Return to the Cloudy Nights Telescope Reviews home page


Astrophotography and Sketching >> CCD Imaging & Processing

Pages: 1 | 2 | 3 | 4 | 5 | 6 | (show all)
Chuck Faranda
professor emeritus
****

Reged: 03/31/07

Loc: Florida
Re: STT-8300 First light report new [Re: pfile]
      #5822855 - 04/25/13 07:14 PM Attachment (34 downloads)

Quote:

does anyone have any examples of what an RBI-spoiled sub looks like? the other night i did a bunch of 30 minute Ha subs with an STT-8300M @ -30C and there's nothing obvious to me that something could be wrong. maybe the target was not bright enough to cause this problem?




Here is a crop of a 600 second dark taken 20 minutes after a 600 second light of a faint nebula (KAF-6303 sensor). No RBI mitigation applied, you can clearly see ghost stars and the nebula.


Post Extras: Print Post   Remind Me!   Notify Moderator  
MattThomas
Vendor (SBIG)


Reged: 07/28/06

Re: STT-8300 First light report new [Re: MattThomas]
      #5822911 - 04/25/13 07:46 PM

Ok, ok... I know no one asked. But here it is anyway.

From an STT-8300 operating at -35C.

We measure dark current by taking a 3 second dark frame and subtracting it from a longer dark frame and looking at the average charge left in the image. In this test I was using a 300 second dark frame.

With no light on the CCD from when it was warm, the charge accumulated in 300 seconds was 2.09 ADU, or 0.75e- (0.0025e- per second)

Then I enabled the RBI preflash and re-did the test. The charge accumulated in 300 seconds was 60.85 ADU, or 21.91e- (0.0730e- per second).

I kept RBI preflash enabled, redid the test and achieved a similar result. Thus by using RBI preflash the dark signal in a 300 second exposure is increased 29x. This will translate into a larger contribution of shot noise to the true noise in the image (as the ~61ADU of dark signal could be subtracted with a dark frame).

I then disabled the RBI pre-flash and continued measuring dark current with 300 second exposures. (Note that these were not necessarily back-to-back measurements - I had to step away from my desk for some of these, so the actual time for the charge to diminish cannot be determined from this test.)

The results were, in order of time from the RBI pre-flash:
11.55 ADU, 4.16e- (0.0139e-/sec)
6.29 ADU, 2.26e- (0.0075e-/sec)
4.99 ADU, 1.80e- (0.0060e-/sec)
3.60 ADU, 1.30e- (0.0043e-/sec)

So, more than 30 minutes after flashing the CCD there is still charge leaking out of the substrate at 70% the raw dark current rate of the KAF-8300.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Chuck Faranda
professor emeritus
****

Reged: 03/31/07

Loc: Florida
Re: STT-8300 First light report new [Re: MattThomas]
      #5823005 - 04/25/13 08:31 PM

Matt,

Are you saying this indicates that RBI is preferable to managing the leakage noise?

Here is an example of the 'noise' RBI can cause (I've seen worse) if no RBI mitigation is applied? I do not own a KAF-8300 camera without RBI mitigation, but I have seen images from others who do. It doesn't appear to be immune to RBI.

Once I the needed scope adapter arrives I will provide some with and without images for us to pick apart.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Peter in Reno
Postmaster
*****

Reged: 07/15/08

Loc: Reno, NV
Re: STT-8300 First light report new [Re: Chuck Faranda]
      #5823152 - 04/25/13 09:11 PM

Quote:

Quote:

does anyone have any examples of what an RBI-spoiled sub looks like? the other night i did a bunch of 30 minute Ha subs with an STT-8300M @ -30C and there's nothing obvious to me that something could be wrong. maybe the target was not bright enough to cause this problem?




Here is a crop of a 600 second dark taken 20 minutes after a 600 second light of a faint nebula (KAF-6303 sensor). No RBI mitigation applied, you can clearly see ghost stars and the nebula.




Whoa!!!! What if you dither the subs, do you get double stars in every image? Do all Kodak chips do that or just yours?

I cool my Sony ICX694 CCD to as low as -15C and image as long as 30 minutes and never seen this issue. Colder than -15C for Sony chips make very little to no difference thanks to their super low noise chips.

Peter


Post Extras: Print Post   Remind Me!   Notify Moderator  
Chuck Faranda
professor emeritus
****

Reged: 03/31/07

Loc: Florida
Re: STT-8300 First light report new [Re: Peter in Reno]
      #5823190 - 04/25/13 09:30 PM

Full frame, front illumated sensors are prone to this. The Sony has an interline sensor.

Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: STT-8300 First light report new [Re: MattThomas]
      #5823509 - 04/26/13 02:02 AM

Hi-

Thanks - that is actually what I was looking for. If at -35C you still get much more charge released in the first 5 minutes than in any time after, then in a long exposure the contribution from rbi leakage is relatively fixed and fairly independent of exposure time. In that sense it acts more like read noise because it is relatively independent of subexposure time for long subexposures. In an exposure above 10 minutes, it appears to add about 25e of uniform dark current. If the read noise is about 9e, then the effective read noise in a long exposure with pre-flash enabled is about sqrt(25+81)= 10.3e.

Obviously this is not actually "read noise" because it wouldn't show in a short exposure - but it is like read noise in that it is independent of exposure time for long exposures.

I don't consider this a huge impact on the read noise - and if the dark current remains so low at -35C then overall this is good for long exposures. With pre-flash enabled, long exposures would act like a slightly higher read noise ccd with fairly small dark current - and presumably no rbi artifacts.

Thanks,
Frank


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: STT-8300 First light report new [Re: freestar8n]
      #5823624 - 04/26/13 05:44 AM

I did a fit to the data provided and even though the exposures weren't back to back, they fit well to an exponential decay of rbi charge on top of a constant dark current. The numbers I get, for -35C, are:

rbi total charge, q: 23.4 e
constant dark current, d: 0.0055 e/s
rbi charge 1/e decay time, T: 150 seconds

where the assumed formula for total dark current in an exposure of time t is:

D(t) = d*t+q*(1-e^(-t/T))

q represents the maximum impact from rbi pre-flash, and I assume that if the pre-flash exposure is weaker then q will be smaller, but then it might allow rbi artifacts to remain. So q by itself might be different from different vendors - but you would need to make sure it was effective at eliminating rbi artifacts.

Anyway - thanks for the data - and it all looks encouraging for 8300 with rbi pre-flash.

One note is that with this nonlinear contribution from the rbi discharge, it would be dangerous to do any scaling of darks, so you would want dark times to match light times. You might also benefit from the use of flat darks here - but bias might still be just as good and possibly better than a scaled dark.

Frank


Post Extras: Print Post   Remind Me!   Notify Moderator  
jerryyyyy
Pooh-Bah


Reged: 10/06/11

Loc: Stanford, California
Re: STT-8300 First light report new [Re: freestar8n]
      #5823952 - 04/26/13 09:59 AM

Guys, could someone be so kind as explain to a poor soul like me with 17 months astroimaging experience what the issue with RBI might be?

I went away from this thread for one day and there are 20 posts on it... WOW.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Jared
Postmaster
*****

Reged: 10/11/05

Loc: Piedmont, California, U.S.
Re: STT-8300 First light report new [Re: MattThomas]
      #5824024 - 04/26/13 10:29 AM

Quote:

<snip>The "detrapping" is a function of time. It can take more than 5-10 minutes at cold temperature for the substrate to fully purge itself of the flooding. Even longer for some CCDs. (Note that the KAF-9000 CCD takes minutes to clear the RBI even at room temperature!!!) This "detrapping" will appear as added unwanted signal (a.k.a. "noise") in your image. In fact you may even get a ghost image of some of the surface treatment (e.g. polishing) of the silicon.<snip>




I was wondering what caused that! I stacked up 50 or so dark frames at -35C on my STXL-6303 with RBI enabled and noticed an obvious ghost image, but had no idea what was causing it. It looked almost like a flat frame from a back-illuminated CCD (where the thinning process leaves a similar "ghost" image).

I'll try a set of darks without RBI enabled, and also measure noise differences with RBI on and off. Sounds like there may be situations--especially narrow band images where read noise is critical--to go without RBI enabled.

Thanks for the informative post.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Jared
Postmaster
*****

Reged: 10/11/05

Loc: Piedmont, California, U.S.
Re: STT-8300 First light report new [Re: jerryyyyy]
      #5824080 - 04/26/13 10:49 AM

Quote:

Guys, could someone be so kind as explain to a poor soul like me with 17 months astroimaging experience what the issue with RBI might be?

I went away from this thread for one day and there are 20 posts on it... WOW.




Kodak KAF chips (and perhaps other full-frame, front illuminated chips) suffer from something called Residual Bulk Image. The basic problem is that when these chips are exposed to a bright subject--something that saturates or nearly saturates the detector at a given pixel--some of the charge goes into the substrate of the chip and then can leak into the wells on subsequent images. Think of it like looking at a bright light for a few seconds with your eye and then looking at a blank wall--you will see the bright light "ghost" on the blank wall. KAF chips do the same thing, and when you stack up images you can get ghosts in the stacked image.

The standard solution to this problem is to "pre flash" the chip with an LED so that the RBI covers every pixel equally. As Matt Thomas was explaining, you get higher noise levels, but no ghosting. It's like adding a second source of dark current as the RBI leaks into the wells.

The second option is to just raise the temperature of the CCD so the RBI leaks out faster--letting it quickly decay to zero. Then, when you re-cool your chip there is no RBI on your next exposure. However, warming and cooling your chip between every pair of exposures eats up a lot of exposure time. It might still be reasonable if you were taking long narrowband subexposures, though.


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: STT-8300 First light report new [Re: jerryyyyy]
      #5824082 - 04/26/13 10:50 AM

Quote:

Guys, could someone be so kind as explain to a poor soul like me with 17 months astroimaging experience what the issue with RBI might be?




You want low noise in an image so you want low read noise and low dark current. You can't control read noise much but you can lower dark current by cooling the ccd. Unfortunately cooling the chip a lot causes it to trap charge, which persists after the image is read out, and it shows again in the next image. It shows typically as weird ghosting or nebulosity around stars - especially when the images are dithered.

So it is a real problem when it happens.

Several years ago some companies offered a pre-flash system that blasts the ccd with a uniform light before each exposure. This doesn't remove the stored charge, but it leaves a uniform charge across the ccd that doesn't show as an image artifact - but it does result in increased noise.

So if you have rbi pre-flash in your ccd, you can either run at a higher temperature so there is no rbi problem, but there is higher noise - or you can run very cool but use the pre-flash - which also introduces noise.

The point of my notes above is that the impact of rbi pre-flash is basically equivalent to a slightly higher read noise - rather than a slightly higher dark current. If you know the various parameters, you can make an optimal choice of what to do.

Based on what I have seen I would do:

1) For short exposures around 5m I would run at a higher temperature to avoid rbi and not use pre-flash

2) For long exposures that are sky limited I also would run at higher temperature to avoid rbi and not use pre-flash

3) But for long exposures that would be limited by dark current rather than sky or read noise, I would run very cold and use pre-flash to avoid rbi. In that case I would effectively be using a very low dark current with a slightly higher read noise.

I am considering a new ccd and if these numbers for rbi and dark current for the 8300 are representative, then the 8300 becomes much more attractive, mainly due to its size, than the sony 694.

Frank


Post Extras: Print Post   Remind Me!   Notify Moderator  
Raginar
Postmaster
*****

Reged: 10/19/10

Loc: Rapid CIty, SD
Re: STT-8300 First light report new [Re: freestar8n]
      #5824431 - 04/26/13 01:30 PM

Does this affect the ST-10 at all?

Post Extras: Print Post   Remind Me!   Notify Moderator  
Alph
Carpal Tunnel


Reged: 11/23/06

Loc: Melmac
Re: STT-8300 First light report new [Re: Raginar]
      #5824704 - 04/26/13 03:38 PM

Quote:

Does this affect the ST-10 at all?



Yes. The KAF-3200 is prone to RBI. You can safely assume that every Kodak NABG CCD will have significant RBI. Anti-blooming gates help to mitigate RBI. They drain excess charge. The KAF-8300 ABG performance is pretty high 1000x and this probably explains why it is more resistant to RBI. RBI shouldn’t be an issue if you don’t saturate pixels and make multiple sub-exposures of the same target over the course of your imaging session. Guide, baby, guide! As Sarah would have said.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Raginar
Postmaster
*****

Reged: 10/19/10

Loc: Rapid CIty, SD
Re: STT-8300 First light report new [Re: Alph]
      #5825606 - 04/27/13 02:06 AM

Alph,

Thank you. I've been wondering what the 'ghosting' is and I think this explains it.


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: STT-8300 First light report new [Re: Raginar]
      #5825643 - 04/27/13 03:04 AM

There have been several threads in CN on problems with RBI using the 8300 ccd. It might help this thread if anyone has a recent example to show with the 8300. The idea that images should avoid any saturation and be precisely aligned in each sub-exposure completely ignores the desire to dither exposures, and the common occurrence of nebulosity containing bright stars - or even galaxies with bright foreground stars. Even an object like the bubble nebula has a patch of pure nebulosity so bright that it could induce rbi artifacts.

At the same time, it is true that you can just run the ccd at higher temperature to reduce the impact of rbi - and that is another place where I find the specs for the 8300 all over the map. The dark current is supposedly related to the sensor itself and should not be different among vendors (as opposed to the read noise which depends on the support electronics) but I am finding values for dark current over a wide range.

If one 8300 achieves 0.004 e/s at -15C and another requires -30C to reach that level, then the first one might do fine without pre-flash while the second requires pre-flash. I would like to see firmer specs on values such as dark current and read noise that are neither optimistic nor pessimistic - so that if you buy such a camera and it does not meet the spec you can return it for one that does.

With rbi pre-flash, I would also like specs such as I derived above - which indicate the total charge released due to pre-flash, and the time constant of its release at a given temperature.

Frank


Post Extras: Print Post   Remind Me!   Notify Moderator  
Hilmi
Post Laureate
*****

Reged: 03/07/10

Loc: Muscat, Sultanate of Oman
Re: STT-8300 First light report new [Re: freestar8n]
      #5825724 - 04/27/13 06:01 AM

Since the topic of RBI of broader interest than purely with the STT-8300 cameras, shouldn't we start a separate topic so that it is easy to find should somebody be searching for RBI related information?

Post Extras: Print Post   Remind Me!   Notify Moderator  
Chuck Faranda
professor emeritus
****

Reged: 03/31/07

Loc: Florida
Re: STT-8300 First light report new [Re: Hilmi]
      #5825745 - 04/27/13 06:53 AM

Quote:

Since the topic of RBI of broader interest than purely with the STT-8300 cameras, shouldn't we start a separate topic so that it is easy to find should somebody be searching for RBI related information?




True, and I'm fine with your suggestion. However, I still have a few questions about how SBIG approaches RBI mitigation as it relates to user settable conditions for flood/flush/integrate. This ties in to the technical questions raised by Frank in a previous post. For example, does the predefined 1 second flash and single flush Matt spoke to accomplish the 100X full well level to ensure all traps are filled as Janesick (2001, pg. 664) recommends? If not, then (for me) this will still be a SBIG specific topic related to the implementation of changes that allow client software access to these default settings.


Post Extras: Print Post   Remind Me!   Notify Moderator  
dawziecat
Carpal Tunnel
*****

Reged: 10/20/10

Loc: Rural Nova Scotia
Re: STT-8300 First light report new [Re: Chuck Faranda]
      #5825821 - 04/27/13 07:55 AM

I remain puzzled as I read through the thread as to what to do about RBI.
First off, I have never personally seen it on the hundreds (thousands?) of subs I've taken with my ST-8300M. Maybe it's there on some and I'm not sophisticated enough to spot it? Could well be.
My camera does not offer pre-flash to mitigate RBI.So, what do I do about it?

1/Run at minus 10C instead of minus 20? Even zero C?
2/Run at minus 20C but turn cooling off for a couple of minutes between subs? Perfectly feasible for long NB subs but not practical for LRGB subs of 5 minutes or so.
3/ Just ignore the whole thing until it actually bites me?

Even if my camera did offer pre-flash, after reading here, I'd be reluctant to use it. Raising the overall noise level, quite irregardless of how "uniformly" you do it, seems counter to what CCD imaging is supposed to be about.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Chuck Faranda
professor emeritus
****

Reged: 03/31/07

Loc: Florida
Re: STT-8300 First light report new [Re: dawziecat]
      #5825882 - 04/27/13 08:48 AM

Quote:

I remain puzzled as I read through the thread as to what to do about RBI.<snip>




My questions were posed because the STT cameras implements enhanced cooling and RBI mitigation. Janesick (isbn:0819436984) details the relationship between temperature and RBI as well as the other things we have touched on here. So, I feel the implementation is important when I compare them to the accepted scientific data on the subject.

I hope my inquiries haven't prompted anyone to think they should fix something that isn't broken for them.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Hilmi
Post Laureate
*****

Reged: 03/07/10

Loc: Muscat, Sultanate of Oman
Re: STT-8300 First light report new [Re: Chuck Faranda]
      #5825902 - 04/27/13 09:04 AM

Does taking multiple short exposures clear the RBI?

Post Extras: Print Post   Remind Me!   Notify Moderator  
Pages: 1 | 2 | 3 | 4 | 5 | 6 | (show all)


Extra information
19 registered and 21 anonymous users are browsing this forum.

Moderator:  Oldfield, droid, bilgebay, WOBentley, dr.who 

Print Thread

Forum Permissions
      You cannot start new topics
      You cannot reply to topics
      HTML is disabled
      UBBCode is enabled


Thread views: 5634

Jump to

CN Forums Home




Cloudy Nights Sponsor: Astronomics