# Optimal Sub Exposure Length

## Recommended Posts

I've just finished watching the excellent talk given by Dr Robin Glover at the 2019 Practical Astronomy Show, which is available on YouTube here if you haven't seen it.

It's a very well set out presentation and I can highly recommend it for those who like myself are delving into the subject a little beyond some of the classic rules of thumb.

However it did leave me with a few questions that I hope some of the knowledgeable membership here might be able to help me with.

Those in the know will recognise that I'm talking about the following equation:

• R = Camera Read Noise
• P = Sky Electron Rate (Calculated from http://tools.sharpcap.co.uk/)
• E = Acceptable % increase in Noise above minimum

When I work this through I get the following exposure times...which seem, well a little long!

* calculated for Atik 383L+ OSC (QE of 57% Pix Size 5.4μm) at f7 with site SQM of 21.89 mag/arc sec2 (e.g. Kielder Star Camp)

I could probably manage 735 sec, but does this mean I'll be getting significant noise which I'll then I'll still need multiple hours of exposures in order to stack enough images to benefit from the noise reduction?

What I'm curious to find out is if there is a way to work out how many images you need to stack to get the same benefit in noise reduction in relation to the minimum noise threshold as a long exposure.  Exposures of 20min plus seem extremely risky.

I'd appreciate some thoughts on this.  I was initially just going to try and knock out as many 5min subs as possible on each target (ideally 30-50 weather permitting) but now I'm a little unsure.

They do say a little knowledge is dangerous! 😛

##### Share on other sites

My honest (but perhaps misguided) opinion is that, while you are wading through all this stuff, you could be doing something more productive if your aim is to produce good images at the end. I really do not believe that the difference between good, bad and indifferent images comes from whether or not they have been made using optimal sub lengths. Two hours well spent on learning something new in processing will give you a benefit worth 10x the difference between a reasonable and a perfect sub length. This is particularly true with CMOS sensors of low read noise. Prior to that, using CCD, it was more of an issue: the critical thing in those days was to ignore all the tech gurus who told you that there was no point in doing 30 minute subs in search of faint stuff - because they were wrong.

Experiment and, above all, learn about processing.

Olly

• 3
##### Share on other sites

Thanks Olly.  My biggest problem is that at the moment I don't have many opportunities for imaging outside of the few weeks I can spend at kielder each year.  So I've not had the chance to tweak and fiddle and test...only to read, tweak and fiddle.  So I'm trying to get as much knowledge about the kit I have so that when I do get out I'm making the most of my time.

I've been going through my software using ascom simulators to test my understanding of my processes and I've had a short test of everything so that I know my hardware is working.  So that's why I'm now focusing on making sure the data I get is the best it can be.

I can delve deep into my processing for the next six months between sessions, when I'm sure you'll see me post about affinity2 or if I take the plunge, pixinsight.

Interestingly though I suspect you unwittingly answered my question with your quip about old ccds needing 30min subs.

I am using an old atik383 OSC so I'm guessing yes, aim for as long as I can reliably guide for. Right?

##### Share on other sites

If you've seen the same video I have I get the impression the general conclusion was not to dwell too much on specifics, as long as your total integrated time maximises signal and averages out noise there will eventually be a limit of minimal returns.

As long as your singular images do not clip data via oversaturation (highly unlikely) you'll be alright. Photons will hit an imaging sensor regardless if the exposure is short but many, or long and few. The benefit of the former is you'll have more frames to average out noise.

##### Share on other sites

It's the increase in noise over the perfect scenario, but if you're at a dark sky site, then 15% increase on very low background noise is still going to give you a much better result than a 1% increase on the optimum result from a bortle 5 site for example.

It's telling you that longer exposures are better, but the actual total noise is still determined by your sky quality, so no you wont have a significant amount of noise given the sky quality you're using to calculate.

##### Share on other sites

Posted (edited)

This is the key slide, 15% increase above that bottom line is still much better than the two lines for the light polluted skies, no matter the sub length

Edited by SamAndrew
• 1
##### Share on other sites

8 minutes ago, SamAndrew said:

This is the key slide, 15% increase above that bottom line is still much better than the two lines for the light polluted skies, no matter the sub length

Thanks Sam that makes sense...I know I need a long total integration time and am aiming for at least 20 subs to get the most out of stacking so now I can not worry too much about pushing past 10min subs.

I just hope the bloody snows gone before I head up on Sunday!

##### Share on other sites

11 hours ago, MG01 said:

Interestingly though I suspect you unwittingly answered my question with your quip about old ccds needing 30min subs.

I am using an old atik383 OSC so I'm guessing yes, aim for as long as I can reliably guide for. Right?

It does depend on the target and total integration time remains the key thing. However, with your Kodak chip I would aim for at least 10 minutes and maybe 15. Certainly 15 for narrowband. Going for 30 minutes with a mobile rig and a limited time window would be a risk likely to bring more cost than benefit.

A friend using your chip was advised by a well-known imager to use shorter subs. She tried it and rejected the idea, returning to longer ones. (I can't remember the precise details.) My own experiments gave me the same results.

Olly

• 1
##### Share on other sites

It’s a shame you have the OSC version of the 8300 sensor and not the mono, but you have to go with what you have.

You have a dark sky so 10 minutes sounds like a good figure to try if your guiding is good enough.

• 1
##### Share on other sites

Thanks for the feedback.  I also wish it was the mono....I think you saw my face palm thread when I realised why I'd managed to pick up such a 'bargain'! 😆🤣

But that's okay.  I'm happy taking baby steps for the time being.

## Create an account

Register a new account

• ### Similar Content

• #### Which telescope focal length optimal for DSO nebula etc ?

By 2Jules,

• 6 replies
• 377 views
• #### Pixinsight weighted batch processing subs different length

By iwols,

• 2 replies
• 232 views
• #### HDR processing without different exposure lengths

By Icesheet,

• 3 replies
• 336 views
• #### Different sub lengths?

By paul mc c,

• 9 replies
• 654 views
• #### Exposure failure

By Bullmastiff,

• 21 replies
• 424 views
×
×
• Create New...