M67/NGC 2682 under very good seeing!

wrote:

On 25 Feb, 01:54, Anthony Ayiomamitis
wrote:

Anthony Ayiomamitis wrote:

Chris L Peterson wrote:

On Sun, 25 Feb 2007 01:43:12 +0200, Anthony Ayiomamitis
wrote:

My master luminance file is available at
http://www.perseus.gr/Downloads/M67-luminance.zipand which is 5.7 Mb
in size. It is based on an average combine of five six-min subs.

Hi Anthony-

Hi Chris,

Using Maxim and IRAF, I measured the average FWHM for stars in this
image at 4.2 pixels, 5.4 arcseconds. With AIP I got values all over the
place. I've previously determined in tests that AIP does not generate
accurate FWHM values. The ~5" FWHM of this image is typical of what I'd
expect for a 30 minute exposure under fair to good seeing conditions.

Are you sure this is just the stacked, calibrated image? The histogram
is completely abnormal. Stars are clipping at about 6500, and the
dimmest values are around 5300. I'd expect to see a range of data from
about 0 to 65535 (the full 16-bit range of the camera). If all you did
was align and average, I'd have to say something went very wrong.

I owe you an apology ... this includes RL and DDP (please see the name
of the file). Damn it!

Please redownload the same zip which has been replaced and includes the
master luminance with only an align and average combine.

PS. Using MAXIM/DL, please check out the FWHM for the star with centroid
303.474 and 698.294 (at 1.176!). Other sample stars are below 1.40.

Enjoy!

Anthony.- Hide quoted text -

- Show quoted text -


I tested it using both MaximDL and IRIS. Maxim gave an average of 2.3
over a sample of 20 stars ad mid-point of the hystogram. Iris fared a
bit tighter, at around 2.1. I'd say that the average fwhm was at
around 2.9", which is ok for clusters or start fields imaging but FAR
from being an excellent seeing.

Andrea, as I mention to Chris in my other reply, the raws were
characterized with lower FWHM's and I forgot to apply the image scale.
The latter, of course, is an automatic 27% error. Ooops!

BTW, there is a lot of thermal noise.
I'd venture to suggest that you create a map of hot pixels and kill
'em before creating the luminance file.

Can I trouble you for two to three sample pixel coordinates for my
education and I will proceed with building a defective pixel map. As I
use CCDSoft for reduction, I will check if it can import such a map. If
not, I will start doing reduction within AIP4Win which can accomodate
such a map.

I also have AstroArt V2 and I will see if it also accomodates pixel maps.

Thanks for the feedback!

BTW2, don't use the value of fwhm reported by Maxim if the reported
flatness value is zero.

Okay, thanks!

Anthony.

On 25 Feb, 12:15, Anthony Ayiomamitis
wrote:
wrote:
On 25 Feb, 01:54, Anthony Ayiomamitis
wrote:

Anthony Ayiomamitis wrote:

Chris L Peterson wrote:

On Sun, 25 Feb 2007 01:43:12 +0200, Anthony Ayiomamitis
wrote:

My master luminance file is available at
http://www.perseus.gr/Downloads/M67-...ce.zipandwhich is 5.7 Mb
in size. It is based on an average combine of five six-min subs.

Hi Anthony-

Hi Chris,

Using Maxim and IRAF, I measured the average FWHM for stars in this
image at 4.2 pixels, 5.4 arcseconds. With AIP I got values all over the
place. I've previously determined in tests that AIP does not generate
accurate FWHM values. The ~5" FWHM of this image is typical of what I'd
expect for a 30 minute exposure under fair to good seeing conditions.

Are you sure this is just the stacked, calibrated image? The histogram
is completely abnormal. Stars are clipping at about 6500, and the
dimmest values are around 5300. I'd expect to see a range of data from
about 0 to 65535 (the full 16-bit range of the camera). If all you did
was align and average, I'd have to say something went very wrong.

I owe you an apology ... this includes RL and DDP (please see the name
of the file). Damn it!

Please redownload the same zip which has been replaced and includes the
master luminance with only an align and average combine.

PS. Using MAXIM/DL, please check out the FWHM for the star with centroid
303.474 and 698.294 (at 1.176!). Other sample stars are below 1.40.

Enjoy!

Anthony.- Hide quoted text -

- Show quoted text -

I tested it using both MaximDL and IRIS. Maxim gave an average of 2.3
over a sample of 20 stars ad mid-point of the hystogram. Iris fared a
bit tighter, at around 2.1. I'd say that the average fwhm was at
around 2.9", which is ok for clusters or start fields imaging but FAR
from being an excellent seeing.

Andrea, as I mention to Chris in my other reply, the raws were
characterized with lower FWHM's and I forgot to apply the image scale.
The latter, of course, is an automatic 27% error. Ooops!

BTW, there is a lot of thermal noise.
I'd venture to suggest that you create a map of hot pixels and kill
'em before creating the luminance file.

Can I trouble you for two to three sample pixel coordinates for my
education and I will proceed with building a defective pixel map. As I
use CCDSoft for reduction, I will check if it can import such a map. If
not, I will start doing reduction within AIP4Win which can accomodate
such a map.

I hope you got the answer. Otherwise I'll repost here.

Andrea T.

On Sun, 25 Feb 2007 14:12:25 +0200, Anthony Ayiomamitis
wrote:

Thanks for the information and feedback Chris! I wish I had kept the raw
files since the values there were less than the master file. Of course,
I would have more confidence on the master file which is based on the
average of the five raws.

Oh yeah, one more thing ... due to my excitement, I was a little too
quick and forgot to apply the image scale to the FWHM I was getting
immediately after each download. Oops!

FYI, with your camera and scope you can calculate the smallest possible
FWHM possible. Assuming your source varies from 400nm to 650nm, that
your optics are perfect, and that there is perfect seeing, you'd get a
FWHM of 0.73 pixels to 0.85 pixels (the former being the actual value
for the central diffraction peak, and the latter the value if that peak
is fit to a Gaussian, which the better FWHM calculators do).

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

I suspect that gravitational interaction isn't sufficient between two
otherwise unrelated stars in an open cluster for this mechanism to work,
but I haven't thought about it that much.

And I suspect that even if several stars were close enough for such an
interaction, the result would be stars flung in all directions, not
dragged into a line.

Maybe they've been lined up for a trick shot, like in snooker ?

--
Boo

Chris L Peterson wrote:
On Sun, 25 Feb 2007 14:12:25 +0200, Anthony Ayiomamitis
wrote:


Thanks for the information and feedback Chris! I wish I had kept the raw
files since the values there were less than the master file. Of course,
I would have more confidence on the master file which is based on the
average of the five raws.

Oh yeah, one more thing ... due to my excitement, I was a little too
quick and forgot to apply the image scale to the FWHM I was getting
immediately after each download. Oops!


FYI, with your camera and scope you can calculate the smallest possible
FWHM possible. Assuming your source varies from 400nm to 650nm, that
your optics are perfect, and that there is perfect seeing, you'd get a
FWHM of 0.73 pixels to 0.85 pixels (the former being the actual value
for the central diffraction peak, and the latter the value if that peak
is fit to a Gaussian, which the better FWHM calculators do).

Cool and thanks for the baseline numbers! Can I get the math behind
these numbers for my education?

Anthony.

wrote:

On 25 Feb, 12:15, Anthony Ayiomamitis
wrote:

wrote:

On 25 Feb, 01:54, Anthony Ayiomamitis
wrote:

Anthony Ayiomamitis wrote:

Chris L Peterson wrote:

On Sun, 25 Feb 2007 01:43:12 +0200, Anthony Ayiomamitis
wrote:


snip


I hope you got the answer. Otherwise I'll repost here.

Well, I checked my master flat (reduced) and there are no cold/hot
pixels. When looking at the histogram, the lowest ADU is 25788, next
lowest is 26765, next lowest is 28851 and so on ... of course, the HUGE
majority (99.86%) of the pixels are between 31751 and 34065 (ie. only 23
pixels are below 31751 but at or above 25788 ADU). Nothing above 34065 ADU.

Thanks.

Anthony.


Andrea T.

On Mon, 26 Feb 2007 00:48:18 +0200, Anthony Ayiomamitis
wrote:

Cool and thanks for the baseline numbers! Can I get the math behind
these numbers for my education?

In a perfect, unobstructed system with no seeing effects, a star forms a
simple diffraction pattern described by a Bessel function. Since that
function is non-monotonic, "FWHM" isn't really defined, but I took it
simply as the FWHM of the central peak. I plotted the function an
determined empirically that the FWHM is about 45% of the full diameter
to the first minimum.

The diameter of the central peak is 2.44 * lambda * f. For your scope, f
is 7.5, so the disk is 7.3um across at 400nm, and 12um across at 650nm.
BTW, this is why you don't normally see diffraction patterns in images-
the patterns for different wavelengths blur together. I took 12um at the
disk size; 45% of this is 5.4um, which is 0.72 pixels with your 7.5um
pixel camera.

If you try to fit the central peak and first ring to a Gaussian, the
resulting function is slightly broader. I did this empirically in
Mathematica, giving a FWHM of 6.4um, or 0.85 pixels.

In reality, the widths are probably a little wider. I simplified the
problem by treating the central peak as if it were produced by a
monochromatic 650nm source. Because the source is broadband, the peak
isn't described exactly by the function I used, but is a convolution of
many functions. However, I think the numbers are close enough to give an
idea of what "perfect" means in your system.

It's worth noting, too, that there is no hard definition of FWHM, which
is why you can't compare numbers produced by different programs.
Algorithms may use fitting or not, and they handle the noise floor and
background differently. Given a purely Gaussian PSF (which is itself a
simplification), which does have a well defined FWHM, I've measured the
FWHM calculation accuracy in the following programs, ordered best to
worst: psfmeasure (IRAF), Maxim (v4), CCDSoft (v5), AIP (v1 and v2).

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Chris L Peterson wrote:

On Mon, 26 Feb 2007 00:48:18 +0200, Anthony Ayiomamitis
wrote:


Cool and thanks for the baseline numbers! Can I get the math behind
these numbers for my education?


In a perfect, unobstructed system with no seeing effects, a star forms a
simple diffraction pattern described by a Bessel function. Since that
function is non-monotonic, "FWHM" isn't really defined, but I took it
simply as the FWHM of the central peak. I plotted the function an
determined empirically that the FWHM is about 45% of the full diameter
to the first minimum.

The diameter of the central peak is 2.44 * lambda * f. For your scope, f
is 7.5, so the disk is 7.3um across at 400nm, and 12um across at 650nm.
BTW, this is why you don't normally see diffraction patterns in images-
the patterns for different wavelengths blur together. I took 12um at the
disk size; 45% of this is 5.4um, which is 0.72 pixels with your 7.5um
pixel camera.

If you try to fit the central peak and first ring to a Gaussian, the
resulting function is slightly broader. I did this empirically in
Mathematica, giving a FWHM of 6.4um, or 0.85 pixels.

In reality, the widths are probably a little wider. I simplified the
problem by treating the central peak as if it were produced by a
monochromatic 650nm source. Because the source is broadband, the peak
isn't described exactly by the function I used, but is a convolution of
many functions. However, I think the numbers are close enough to give an
idea of what "perfect" means in your system.

It's worth noting, too, that there is no hard definition of FWHM, which
is why you can't compare numbers produced by different programs.
Algorithms may use fitting or not, and they handle the noise floor and
background differently. Given a purely Gaussian PSF (which is itself a
simplification), which does have a well defined FWHM, I've measured the
FWHM calculation accuracy in the following programs, ordered best to
worst: psfmeasure (IRAF), Maxim (v4), CCDSoft (v5), AIP (v1 and v2).


Thanks for the illuminating (no pun intended) discussion and insights!

Anthony.