Affiliation
American Association of Variable Star Observers (AAVSO)
Thu, 07/12/2012 - 14:19
Well, after years of "good service" with my old Thomson CCD based camera, it is time to look for a new CCD. Before deciding what kind of camera, I would like to ask the group if anybody tried KAF8300 for photometry. I looked in the web, but apparently nobody used it for photometry purposes. I know that there are some cons (ABG, small pixels), but it has other pros (cost).
Anybody tested it?
Thank you in advance.
Alessandro Maitan
While I don't use the KAF8300, I just wanted to say that I was very disappointed when I saw the specs for the new SBIG cameras. Some of the same issues that you mentioned caught my eye immediately. In looking at their website, it looks like the older model cameras are being de-emphasized if not being discontinued soon in favor of "pretty picture" cameras. Apparently that is where the money is.
You may be able to pick up a more suitable camera on the used market for less money .... just a thought. I'll be hanging on to my old parallel port ST9 as long as I can.
....Tim
Looks like they are streamlining by using just one CCD.
Unfortunately, they picked one with just 25,000 e- well capacity.
Not the best for science.
I think you can make pretty pictures with 100,000 e- well depth as well.
THey do have the one $12,000 one though.
Guess I'll keep the ST8 and ST9 running.
I should not have given the series ST6 away either.
Ray
Ray,
Yes, Cameras seem to be getting expensive compared to the old SBIG ST series. Have you looked at any of the QSI 500 or 600 series cameras? They don't have what I am looking for, but they seem to have some well-constructed and well-cooled cameras. I have heard comments from those who should know, that the QSI filter wheels have much more reproducible positioning than many.
Brad Walter, WBY
One negative with the SBIG ST-8300M is that when the desiccant needs to be replaced, the camera will need to be taken apart versus the "plug and chamber" dessicant setup of the older SBIG series. Though not an immediate negative, it may become one when the dessicant needs to be changed. I can not speak for other CCD camera brands. The engineering of this model seems to accentuate cheaper costs over function. Kevin - PKV
One thing to keep in mind is that the max linear pixel depth for this camera for photometry is about 13000 ADU. The specs for the camera on the OPT website claim a 64.4 linear dynamic range. Since the read noise is 8 e- and although FLI doesn't give the gain, I am assuming with a saturation of 25,500 e- the gain is one. the linear criteria is normally 1% for amateur cameras so I am further assuming the linear dynamic range is determined by the max ADU for 1% linearity. Dynamic range is 20*LOG(ADU/RDNOISE and 64.4= 20*LOG(13277/8).
I did a little analysis to see how magnitude, and SNR varied with Max ADU of a typical star image for my set up. I did this simply by looking at a good image I had of M67 and taking photometric measurements including magnitude, SNR, CCD error equation results and Max ADU measurements of every star in the FOV. The results will vary with seeing, sky background, plate scale focus, and gain of the camera. My camera has a gain of 2.3. Therefore, to compare to a camera with a gain of 1.0, I converted all ADU to e-. FWHM in my area is often in the 3 arcsec range which equates to a FWHM of almost 5 pixels with my telescope and 9 micron pixels. A night with 3 pixel FWHM is an outstanding night. With ~4.8 pixel FWHM I need a max flux of about 1950 e- compared to a background of around 575 e- to get SNR above 100. the range from Max e- of 1950 to 13000 under these typical conditions would give me a useful linear range of about 2.2 magnitudes between the dimmest star of interest with SNR of 100 to the brightest star of interest that would remain in the linear range. However, I am often trying to achieve 2-3 millimag precision which means I need SNR of 400 or better. For my equipment that means the dimmest star of interest needs max flux of 6900 e- which gives me a useful linear range of only about 0.5 magnitudes to 13000 e-.
Therefore, a 25,500e- full well depth camera would not be a good choice for me even if the pixels were larger than the 5+ microns of the 8300 CCD. Except for the very small array size the KAF0401 chip I am using is a good choice for the precision I am trying to achieve. In larger format chips the 6303 would be a good choice even though it is a lot more expensive than the 8300.
All of this long winded explanation is to give background for my suggestion that you think about what kind of photometry you want to do, the conditions in your area and all of the other variables I mentioned above before buying a camera with such a shallow well depth. Of course you can always take multiple images and bin the data, but that increases the time and trouble of making observations.
I have nothing to do with FLI, but I suggest looking over their offerings. the microline cameras have a lot of features in a small, light package. I know several people who use them and haven't heard a complaint. They are light enough even in the larger formats that you don't have to replace a lot of your image train to handle added weight compared to older smaller format cameras. You also don't have to worry about desiccant replacement. The CCD chamber is permanently sealed.
Good luck on your camera hunt.
Brad Walter, WBY
It is possible to linearize CCD data when it's above the linear range...but that requires analysis of your camera, and post-processing...and can only help linearize data that is 'somewhat' above the linear limit (depends on the particulars of your CCD).
We may see data-linearizing happening more often...if we only have small-pixel CCD's available in the future. (But it may be unrealistic to expect 0.002 magnitude precision with linearized data. Real-world testing will answer this question.)
SIBG quotes a gain of 0.37e-/ADU.
Note that you can often bin KAF chips 2x2...and the readout node has a larger capacity than individual wells...so that you get a larger effective pixel with a larger effective well depth. I have not looked at the KAF8300 data sheet in detail to see if this can be done with this chip.
>Note that you can often bin KAF chips 2x2...and the readout node has a larger capacity than individual wells...so that you get a larger effective pixel with a larger effective well depth.
On chip binning doe not help you with non linearity. each of the individual pixels that you bin together will become non linear above approximately 13,000 e-. So you can have a 2x2 binned readout that is less than 52,000 but if one of the 4 binned pixels is over 13,000 it will be in its non linear range. The effect of the non linearity is divided by up to 4, but it is there, regardless.
I am using KAF-8300 (in QHY-9 ccd camera). I am very satisfied with this model . The worst issue about it is its high thermal noise compared to other ccd chips, wich requires double stage cooling.
I chose this detector because it allows a large field of view (40x30 arcmin) with my optical tube (8" RC 1600mm fl) thanks to its 18x13mm size. Although this ccd has pixels too small for my focal length and average seeing, it fits very well to my optical system at 3x3 binning.
Because it's an ABG chip, you have to be careful to adjust exposure times to not enter non-lineal response level.
I cannot compare it to Thompson or other Kodak chips because I didn't use none of these. I come from Sony ICX, smaller but less noisy, and this KAF-8300 is my first full frame, serious detector.
KAF-8300 is able to be used for photometry but perhaps if I would have a wide budget I would choose a non-ABG ccd equivalent in size.
First of all I'd like to thank you everyone that is partecipating in this discussion. Just a few words on what I'm doing. I'm partecipating in a pro am collaboration with Asiago National Obesravtory called ANS (Asiago Novae and Symbiotic Star Collaboration) with two friends. We mainly observe symbiotic stars and novae and we normally use our group telescope (a 16" newton scope equipped with a KAF261 CCD camera and UBVRI filters) and a 10" SC with VRIby filters. Now we would like to improve the perfromance of our CCD camera with something better than the aged Thomson TH7895M (512x512 pixel 19 um) with a max QE of 40% and blue cut off at 400 nm (practically is blind at blue mag). Our target is observing our target with total errors less than 0.02 mag (not the instrumental error). The idea of using a CCD Camera equipped with KAF 8300 is only bound to the "fairly" low price without going on the second hand market. However as far as I understand KAF8300 based CCD cameras are mainly used for nice photo.
On the other hand, I've the experience in reducing data in binned 2x2 images (taken with a KAF402Me CCD) with good results therefore considering the SC focal length, using binning 2x2 or 3x3 will be a must.
Alessandro Maitan
As you made mention of camera performance in blue, I can tell you, in favour of KAF-8300, that its response in B is pretty well (not as good as in V and R of course).
A question not related to this posting, ¿Would it be possible to get some information about ANS in some website?
Regards
Miguel
Alessandro Maitan wrote:
"On the other hand, I've the experience in reducing data in binned 2x2 images (taken with a KAF402Me CCD) with good results therefore considering the SC focal length, using binning 2x2 or 3x3 will be a must."
I guess I don't understand this. Why is binning this chip a "must?" Do you have bandwidth problems? Binning changes your arcsec/pix and image size (in KB) but doesn't help with fov issues, for example.
I use the KAF 402 chip (in an older ST-7 and a newer ST-402) with good results. I've used them with both 10 and 14-in SCTs. The "trick" is to use a focal reducer. With the 10-in f/10 a 0.5x focal reducer produces 1.48 arcsec/pix and a fov of 18.9x12.6 arcmin fov. With the 14-in F/11 I use a 0.33X focal reducer which provides about the same.
Jim Roe, Bourbon, MO
I just got an answer from SBIG on my question, why there is no affordable CCD cam like the ST-series
left in their programm. They will have at least one (probably more) CCD cams in a few month. At least one with the same chip as the ST-8 had. It will only have the new STT-housing and everything around it (self guiding, better cooling and so on).
So maybe it is worth to wait for a new real measuring cam instead of dealing with the problems of a "pretty-picture-cam".
greetings
katy
I have done photometry with the Atik 383L+ which uses a KAF8300 CCD (mono). The results were very good. I binned 2x2 so I checked the linearity and found it to be linear to at least 65000ADU (R^2=0.9999) (less than 1% difference in any test measurement from linear response past 65,000 ADU). I have not testes un-binned linearity but a test on the Atik website (sorry don't have the link) showed it was linear to 65535. Don't let the ABG dissuade you (at least as far as the Atik implementation is concerned). V band differential photometry results typically have magnitude standard deviations of a series of observations of < 0.01 magnitude and agreed with published magnitudes to within 0.02 mag. I think if you are sensible then it is fine - I mean keep some headroom in the ADU to avoid sub-pixel saturation.
Cheers
Bob