56 hour sub?

[Demonperformer]

Honestly, the things I start to think about when the weather is bad and I can't get outside ...

Assumptions:
I live within the arctic circle, so I have part of the year when the sun never rises.
Weather conditions do not interfere with imaging during the time (!!).
My telescope and ccd camera will go to mag 13.0 on a 20 sec sub, and my OAG guiding is excellent.
I want to image stars down to mag 23.0 at a point close enough to the pole that it does not set.
The electrics and mechanics of my system are able to handle the sort of temperature I am likely to have.

The simple math would suggest that a 55.6 hour sub would gather sufficient photons to reveal stars at mag 23 (10000x the duration giving an additional 10 magnitudes), but given the above assumptions, would I actually be able to get a sub of this length in practice?

And what would be the first problem that would be likely to limit the ultimate length of sub I could obtain?

Or indeed, is there an ultimate limit to the length of a sub that could be taken in this situation?

[ronin]

Not many places in Yukon, Alaska or NT to try it out that are far enough North, best looks like Norway. Try Alta and Hammerfest, Hammerfest being the most northerly. Better is Svalbard, Spitzbergan - it is further North still.

Check the hurtigruten ferries for a timetable, no idea about accomodation, may get shots of the aurora as a bonus.

Expect a report from you in about 2 weeks, maybe 3. :grin: . Oh yes and some good 56 hours images. :rolleyes: Don't dome back without the subs, :eek: .

P.S. Hope the weather improves so you can stop thinking. :p

[Naemeth]

Cooling is not likely to be an issue, it's polar alignment that seems the biggest issue and making sure guiding is spot on. I think you probably would be able to do it, but that you might end up with very over exposed bright stars.

[michael.h.f.wilkinson]

You could in principle do this, but the dark current may become a problem, as will saturation effects in bright areas. If the readout noise can be kept in check, I wonder if a long series of shorter subs would not be better in practice. Let us assume a dark current of in the order of 1 electron at your normal exposure time (this is pretty good). At 10,000 times the exposure time you would be looking at 10,000 electrons on average. The standard deviation of the dark current would be the square root of that number, i.e. 100 electrons. The average I can subtract out using darks, but the random part (the standard deviation) I cannot. This random part of 100 electron std.dev. is much higher than typical readout noise (just a few electrons). This means that if I take many short exposures and add them up, I would end up with practically the same S/N but a much higher potential dynamic range, because I avoid saturation effects.

Besides, 56 consecutive clear hours might not be readily available that far north.

[slickfonzy]

Would you also need a setup that would allow you to track continuously without needing to flip at the meridian?

[michael.h.f.wilkinson]

Would you also need a setup that would allow you to track continuously without needing to flip at the meridian?

Actually, at the pole, an Alt-az mount tracks perfectly. Slightly away from the pole, just put a fork mount at a slight angle and away you go.

[opticalpath]

Hmm, maybe 56 hours would be pushing it, even up there   

Detection of very faint objects is ultimately to do not just with total photons captured from the object, but with signal to noise ratio. Noise from all sources results in uncertainty and the question is: can you distinguish the object signal from the noise signal with certainty? (S/N ratio is a measure of that uncertainty.) In any long integration you have to add up all the noise contributions as well as the photons you expect from the target object. In very long integrations, sky background eventually becomes a major noise factor, even in very dark skies. Then there's critical sampling, diffusion caused by seeing/ tracking .... etc.

There's an excellent explanation of this stuff in a paper by Herbert Raab:
http://www.astrometrica.at/Papers/PointSources.pdf

As long as your subs are long enough to lift the background brightness clear of the noise level, it's generally better not to go much longer but to take multiple shorter subs. Way back in 2004 I was chasing very faint (mag 19-20) Kuiper Belt Objects with a lowly raw-modified 8-bit webcam. The faintest I ever reached in good seeing was Vmag 20.5, and that was with many exposures of only 20 seconds, in quite 'bright' suburban skies.

Adrian

[Demonperformer]

Thanks for the replies.

So, noise would still eventually become an issue.  And I certainly had not considered the 'meridian flip' problem.

Adrian: Mag 20.5 with 20 second subs - under suburban skies?? I am definitely going to have to give that a go.  Will have to give that paper a thorough read!

Thanks.

[opticalpath]

Thanks for the replies.

So, noise would still eventually become an issue.  And I certainly had not considered the 'meridian flip' problem.

Adrian: Mag 20.5 with 20 second subs - under suburban skies?? I am definitely going to have to give that a go.  Will have to give that paper a thorough read!

Thanks.

With only 8 bits, these cameras saturated quickly so exposures had to be short - no choice!   Steady seeing and good tracking was critical for extremely faint stars/ dwarf planets though .... you need those precious few photons to fall on just a few pixels and not be spread around too thin.  Good fun to see how low you can get!

Adrian

[ollypenrice]

There would be the tiresome task of taking 15 fifty-six hour darks as well... 

Olly

[JohnC64]

There would be the tiresome task of taking 15 fifty-six hour darks as well... 

Olly

:grin: :grin:

With the comments on the image of - "Cracking Lum  - are you thinking of getting RGB with this??" and "Its a bit noisy, Needs more subs"

Very interesting, thought provoking original post though

John

[FraserClarke]

Or indeed, is there an ultimate limit to the length of a sub that could be taken in this situation?

The Aurora...

As others have said, you'll get very little benefit from using longer exposures once you reach the regime that background flux (be that dark current or sky flux) dominates over the read-noise.

That far north, they sky will actually be quite bright because of the auroral emission, so you'd saturate long before 56 hours.

Even at a good dark site with no moon, the sky has a brightness of about 22 mag per square arcsecond. So assuming you have a standard kind of set-up with 2" pixels, the sky itself is equivalent to having a 20.5 mag star in every single pixel... Once you start trying to image things fainter than the sky (which is totally possible), your exposure times start going out a lot quicker than the simple photons-from-the-target calculation...

Oh, and you'd have done 2.5 rotations of the mount and ripped all the cables out

[Gina]

What a fascinating thread

[dph1nm]

(10000x the duration giving an additional 10 magnitudes)

Err , no - it goes as the square root of the exposure time, so you only get 5 mags improvement in depth.

NigelM

[ollypenrice]

Or even, 'I really want to get this one done before the sun goes red giant...'

Olly

[jackrussell0232]

I think it would get a bit tiring - going back to check the guiding every 5 minutes, 670-odd times!

Or is that just me?

[Demonperformer]

 There's an excellent explanation of this stuff in a paper by Herbert Raab:

http://www.astrometrica.at/Papers/PointSources.pdf

I have now had the time to devote a couple of hours to digesting this paper and playing with some figures.

The first obvious thing to say is that there are a couple of figures that I have no idea of how they relate to my situation, so I have had to use the figures quoted in the paper - they may be way off. These are FWHM (which is quoted as 4"), the amount of signal photons on the peak pixel (which is quoted as 29%) and the sky background (quoted as 18).

All that having been said, when I plug my own setup figures (see below) into the spreadsheet, I get a 20 mag star giving me a peak SNR of 4.7 on a 100 second exposure. And restricting myself to a minimum PSNR of 4.5, I could get down to mag 19 with a 20 second exposure. If I accept the PSNR figure of ~3 in the paper, this would push 20 sec exposures to mag 19.4.

Allow 2 magnitudes worse sky brightness (16) and half the peak pixel percentage (15%) [changing FWHM does not seem to alter the result at all], and this still gives me a limiting magnitude of 17.8 (PSNR 4.5) or stretching to 18.2 (PSNR 3.1) on a 20 second exposure.

I really had no idea I could achieve so much with my gear from my back garden!

The figures I am using for my equipment are:

scope {8SE with 0.5 focal reducer}:

diameter: 0.2 m

f-ratio: 5

central obstruction: 31%

so scope area: 0.021 sq m

camera {SXV-H9}:

camera pixel size: 6.4 um

angular pixel size: 1.32" [from CCDcalc]

pixel area: 1.7424 sq"

dark current: 0.1 e/s/px [from SX website]

readout noise: 7 e/px [from SX website]

QE: 65% [from SX website]

Thanks.

[Steve Ward]

Another small problem ...  

Astronomical twilight finishes when the Sun dips 18 degrees below the horizon and night begins , at the Poles the Sun never gets lower than 23 degrees below the horizon so even the days of " total darkness " never get truly dark ....