Read noise, light pollution and a DSLR (Canon60d)

[MalcolmP]

Re-posted in the right ? forum !

Questions :-
(1) How can I measure my read noise? *
(2) Given a lp level **(of eg. 20mag/sqarcsec) how long does my sub exposure need to be for my rn to be 5x *** lower than the lp

The rest is by way of chat and can be discarded !
Discussion! :-
* I have googled and found generic specs in the region of 3e per somethingorother at iso800, but how typical is that of 'most' 60d ?

** lp is often given in terms of (not very useful bortle units or) SQM units of mag/arcsec^2 So how do we convert that to electrons in the sensor.
I can measure lp SQM using ASTAP

***5x is often quoted to reduce read noise to insignificance when evaluating total integration times

Why? :-
Up until now I have been entertaining myself chasing supernovae and other dim objects like asteroids and variable stars using the 60d + 50 or 135mm lens fixed on a tripod. Several hundred subs have been interesting  but now I am wondering if I should progress to an alt-az goto for 10sec subs or to a polar aligned device for even longer subs?
Is 10sec going to be long enough to suppress read noise @ typically 20m/as^2 , maybe I am already suppressing read noise with 100 or 1000subs of 5sec or 2sec ?

I have of course googled, but most of it rapidly goes over my head or is not relevant to a specific dslr
Sorry for the long post, I blame too much thinking time on my hands waiting on the weather thanks for reading ! 
 

[symmetal]

The easiest way to get actual camera sensor info is to use Sharpcap's Sensor Analysis, (the free version will provide this) which gives read noise and a host of other info, and also gives optimum exposures based on read noise swamping. The 60D can be set to provide raw image data so should work ok for sensor analysis.

As to why 5x, it's because at around that level the read noise contribution to your image is effectively made insignificant compared to the noise you're getting from the image's sky background. For a detailed discussion you can read this thread if you have a spare weekend. 😁 Measuring the sky background ADU values from one of your images lets you calculate the sky background noise contribution (after removing the average camera bias value) and the rest follows as detailed in the thread topic. It's easier though to let Sharpcap do the hard work. 🙂

10 sec subs I'm sure won't be long enough to swamp the read noise unless you have very light polluted skies, like central London, and certainly not at SQM 20, so to counteract this it would mean having to have a longer total integration time, in order to match the results of the total integration time of read noise swamped exposures, as far as image noise is concerned.

An EQ mount would be much more useful to use for fainter objects like DSOs, as longer exposures can be made without dealing with the image rotation problem you get with an Alt/Az mount, though depending on the EQ mount (ie. how much money you want to spend), and your lens focal length you will likely need a guide scope for exposures more than around 30s.

Supernovae and variable stars can likely be captured with short exposures on an Alt/Az mount but you're limiting the type of objects you can image. You'll likely want to expand your options over time, so an EQ mount is a better choice in my opinion.

Alan

[MalcolmP]

4 hours ago, symmetal said:

The easiest way to get actual camera sensor info is to use Sharpcap's Sensor Analysis,

this thread if you have a spare weekend.

Thanks Alan, I thought I was probably not near swamping the read noise with my skies,  until Vlaiv mentioned millisecond exposures in a recent topic
So I began to wonder since I had no feel for it.
20sqm is the best I have seen (in my rural back garden, nearest street light a few miles away) since I have had the capability of measuring it with ASTAP,, but not the best skies I have seen historically by MK1 eyeball.

Sadly Sharpcap will not run on my old window7 machine it objects that I do not have a required instruction set in my cpu !

Lots of rainy day reading to be had in that link but I fear it has lost me already in the second post - I know not how to measure my median ADU though I understand it is common practice amongst real APers with dedicated cameras  
 

[symmetal]

Pity you can't run Sharpcap. if somebody reading this topic has a Canon 60D and can post the data they obtained from Sharpcap it would help a lot.

What capture program are you using to get your images, as the dedicated astro ones  provide an image analysis which gives the mean, median, max and min pixel values for either the whole image or a smaller area like a 21x21 pixel area. Unless you have a lot of nebulosity in your images the median ADU will be the ADU of the background dark sky. This is the figure you need for doing read noise swamping calculations. You can usually get a direct readout displayed of the xy coord and pixel ADU value by moving your mouse over the image dark sky background. The 60D is a 14 bit camera I believe though the ADU displayed will usually be 16 bit so will vary between 0 for black to 65535 for white.

As well as the read noise you need to know the camera gain in electrons/ADU at various ISO settings to determine optimum exposures. Sharpcap would have given this information so hopefully someone can provide the data.

Alan 

[MalcolmP]

9 hours ago, symmetal said:

 has a Canon 60D and can post the data they obtained from Sharpcap it would help a lot.

What capture program are you using

mean, median, max and min pixel values for either the whole image or a smaller area like a 21x21 pixel area.

Or even something similar to a 60d would be a ball-park for starters,
Just an intervalometer (camera on a pan/tilt fixed on a tripod) followed by DeepSkyStacker and Gimp.
Turns out Gimp can do something like that if I import the raw CR2 in 16bit, else it defaults to 8bit, thanks for the prompt.

 

Edited February 23 by MalcolmP

[ollypenrice]

It doesn't take long to run sets of exposures of various lengths in order to find out what works best for you. It is also, let's remember, a necessary part of the scientific method....

Olly

[symmetal]

It's worth taking a look at capture programs like APT or NINA. APT is more geared towards DSLR cameras, with a free trial period but is very reasonably priced. It's probably easier to work with for somebody new to using capture programs. NINA is free. Both will give a lot of image information that will be useful to you. NINA will give optimum exposures info but I believe it needs the sensor data from Sharpcap to work them out.

It may be worth starting another topic asking if anyone has 60D, or similar sensor data from Sharpcap as you'll likely get more people seeing it.

Alan

[MalcolmP]

4 hours ago, ollypenrice said:

It doesn't take long to run sets of exposures of various lengths in order to find out what works best for you. It is also, let's remember, a necessary part of the scientific method....

Yes.
But you may have missed, in my first post, why I wished to evaluate rn and lp :-
"Up until now I have been entertaining myself ,,,using the 60d ,,, fixed on a tripod. ,,,  but now I am wondering if I should progress to an alt-az goto for 10sec subs or to a polar aligned device for even longer subs?" 
So yes I would if I could but I cannot yet do as you suggest. 
Thus far I have determined that 2 and 5 sec exp at iso800 work best for me at 135 and 50mm fl  without tracking &/or guiding.
 

Edited February 23 by MalcolmP

[MalcolmP]

4 hours ago, symmetal said:

It's worth taking a look at capture programs like APT or NINA.

Thanks, yes already checked them out, NINA only supports win10&11.  APT does run and may be useful in the future after I buy a bigger monitor to read its screen !
Meanwhile I have found a YTube vid describing a method to measure rn with Siril which also runs on my win7. So I'll go play with that for a while

 

[MalcolmP]

Just thinking aloud and because I feel like explaining, some may be wondering, why I raised my 2 original questions :
There was recently elsewhere a very long post about the effects of various noise sources on the quality (ie.SNR) of an image obtained with many stacked subs.
Also posts raising the question of total integration time, you know " are a few long subs better than lots of short subs " are very common and often end up very long with much theory !

In my case I am well aware that the more subs the better, regardless of their individual length without worrying too much  about the contribution of read noise (rn), (one does with what one has to and soldiers on) but realising that it was probably significant in my case of very short subs (2sec and 5 sec) and that I should probably take steps to mitigate that, like an az goto or a full eq setup.

However I do like to be able to 'do astro' quickly between clouds or worse ! So to be able to plonk a camera on a tripod and capture a passing supernova in galaxy far far away while also preparing roast ducks and entertaining guests at the same time was very satisfying  https://stargazerslounge.com/topic/419292-caught-another-sn-with-my-simple-dslrlens-fixed-on-a-tripod/

Knowing that recent DSLRs are reported to be very noise free and not needing darks to be used, (allegedly ! which also often generates long topics ), yes, I know, that is not related to rn but  hear me onwards , ,  and because in that topic I refed above there was mentioned millisecond exposures, I got to wondering if read noise was now also less of a problem for exposures of many seconds than had  previously been thought, you see.

So, I decided steps to find out my practical individual noise levels might be interesting to do ! Hence the op.
It is (has turned out to be) quite an involved process and I may throw up my hands in horror and wait till I obtain a new computer with which to run SharpCap and give up on this quaint Victorian notion of educating myself

There is an alternative :- I could use 'her' computer, but you know - - - "It was ok until you did ,,, "

Ok, we can all go back to sleep now  thanks for reading.
 

 

Edited February 24 by MalcolmP

[symmetal]

From your last post Malcolm, I think that the Star Adventurer would be the best option for you. The first two are not Goto though. Even if you don't polar align but just set it to your latitude and in the home position point it roughly northwards, you would be able to use significantly longer exposures than you can get with a tripod before your stars get elongated. Alt/Az mounts need to track in two axis and at varying speeds depending on where you're pointing, as well as giving field rotation at longer exposures, while EQ mounts just track in one axis at a fixed speed so should always give better results, other things being equal. 

Star Adventurer Mini Bundle £249

Star Adventurer 2i Wi-Fi Pro Pack £335

Star Adventurer GTi £469 with Goto

DSLRs not needing darks is more to do with them not being able to set a fix sensor temperature, and as darks vary with temperature, using them tends to produce worse results then not using them. 🙂

Alan

[MalcolmP]

4 hours ago, symmetal said:

Even if you don't polar align but just set it to your latitude and in the home position point it roughly northwards, you would be able to use significantly longer

Well done, you are telepathic (** see below)

Yes alt-az do need to, and alt-az-goto (as per eg. the S50) do it to track the object gone to There was once upon a time on the web a well researched paper(pdf) plotting for many latitudes the acceptable(for defined limits) field rotation at all manner of pointings in the sky.
(I can no longer find it on the web or the safe place I put it on my HD !!lol)
The best being due east and due west where the az tracking is minimum and most of the motion required to keep track is vertical-ish such that 30sec exp were reasonable. For south at 45deg el I think was worst at 10sec or a bit less.
Certainly the altaz of the S50 smart is keeping folks happy with 10sec for field rot.

But having said all that I thank you for the Star Adventurer suggestion/options.
I have often hovered my finger over offerings in the for sale section but never done the deed, always meaning to look up which model/pack/addons does what
So I think from what you have listed a mini would be enough for me.
Lots are posted on the forum about getting polar aligned accurately, but never anything about **^ needed accuracy, so that was going to be one of my future posts ^** especially as I cant see polaris from my fav spot in my garden.
If plumbob and compass-line-on-ground was going to be good enough for my humble needs to get beyond 2 and 5sec that would  be a good start.

Re darks, I'll come back on that later [ if you are game for more (!!) ]  suffice to say for now that I have been using lights and just a hundred bias in a master bias.
 

[MalcolmP]

Part 2 Darks and Bias, This is mostly for @ollypenrice to giggle over,
 Ah yes I see, thanks for the clarification that makes sense as I did find that temps were all over the place.

When I first got a 60d I did do a sort of semi-scientific suit-me test (#1 Olly will be pleased to know  ! ) by doing stacks of 100each, lights only, lights+darks, lights+dark+bias and lights+bias. I found that there was no (or little) significant difference in the last 3. So I began to use only L+B as this suited me in terms of effort and time saved.
I am sufficient of a scientist to be aware of observer bias, systematic error and the need for double blind, but hey this is a sport init ! so chi^2 coef et al can go elsewhere.
To do later was to increase the subs to 500 and then 1000 but roundtuit failed (#2 Olly can scold me for weak moral fibre ) and I stuck with L+B ever since. 
Another as yet unfulfilled ambition was to sort darks into exif temp groups since lights are reasonably predictable per run, it was darks that were the main prob.  but but,
I think I have read that ASTAP can match darks with lights by temp. on the fly but I may have misunderstood
 

Edited February 25 by MalcolmP

[sharkmelley]

On 22/02/2024 at 18:12, MalcolmP said:

Questions :-
(1) How can I measure my read noise? *
(2) Given a lp level **(of eg. 20mag/sqarcsec) how long does my sub exposure need to be for my rn to be 5x *** lower than the lp

 

Your Canon 60D read noise at ISO 800 is approx 3e:

https://www.photonstophotos.net/Charts/RN_e.htm#Canon%20EOS%2060D_14

If you want the light pollution noise to be 5x higher then you want it to be 15e.  That means you need to collect 225 (i.e. 15^2) electrons.

http://tools.sharpcap.co.uk/ will allow you to approximate your sky background rate (e/sec) which will depend on the f-ratio of your lens/scope.  You can then work out how long your exposures need to be to achieve 225e.

[MalcolmP]

2 hours ago, sharkmelley said:

Your Canon 60D read noise at ISO 800 is approx 3e:

https://www.photonstophotos.net/Charts/RN_e.htm#Canon%20EOS%2060D_14

If you want the light pollution noise to be 5x higher then you want it to be 15e.  That means you need to collect 225 (i.e. 15^2) electrons.

http://tools.sharpcap.co.uk/ will allow you to approximate your sky background rate (e/sec) which will depend on the f-ratio of your lens/scope.  You can then work out how long your exposures need to be to achieve 225e.

Brilliant !
Excellent, perfect.  Many thanks.👍

 

For my fav 135mm f2.8 lens the answer =2.99 e/pixel/s
thus 255/3= >85sec for a good sky
or 255/50= >5sec for a 17sqmB9 sky !
Now I can spend a cloudy day tabulating info for my various lenses and skies

[MalcolmP]

5 hours ago, sharkmelley said:

https://www.photonstophotos.net/Charts/RN_e.htm#Canon%20EOS%2060D_14

 

 

Random extra just thinking out louds :
When I was new to my 60d I googled  many sites to find info on optimum ISO. (sensitivity vs. dynamic range) and it seemed at that time the curves bottomed out in the region of 1600, so coservativly 800 was chosen, However looking at the curve on that site and considering that for a  faint specific object*, not a wide range pretty picture, dynamic range is not the major issue, it looks like I should be doing some experimenting* in the 1600 region

* *just proves the old adage that extra heads on a forum are better than one crying alone into his beer  
 

[sharkmelley]

If you choose ISO 1600 then shorter exposures will be sufficient to achieve your chosen 5*read_noise criterion.  

[MalcolmP]

Thanks The difference (down to 2.3e) may be marginal     something more to investigate, but  I must not get too OCD about this ! 
If I get another look at the SN when the moon gets out of the way it might be an interesting test because it is fading quite rapidly !

 

Edited February 27 by MalcolmP

[MalcolmP]

A happy couple of hours spent learning how to drive LibreOfficeCalc and plot pretty graphs therein
Nice exponentials.