Jump to content


Photo

Considering an SBIG 4020

  • Please log in to reply
20 replies to this topic

#1 Gregk

Gregk

    Vanguard

  • *****
  • Posts: 2426
  • Joined: 19 Nov 2006
  • Loc: San Tan Valley, Az,

Posted 26 August 2013 - 11:49 AM

Hi

I am considering purchasing an SBIG 4020 Mono camera
It has the Built in FW and Guider. It is an older model 2008
with USB 1.0. IT has a 15mm x 15mm sensor.

Any one have any experience with this Camera? I am not doing research just AP (galaxies, nebs and clusters)

Well Debth seems to be twice that of the 8300 chip

Greg

#2 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 26 August 2013 - 12:56 PM

I think you mean a 4020. With the KAI-4020 sensor. 7.4um pixels, interline sensor, lower quantum efficiency than the 8300 (e.g. 30% at H-a line). For pretty pictures, it would take almost twice as long to capture as an 8300.

#3 Gregk

Gregk

    Vanguard

  • *****
  • Posts: 2426
  • Joined: 19 Nov 2006
  • Loc: San Tan Valley, Az,

Posted 26 August 2013 - 03:40 PM

Thanks I corrected the model #

I reviewd both Quantum charts and they seem almost
Identical. Were did you see the large delta re HA
at 640nm theyare both around 30% QE

Greg

#4 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 26 August 2013 - 03:47 PM

only the "no microlens, clear glass" version of the 8300 has 30% QE at the H-a line.

All the versions with microlens (clear glass and MAR glass) are at around 45% to 50% at H-a. To the best of my knowledge all the common 8300 cameras (SBIG, Atik, etc.) use the microlens version.

ref: page 17 of http://www.ccd.com/pdf/ccd_8300.pdf

#5 Gregk

Gregk

    Vanguard

  • *****
  • Posts: 2426
  • Joined: 19 Nov 2006
  • Loc: San Tan Valley, Az,

Posted 26 August 2013 - 06:43 PM

Ok thanks Not sure what I was looking at then.

#6 schmeah

schmeah

    Apollo

  • -----
  • Posts: 1403
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 26 August 2013 - 09:05 PM

I think you mean a 4020. With the KAI-4020 sensor. 7.4um pixels, interline sensor, lower quantum efficiency than the 8300 (e.g. 30% at H-a line). For pretty pictures, it would take almost twice as long to capture as an 8300.


I don't think that is strictly correct. The QE at Ha is 45 vs 32% for the 8300 vs the 4020. However, the 7.4 micron pixels of the 4020 capture 1.88x as many photons per pixel as the 5.4 micron pixels. So doing the math, the number of electrons generated per pixel is actually higher for the 4020. Also the dark noise and read noise are lower for the 4020/4022 favoring this chip in terms of SNR per pixel. Anyway I've been pleasantly suprised by the Ha sensitivity of my QSI 540 (4022 chip). So unless you are imaging at very short focal lengths and can realize the potential resolution benefit, I don't see any advantage for the 8300 in terms of producing pretty pictures.

#7 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 26 August 2013 - 10:04 PM

There's been a very long discussion of QE and pixel size. Suffice it to say that pixel size has no bearing on sensitivity. QE trumps pixel size and is the most important metric (the others being read noise and dark noise).

I know it seems counter intuitive and that big pixels are better. But this is not so. I can't repeat the arguments but head over to Ron Wodaski's Yahoo group for the bloody details.

#8 avarakin

avarakin

    Surveyor 1

  • *****
  • Posts: 1884
  • Joined: 13 Jul 2009
  • Loc: Parsippany NJ, USA

Posted 27 August 2013 - 12:29 AM

I would be concerned about USB 1: image downloads could be very slow.

Alex

#9 schmeah

schmeah

    Apollo

  • -----
  • Posts: 1403
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 27 August 2013 - 07:39 PM

There's been a very long discussion of QE and pixel size. Suffice it to say that pixel size has no bearing on sensitivity. QE trumps pixel size and is the most important metric (the others being read noise and dark noise).
I know it seems counter intuitive and that big pixels are better. But this is not so. I can't repeat the arguments but head over to Ron Wodaski's Yahoo group for the bloody details.




Forgive me if I missed that discussion. So if you would be so kind as to give a link to it ... Couldn't find it on a search here, and I'm really not interesting in joining another group in order to search it there. Or a simple summary in your own words would suffice ... You don't have to provide all the bloody details.

#10 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 27 August 2013 - 08:38 PM

Derek,

The gist of it is that a given extended object or star produces a fixed photon flux.

Let's say the resulting image circle is 20mm in diameter. The total photon flux over that entire circle is going to be fixed. So regardless of having large pixels or small pixels, the total photons you'll collect over the whole sensor is the same.

Hence, higher QE would always trump lower QE (because it's the conversion from photons to electrons).

Of course if you have fewer, larger pixels, you'd collect more photons (and hence more electrons) per pixel at the expense of lower resolution. So sure, you're getting more signal per pixel, but that's still one pixel of information.

As Stan Moore has said..

"The true mission and metric of a camera is how much information it can capture from the virtual image presented to it."

Less resolution = less information.

My mistake was assuming big pixels = more electrons, hence better signal-to-noise ratio, etc. But information isn't in the individual pixels, it's over the entire sensor. Low QE means you're converting less of the total incident photons in the virtual image, irrespective of all other factors. So the overall information over the entire sensor is going to be less.

#11 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 27 August 2013 - 09:00 PM

The above of course is assuming that your focal length is such that you are not oversampling with the smaller pixel sensor. Hence with the same FL you would be undersampling with the big pixel sensor (and hence throwing away detail).

Most people I know image at under 1000mm. At these focal lengths a 7.4um pixel like the 4020 would be > 2" per pixel, which I believe is undersampled. The 8300 would give around 1.25" to 1.5" per pixel, which is about right.

Now if the OP is using a 2000mm or 3000mm scope, then you're not getting any more detail with the 8300 since you're really oversampled. Hence in that case I would say the area of the 4020 pixels would trump the higher QE of the 8300 (the 8300 is converting more photons, but there are less electrons per pixel, and you're not getting any additional detail out of those numerous small pixels to compensate).

#12 schmeah

schmeah

    Apollo

  • -----
  • Posts: 1403
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 28 August 2013 - 03:40 PM

Thanks Orly. That's a nice succinct summary and I think I understand what you're saying. But regarding the initial response to the OP, I think it may be a stretch to say that it will take "twice as long" to make a pretty picture with the 4022 compared with the 8300 without knowing more about his imaging scopes and intended target. The 8300 may have a higher QE at Ha, but this is not true at other wavelengths, ie in blue and green. And he did say that he was interested in shooting galaxies and globs suggesting that he will be shooting at longer focal lengths where the larger pixels would be advantageous. So perhaps a better statement would have been, "if you plan to shoot predominantly emission nebulae with a short focal length system and your scope is well matched regarding focal length, seeing conditions and pixel size, then it may take a bit longer with the 4020 chip." But that's ignorIng other factors such as CCD noise characteristics, well depth etc. I don't know a lot about this stuff, but at least for me sometimes a simple subjective comparison is helpful. I recall when switching from a DSI Pro ll to a Pro III I found the DSI Pro III to be much less "sensitive" at Ha, requiring significantly longer exposures than the Pro II. The only major difference to my recollection was the smaller pixel size of the Pro III as they similar QE curves.

#13 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 28 August 2013 - 07:36 PM

I also remembered this thread - http://www.cloudynig...6045108/page...

where Mike Miller says "avoid interline." They're talking about the 11002, but that chip is very similar to the 4020, albeit it's larger.

#14 Gregk

Gregk

    Vanguard

  • *****
  • Posts: 2426
  • Joined: 19 Nov 2006
  • Loc: San Tan Valley, Az,

Posted 29 August 2013 - 02:29 PM

Something I picked up from the SBIG site


Well "depth" is usually measured in electrons.

Normalizing for area reveals that the 8300 is slightly "deeper":

4020:
Pixel Size = 7.4 x 7.4 microns
Full Well Capacity (NABG) = 40,000 e-
= 730e-/u^2
(ABG would be considerably shallower)

8300

3326 x 2504 pixels @ 5.4 microns
Full Well Capacity = 25,500 electrons
so 874e-/u^2

#15 jaddbd

jaddbd

    Viking 1

  • -----
  • Posts: 877
  • Joined: 18 May 2005
  • Loc: Woodbine, MD

Posted 29 August 2013 - 03:28 PM

I have an 8300 and a 4022. At long FL (F8/2000mm), with the mediocre seeing conditions at my site, a crude side by side comparison gives the edge for me to the 4022 (pixel res. of .76 arc sec) over the 8300 (pixel res. of .56 arc sec) with Luminance data. Single subs show a bit better data/contrast on the faint wispy stuff. The 8300 does better with Ha data and especially better with S II data. Narrowband I generally shoot at a shorter FLs (1530 or 650). I prefer the 8300 for my refractor also. Differences are mostly subtle in a my configurations and conditions (orange zone surrounded by red, seeing mid 2s to low 3s mostly). Better seeing and darker skies might tell a different story. I would not totally sell the 4022 short on it's narrowband abilities either.

John D

#16 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 29 August 2013 - 09:45 PM

Hi John, as I mentioned above, if your pixel scale is small enough that the 8300 would be oversampling, I think the 4022 would be more competitive. 0.76" vs 0.56" are both oversampled, so I can understand why you'd prefer the 4022.

One other thing, AFAIK all 8300s have anti-blooming. So comparing well depth of a NABG 4020 and an 8300 is not quite exact.

#17 jaddbd

jaddbd

    Viking 1

  • -----
  • Posts: 877
  • Joined: 18 May 2005
  • Loc: Woodbine, MD

Posted 29 August 2013 - 10:35 PM

Hi Orlyandico,

The 4022 is anti-blooming also (QSI640)...

Plus I usually run the larger scope with a CCDT67 reducer giving an FL of 1530mm and scale of 1.12" with the 4022 which seems to match well with the typical conditions here.

JD

#18 SL63 AMG

SL63 AMG

    Viking 1

  • *****
  • Posts: 864
  • Joined: 21 Dec 2009
  • Loc: Williamson, Arizona

Posted 30 August 2013 - 02:02 AM

Orly wrote:

My mistake was assuming big pixels = more electrons, hence better signal-to-noise ratio, etc.


I mean no disrespect as I am a novice at this stuff, but if your statement is true why is everyone binning their RGB subs in order to obtain better signal to noise ratio with shorter exposure times?

Personally, I prefer to shoot all my subs binned 1x1 and many people tell me I should bin my RGB 2x2.

Why? :question:

#19 schmeah

schmeah

    Apollo

  • -----
  • Posts: 1403
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 30 August 2013 - 10:46 AM

Derek wrote:

My mistake was assuming big pixels = more electrons, hence better signal-to-noise ratio, etc.


I mean no disrespect as I am a novice at this stuff, but if your statement is true why is everyone binning their RGB subs in order to obtain better signal to noise ratio with shorter exposure times?

Personally, I prefer to shoot all my subs binned 1x1 and many people tell me I should bin my RGB 2x2.

Why? :question:


Hi Dave, actually you misquoted me on that one. That was Orly's statement, not mine. Anyway I don't think any one is arguing the fact that larger pixels = more photon's per pixel and therefore all other variables being equal ie: CCD noise characteristics and QE that the larger pixelled CCD would be more sensitive and require less exposure time, hence the rationale for binning. The question is at what point does a higher QE trump the pixel size advantage when comparing different CCDs.

#20 Peter in Reno

Peter in Reno

    Fly Me to the Moon

  • *****
  • Posts: 6195
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 30 August 2013 - 11:43 AM

It's true that more photons may hit larger pixels but low QE means the photons are hitting the pixels and not registering.

Peter

#21 SL63 AMG

SL63 AMG

    Viking 1

  • *****
  • Posts: 864
  • Joined: 21 Dec 2009
  • Loc: Williamson, Arizona

Posted 30 August 2013 - 12:51 PM

I apologize. I had the right quote wrong name. I fixed it.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics