How to Get Less Noise

[JohnSadlerAstro]

Hi,

I have a Canon EOS 1000d, which I modded back at the start of the year. I usually capture subs of about 3-5mins duration at ISO 800. However, I'm wanting to image the Deer Lick group with Stephan's Quintet, and I'm going to need to go right the way to 5 min subs for them. 

I've noticed quite high noise levels at ISO 800, is this the optimal speed to use? Would it be better to capture subs at ISO 400, or would they then be really underexposed? Would I be better doing 3 min subs at ISO 1600?

I find ISO a little bit confusing, as far as I know it's a measure of how the camera's computer assigns brightness from the raw pixel values? ? 

Thanks for looking!  

John

[ollypenrice]

Have you considered a cooler box?

Olly

[Craney]

There was a DSLR cooler box for sale on this very forum not so long ago...

 

 

[Dragon_Astro]

The higher the ISO, the more noise/grainy look you'll get. ISO 400 was the sweet spot when I used my 450D

[JohnSadlerAstro]

Hi,

Yes I have considered the cooler box, noticed it for sale the other day, I'm just concerned that it wont make much difference. Reading though the test results it seemed like the results were very subtle.  

If I were to start taking my subs at ISO 400, would I have to switch to 10 min exposures? Or could I still continue at 5 min max?

According to this website: http://dslr-astrophotography.com/iso-values-canon-cameras/   , ISO 200 is the best setting. Surely though, that would mean I'd need exposures of about 40 mins long to be the equivalent of a 5 min sub? ?

John

[Dragon_Astro]

4 minutes ago, JohnSadlerAstro said:

Hi,

Yes I have considered the cooler box, noticed it for sale the other day, I'm just concerned that it wont make much difference. Reading though the test results it seemed like the results were very subtle.  

If I were to start taking my subs at ISO 400, would I have to switch to 10 min exposures? Or could I still continue at 5 min max?

John

I think you'd still be ok with 5mins, but there's no harm in trying 10....see how the noise is? The longer a dslr is exposing, the warmer it gets though. You might find 5mins works best, but will have to take a few more subs to compensate.

[MarsG76]

I expose at ISO800 and 1600 using my 40D, which is older than your camera, and find that the noise levels are acceptable.... although you have a number of things to consider, one is the number of subs you stack, the more the better, and the lower your noise will be,,,,

The other thing is that you should preprocess all of your subs before stacking... you can use dark frames, flats and Bias frames but I find that I do not need to use darks when I convert CR2 to TIFF before stacking using the Photoshop Camera RAW filter as it removes a lot of the the noise in each sub... of course I'm presuming that you're exposing in raw format.

 

I expose subs for 30 minutes per sub with this method (when imaging through narrowband filters) and the noise is manageable.

 

[Dragon_Astro]

Personally I wouldn't bother with the dslr cooler, as you've seen for yourself the temp drop wasn't that much really. You could manage the noise yourself adjusting the ISO.

[newbie alert]

You could do a cold finger mod?

Only way to reduce noise is to cool that sensor down..mid winter it's not too bad..mid summer almost not worth it...suprised you can image for 5 mins at iso 800..

[Scooot]

1 hour ago, JohnSadlerAstro said:

If I were to start taking my subs at ISO 400, would I have to switch to 10 min exposures? Or could I still continue at 5 min max?

John

Hi John,

I think I would try changing to ISO 400, leave the exposure at 5 minutes but take twice as many so your integration time is the same. You'll improve your SNR and dynamic range. Taking more subs will increase the signal of the the very faint stuff enough so that the chip registers it. So even though the exposure is halved I don't think you'd lose as much as you might think.

[coatesg]

You only need to expose long enough to ensure the background noise level (shot noise) is significantly greater than the read and thermal noise. 5 min should be enough unless you are under very dark skies - 2-3 min is probably enough in average skies. It's the total exposure that's important once background noise >> read noise, rather than the length of the subs. 

At lower ISO (ISO = gain, not sensitivity), you should get less fixed pattern noise and a wider dynamic range (so stars won't saturate as quickly) - but you choose the lowest ISO such that the upstream noise swamps downstream noise in the camera (see: http://dslr-astrophotography.com/iso-dslr-astrophotography/).

[alacant]

Hi

I find that BY FAR the best noise reduction on my 700d is obtained by dithering between exposures. Here's my Stephan with 6 minute exposures and the sensor well above 30º throughout. Without the dither, I don't think it would get anywhere near. Lose any dark frames you may have. Use bias and flat frames along with your light frames only.

BTW, you're gonna need a long focal length. Even at 1200mm, it's small... You've set yourself a good challenge;)

HTH, good luck and do post your results.

[Stub Mandrel]

Consider a cold finger. I made one for my 450D and the results are remarkable. Stretched darks show virtually no noise whatsoever.

This is a master dark for 300s converted to 16-bit PNG. It's only 73kb because there's virtually no variation in it to compress. Zoom in and you will find a handful of hot pixels! You could ask if I need to use darks at all:

[michaelmorris]

5 minute subs at ISO 800 with a Canon 1000D is perfectly doable.  Take at least 20 subs plus darks , flats and bias and combine them in Deepskystacker.  Dithering between exposures also helps a lot with removing fixed pattern noise. 

(I'm not going to comment on the performance of DSLR coolers in this thread as it might constitute selling on the forum.)

 

[JohnSadlerAstro]

Hi,

Would something like this work for the cold finger mod? Would a mod be possible on a £25 budget? Also, what is the risk factor? If I break the camera at this point it could be a bit awkward--but if its similar to the IR filter mod im ok with that.  

https://www.amazon.co.uk/KKmoon-Thermoelectric-Refrigeration-Conduction-TEC1-12706/dp/B074DVY4D5/ref=sr_1_2_sspa?ie=UTF8&qid=1536664938&sr=8-2-spons&keywords=peltier+module&psc=1

https://www.ebay.co.uk/itm/Genuine-COPPER-Strip-0-7-mm-Sheet-for-Model-Makers-Hobbyist-Guillotine-Cut/291690141167?var=590718503629&hash=item43ea1655ef:m:mc83svYNvhenbYew98C5WyQ

John

[Stub Mandrel]

I used the TEC out of an old mini-fridge. I found a rather better heatsink. There are some good guides online, one has a template for the copper.

I think my copper was probably 0.7mm.

Nothing was riskier than the astro mod, in fact less chance of a sensor breakage.

Trickiest bit was making a loop of thin resistance wire around the sensor cover plate, and experimenting to get the right resistance to control the current in the wire to the lowest the keeps the sensor free of ice.

I used a 3D printed cover with foam in it to insulate the back of the finger and foam between the cooler and camera body - you need to make sure you aren't cooling outside the camera more than you need to.

The sensor in the camera shows a minimal temperature drop, but clearly the sensor is way below zero as without a heater wire it ices up in a few minutes.

With apologies to Michael, I think they are much more effective than cool boxes for the obvious reason that the suck heat directly out of the sensor and its associated circuitry. The best bit about the cold finger is you can easily see the screen and access all the controls as normal, although the USB and remote connectors can only be removed by partially removing the cooler.

I now have a plate fitted beneath the camera that zip ties all the wires in place as serious strain relief.

With original heater resistor:

This is a really good heatsink, I think it's out of an old PC. The fan blows the warm air away from the camera:

Hot melt to protect from condensation/insulate:

The copper plate:

Failed cool box, foamex is just too thin to work for insulation:

 

[michaelmorris]

4 hours ago, Stub Mandrel said:

With apologies to Michael, I think they are much more effective than cool boxes for the obvious reason that the suck heat directly out of the sensor and its associated circuitry. The best bit about the cold finger is you can easily see the screen and access all the controls as normal, although the USB and remote connectors can only be removed by partially removing the cooler.

No apologies needed, a cold finger direct onto the back of the chip is always going to be more effective than a cool box.  The advantage of the cool box is that you don't have to dismember the camera first, so much better for techo-numpties like me!

[JohnSadlerAstro]

3 hours ago, Stub Mandrel said:

I used the TEC out of an old mini-fridge. I found a rather better heatsink. There are some good guides online, one has a template for the copper.

Hi,

How easy/expensive is it to get hold of something like that, do you know?

John

[ollypenrice]

7 hours ago, alacant said:

 

BTW, you're gonna need a long focal length. Even at 1200mm, it's small... You've set yourself a good challenge;)

 

Some care is needed here. With this camera a 1 metre focal length is going to be working at 1.18 arcseconds per pixel. What resolution will your guiding and your seeing support? If it won't support anything less than this then adding focal length will just add 'empty resolution,' meaning more size without more detail. I've imaged this target with a long FL (2.4 metres) and a moderate (1 metre) in different cameras, the long giving 0.63"PP and the moderate 0.9"PP. There is no difference in real detail captured. The seeing imposed the limit. This was at 1 metre with 4.54 micron pixels.

Olly

[Stub Mandrel]

19 minutes ago, JohnSadlerAstro said:

Hi,

How easy/expensive is it to get hold of something like that, do you know?

John

I just rescued one from the trash.

A 40-W tec/fan/heatsink combo off ebay or amazon is fine, but choose one with a decent heatsink.

Ideally, but a separate 40W TEC and a hefty CPU cooling heatsink and fan off ebay, the total cost will probably be less.

 

[JohnSadlerAstro]

47 minutes ago, ollypenrice said:

Some care is needed here. With this camera a 1 metre focal length is going to be working at 1.18 arcseconds per pixel. What resolution will your guiding and your seeing support? If it won't support anything less than this then adding focal length will just add 'empty resolution,' meaning more size without more detail. I've imaged this target with a long FL (2.4 metres) and a moderate (1 metre) in different cameras, the long giving 0.63"PP and the moderate 0.9"PP. There is no difference in real detail captured. The seeing imposed the limit. This was at 1 metre with 4.54 micron pixels. 

7 hours ago, alacant said:

BTW, you're gonna need a long focal length. Even at 1200mm, it's small... You've set yourself a good challenge;)

HTH, good luck and do post your results.

Hi,

I'm not expecting anything like your image, more of a widefield, though. I believe my setup works with 5.75 micron pixels, making it about 1.8"/px. As NGC7331 (The main target) is about 8' across, I should in theory get about 240 pixels covering it from one end to the other. Imo that should be enough for a little bit of detail. I did capture 8 poor weather subs as a concept test which had some big collimation/coma issues (which I think are now resolved) and they were reasonably promising. If I get a few hours I'm reasonably hopeful it will turn out pleasant. My main worry was the noise, it just seemed so awful, even for only 24 mins. ? 

John

 

[michael8554]

Hi John

Unfortunately the 1000D is one of the noisiest Canon dslr's, so I would sell it, buy a 450D body about £90, and mod it yourself now you have the experience ?

600secs at ISO 800, or up to 340secs at 1600.

Dither, Flats, use Bias as Darks in DSS.

Michael 

[JohnSadlerAstro]

I'd love to do that, but idk if anyone would actually buy a rattling old 1000d with the IR filter ripped out! ?

EDIT: Although there is the old 55-200mm telefocal lens I never use now...

John

 

 

[Stub Mandrel]

22 minutes ago, JohnSadlerAstro said:

I'd love to do that, but idk if anyone would actually buy a rattling old 1000d with the IR filter ripped out! ?

John

 

 

I did ?????

[ollypenrice]

Awe, by day I shoot with my old 1000D. I think I might cry...

lly