What are the units for FWHM?

[frugal]

I have been reading around focusing and matching pixel size to focal length, and one of the statements that I keep reading is that for most people the seeing will limit FWHM to about 3.5. However the units are never given. Is this because it is a unitless value, or is it assumed that the reader will know the units?

I had thought that it was arc seconds, but I use Backyard EOS which does not know the focal length I am using so it must be working in pixels.

As an example. I can generally get a FWHM score from Backyard EOS of about 3.1. With a 300mm camera lens I am sampling at about 4.5 arc seconds per pixel. So is my actual FWHM 14? Which seams a bit rubbish

Sent from my GT-N7000 using Tapatalk

[sologuitarist61]

Sorry, been into photography all my life and never come across the term, perhaps if someone comes up with the actual words that the letter stand for I might get some idea of what you are actually talking about!

Have looked on the web and Wikipaedia comes up with this http://en.wikipedia.org/wiki/Full_width_at_half_maximum but not being a mathematician I have no idea what it all means.    

[Hitesh]

Fwhm is full width half maximum. I think they are arbitrary units not too sure but I know someone more knowledgable will come along in a bit and explain it all properly

[ronin]

FWHM will be a unit of length (Width is a length), I assume it refers to pixel size so the most common one used is micron, μm.

[Stephen]

In astro software terms I believe it's unitless - though it's a count of pixels if that helps any

[frugal]

If it is unitless then how can people make blanket statements like "you can never get better than 3.5 due to seeing"? In that case all I need to do is use a camera with smaller pixels which does not seem right.

Sent from my GT-N7000 using Tapatalk

[Physicist13]

FWHM is often used in Physics. Its units are the same as the units you are plotting against (ie the x axis units).

For astro-photog the usual is in arcsec but equally it could be in pixels, purely depends on what the user/software deem to be the x asis units. They should specify.

P

[Stargazer33]

Does this explanation help?

http://www.noao.edu/wiyn/images/fwhm.html

[IanL]

Ideally the author or the software should spell this out.  If not you can usually deduce it from the context:

- If your software doesn't ask for any additional information, then you can reasonably assume the FWHM is measured in pixels as there is no way for the software to calculate any other unit.  So really all you're aiming for is the smallest number you can manage for a given star to achieve best focus for example.

- If you have the opportunity to provide the pixel size of the camera (in μm) or the focal length of the scope/lens (typically in mm but could be in some other unit of length), then the FWHM will usually be in arcseconds. The software can use these two parameters to convert from pixels to arcseconds as follows:

FWHM_arcsec = FWHM_pixels x ( pixel_size_μm / focal_length_mm) x 206.3

FWHM in arcseconds is more useful since you can directly compare images taken with different cameras and scopes/lenses.

[frugal]

Cheers,

So with a FWHM of 3.1 and a arc seconds per pixel of about 4.5 (from your excellent Imaging Toolbox Ian) I have a FWHM of about 14 arc seconds. That would certainly explain why my images do not look particularly crisp and sharp. Another reason why I need to upgrade to a better lens or a proper telescope

Sent from my GT-N7000 using Tapatalk

[RikM]

Cheers,

So with a FWHM of 3.1 and a arc seconds per pixel of about 4.5 (from your excellent Imaging Toolbox Ian) I have a FWHM of about 14 arc seconds. That would certainly explain why my images do not look particularly crisp and sharp. Another reason why I need to upgrade to a better lens or a proper telescope

Sent from my GT-N7000 using Tapatalk

Take a shorter exposure and check it again

I use FWHM both for focussing and as a check that focus isn't drifting or that the seeing isn't getting worse during the run. If you focus using 1 sec exposures you will have much lower FWHM values than for 10min subs. I normally use 6 sec exposures for focus and get readings of 2-3 and then for 10min subs readings of 4-7. I think this is in pixels but Maxim can display either pixels or arcsec depending on which box is ticked, so I'm not 100% sure. My images normally look a little soft.

[frugal]

I can get a FWHM of 3.1 even with live view with my 300mm lens

Sent from my GT-N7000 using Tapatalk

[ollypenrice]

I think (but don't guarantee!!) that what really matters is that your tracking should be good enough to keep errors at the imaging camera below a pixel. If you were using an OAG and a guide camera with the same size pixels as the imaging camera then this would be easy to see at a glance, because sub-pixel on the guide trace would mean sub-pixel on the image.  When this isn't the case you could work out the resolution of both the guide rig and the imaging rig in arcseconds per pixel and compare them. If you are on, say, 4 arcsecs per pixel at the guider and 2 arcsecs per pixel on the imager then your guide trace would need to show no more than half a pixel of drift to keep the imager within tolerance.

If this simple view of things is erroneous then I won't cry if somebody tells me so! As a practitoner I have to settle for as small an FWHM as possible and as good a guide trace as possible. If the seeing or guiding are giving bad values the trick would be to shoot colour rather than luminance that night. OSC users might capture bad data on one night and better on the next, then use layers with a bit of cunning. Make a full stack and a stack of only the good data. Stretch both, put the good stuff on top, select the faint low detail parts of the image and erase them from the top layer to get the deeper, low detail, low noise parts from the deeper set.

Olly

[lensman57]

I have been reading around focusing and matching pixel size to focal length, and one of the statements that I keep reading is that for most people the seeing will limit FWHM to about 3.5. However the units are never given. Is this because it is a unitless value, or is it assumed that the reader will know the units?

I had thought that it was arc seconds, but I use Backyard EOS which does not know the focal length I am using so it must be working in pixels.

As an example. I can generally get a FWHM score from Backyard EOS of about 3.1. With a 300mm camera lens I am sampling at about 4.5 arc seconds per pixel. So is my actual FWHM 14? Which seams a bit rubbish

Sent from my GT-N7000 using Tapatalk

The absolute measurement unit is in pixels but  in AP arcsec is a more appropriate unit. On a night with excellent seeing and transparency  you get smaller FWHM values for a target with a specific set up than you'd get on a bad night.

A.G

[ollypenrice]

What I don't follow is that, if it is in pixels, why do I get readings like 0.85 and 0.87 or 0.99 as I did last night? (Fabulous seeing but mediochre transparency, a natural enough experience.) How can it divide the indivisible pixel?? Sorry to be thick!

Olly

Edit. And then it dawned on the silly old beggar. The widest part of the star's bell curve is much more than a pixel and the top can be calculated as a notional point and half way between comes out as 0.88 or whatever. I will leave my rambling thought processes posted for your amusement! 

[lensman57]

What I don't follow is that, if it is in pixels, why do I get readings like 0.85 and 0.87 or 0.99 as I did last night? (Fabulous seeing but mediochre transparency, a natural enough experience.) How can it divide the indivisible pixel?? Sorry to be thick!

Olly

Edit. And then it dawned on the silly old beggar. The widest part of the star's bell curve is much more than a pixel and the top can be calculated as a notional point and half way between comes out as 0.88 or whatever. I will leave my rambling thought processes posted for your amusement! 

This is an interesting read, sort of turns everything on its head.http://www.astro-imaging.com/Tutorial/MatchingCCD.html

Regards,

A.G

[Scorpius]

This is an interesting read, sort of turns everything on its head.http://www.astro-imaging.com/Tutorial/MatchingCCD.html

 

Regards,

A.G

Very interesting. Haven't done the math but starting to wonder if the 8.6 micron pixels in my G3 mono CCD coupled with my ES 480 FL scope will lead to under sampling. Oh well, since I can't afford a new camera at this point, guess I'll go with what I got and hope under sampling doesn't become an issue. Thanks for the link!

Regards,

Scorpius

[drjamesr]

Last resort - refer to instructions!

Page 29 of the BY EOS user manual gives a comprehensive explanation of the calculation, not that I fully understand it. If anyone else does please post. The figure is a pixel measure calculated in terms of the fraction of saturated pixels at HW - so not sure arc seconds can be derived from it in a simple way at least. See

http://www.jtwastronomy.com/products/guides/backyardguide.pdf

[Merlin66]

Rob's paper quoted above is an excellent summary of the FWHM v's resolution V's CCD pixel.

It also applies to spectroscopy - and the resolving "power" of the spectroscope.

[KyleStoke]

This is an interesting read, sort of turns everything on its head.http://www.astro-imaging.com/Tutorial/MatchingCCD.html

Regards,

A.G

This is interesting, working out the seeing rating seems to be the hard bit