so it produces an image 4 times larger. If that CCD has a QE of 77%, those same 1,000,000 photons per second will turn into 770,000 photons per second in the CCD, or roughly 0.94 photons per pixel per second. So each pixel catches 34% of the photons that the larger sized pixels does. In that respect: yes the larger sized pixels ARE more sensitive.

But what happens if we either bin the smaller pixels, or use a shorter focal length telescope to match the same image scale we achieved with the 9 micron pixels? In that case, presuming the same aperture size (so the same 1,000,000 photons per second on the CCD) but half the focal length, we're back to the original image scale of 1.86 arcseconds per pixel - this time for the 4.5 micron pixels. Also, the DSO we are imaging will now only illuminate the same number of (smaller) pixels as in the original example: 640 x 320, so the resulting image we get on our LCD monitors will be the same size as in the original case. Here's the good bit. But with the higher QE of 77%, the CCD catches 770,000 photons per second across the same number of pixels that the 9 micron CCD did - instead of the 550,000 due to the different QE of the two chips. Hence each pixel now collects 3.85 photons per second (c.f. 2.75) - a 40% increase, but still producing the same sized image.

So the conclusion (if anyone's still reading) is that for the same image scale, the CCD chip with the higher QE will catch more photons in a given time.

[swag72]

Here we go again - How many times do I ned to change my mind?!!!!

So with a 500mm scope you're effectively saying that with the 460 I'd collect 40% more photons?

[pete_l]

Here we go again - How many times do I ned to change my mind?!!!!

So with a 500mm scope you're effectively saying that with the 460 I'd collect 40% more photons?

The example I gave was hypothetical. The pixel sizes for the 4000 is 7.4u - (not the 9u I used to make the example easier) and for the 460 it's 4.54u. However if you compare the 4000 and the 460 with each camera operating at the same image scale, then yes, if I got my sums right: I reckon that in the same exposure time. with the same image scale and with the same telescope aperture, the 460 will collect more photons and the proportion will be that of the QEs of the two cameras. That would require the 460 to be in a telescope+reducer combination with a focal length of 4.54/7.4, i,e, 60% of the one that's comparable to the 4000 + telescope.

So far as captured photons goes, for the 4000 and the 460, the relationship is 77 / 55 = 1.4 times as many in the same time, for the same image scale, for the higher QE chip.

I'd strongly advise you to work through this to your own satisfaction as I am notorious for barking up the wrong tree, making false assumptions and (sometimes) even working in base10, when I thought I had my calculator using hexadecimal.

[Black Knight]

So the conclusion (if anyone's still reading)

I'm still reading. I'm finding this whole thread very interesting and um, illuminating

[ollypenrice]

I want both Olly, as wide a field as possible and for narrowband predominantly. Faced with equal importance .......... now what? Have you known people to do NB with the 4000?

Of course you want both - you're a woman!

I do NB in the 4000 mono, mostly Ha and sometimes a bit of O111.

Olly

Ha and O111 combined in greyscale, Tak FSQ85, 2 panel.

200mm camera lens, Ha:

[ollypenrice]

PS, I'm sorry but I can't remember the exposure times. The Veil, which I think has gone fairly deep, was not extravagent, maybe 2.5 hours per filter per panel at F5.3.

I can confirm that 1.25 filters are fine, even at F3.9, for the 4000.

Please note that I'm not flying a kite for the 4000, here. The 460 seems great, though I want more and more and more chip size to feed my widefield cravings...

Olly

[swag72]

They look nice Olly - Me likey!!!! Changed my mind ................ AGAIN!!!!!

[ollypenrice]

They look nice Olly - Me likey!!!! Changed my mind ................ AGAIN!!!!!

But of course: you're a woman! (I never change my mind. Well, I did once. Or did I? No, maybe not. I'll ask my wife...)

Olly

[jnc71106]

Surely the f ratio your currently imaging at must be accounted for. The 314l+ is a good cam with very good sensitivity I've seen the amount of time you spend imaging Sara surely moving to a less sensitive cam will mean having to spend even longer on your targets

If the 460ex is as good if not better than the 314 it would be the logical choice

Not that this hobby is ever logical mind you