Bortle 7 skies - should I take 180s or 300s exposures?

[Steve143]

Hi,

I'm imaging with the ZWO 533MC Pro in Bortle 7 skies. I'm guiding and dithering and can get 300s exposures without any issues but would I be better off shooting 180s exposures given the quality of my skies? I seem to recall someone say that 180s exposures in such light-polluted skies are better than 300s exposures. However, during post-processing my images from 300s exposures look ok - just haven't tried 180s exposures yet to see if there is any significant difference. I also shoot with an L-Pro or L-extreme depending on the target. 

Thanks,

Steve 

Edited September 5, 2023 by Steve143

[Elp]

It depends on your light pollution level, only you can determine that. Also depends on your target.

[vlaiv]

Depends on your setup and best approach is to measure it.

[vlaiv]

Just now, Elp said:

It depends on your light pollution level, only you can determine that. Also depends on your target.

Does not depend on target unless target is brighter than light pollution (which is rarely the case).

Read noise has to be compared to highest source of noise that depends on time (dark noise, target shot noise or LP / sky shot noise, last one is usually the highest when we use cooled cameras).

[Elp]

Say you're imaging the squid (I know different equipment), because its O3 signal is so faint, is it not better to long expose to get better pixel activation (full well) rather than shorter if you compare both exposure times (everything and total time being equal) because stacking is an averaging process?

Edited September 5, 2023 by Elp

[vlaiv]

7 minutes ago, Elp said:

Say you're imaging the squid (I know different equipment), because its O3 signal is so faint, is it not better to long expose to get better pixel activation (full well) rather than shorter if you compare both exposure times (everything and total time being equal) because stacking is an averaging process?

I'm saying that decision on individual exposure does not depend on how faint your target is.

Here is simplified explanation. Say you take one 300s exposure and you take two 150s exposures and add those 150s exposures (not average, but simply add them, although averaging and adding is really the same - but let's keep it simple).

Only difference between those two exposures - 300s in one go or 300s in two goes will be how much read noise there is. Everything else is the same - in both cases you accumulated 300s of everything - target signal, sky signal, dark signal ... it all grows linearly with time and it does not matter if you split accumulation into multiple parts. All except read noise.

Read noise grows with number of exposures as each exposure adds one "dose" of read noise. Ok.

Now onto the next step.

Noise adds like linearly independent vectors - like square root of sum of squares. Same thing as Pythagoras theorem, right?

In above image AC is larger than both BC and AB, so is FD compared to EF and DE - but notice one thing - as one side of triangle becomes much shorter than the other - hypotenuse starts being just a bit longer than other side.

This is the key for determining sub duration - you want hypotenuse (total noise) to be almost the same as largest noise source (LP noise) - and this happens when short side of triangle (read noise) is much smaller than long side of triangle (LP noise).

This only works when stacking - it makes no sense for single exposure. Final difference in noise in stack (signal will be the same regardless - there is certain amount of signal in 300 seconds how ever you slice it) will be when read noise is sufficiently small compared to LP noise. Ratio of the two is important, and I advocate ratio of 5 - have your LP noise at x5 that of read noise then total noise will be just:

total = sqrt(lp_noise^2 + (lp_noise/5)^2) = sqrt( ln_noise^2 * (26 / 25) = lp_noise * sqrt(26/25) = lp_noise * 1.0198 = lp_noise increased by ~2%

You'll be hard pressed to see difference of 2% in noise increase - if read noise is 1/5 of LP noise - it will be like LP noise is %2 higher and there is no read noise.

Makes sense?

In any case - target brightness does not play a part in above explanation unless target is brighter than the sky and then we should use target shot noise to swamp read noise rather than sky shot noise / lp noise.

[Steve143]

Wow!! Thank you so much! This is incredible. 

I shoot with an 80ED-R and all controlled with the ASIAIR Plus. Is there an easy way to measure LP and read noise with this setup?

Thanks again. 

[vlaiv]

1 minute ago, Steve143 said:

I shoot with an 80ED-R and all controlled with the ASIAIR Plus. Is there an easy way to measure LP and read noise with this setup?

Not really sure, but all it really takes to measure it is calibrated sub form one of you previous recordings and knowledge of gain you used.

ZWO publishes read noise vs gain for their models - so you can use that:

and all you need to do now is to measure sky background in electrons - which is a bit tricky part because you need to properly convert ADU units that you measure back to electrons via e/ADU which is also published for selected gain by ZWO:

Once you have average background signal and read noise - then it is easy.

Say your background is 200e and your read noise is 1.5e. Important thing to note is that LP noise is equal to square root of sky signal so LP noise will be ~14.14 (square root of 200). That is ~x10 larger than read noise, so you can halve LP noise which means 1/4 of background signal (signal is square of noise) - In above example you can use 1/4 of exposure time you used to make that sub which has 200e background and 1.5e read noise.

Makes sense?

Btw - there are couple of threads on SGL that go into depth how to measure background / convert from ADU back to electrons (it's sometimes tricky as it depends on bit depth of your camera and so on) - so do a search and see what you can find. There might even be software that will do above for you. I've glanced over some discussions - maybe SharpCap can do it or some other similar software?

[Steve143]

Thanks again. I always use a gain of 101 with the 533MC Pro as it seems to be the optimum gain for this camera. 

[vlaiv]

2 minutes ago, Steve143 said:

Thanks again. I always use a gain of 101 with the 533MC Pro as it seems to be the optimum gain for this camera. 

In that case - just check ADU value of highlight in your image (Star core) if it's 65000 something something - divide background value with 4 to get electrons, else if highlight is around 16384 then use number you measure as average / median background (select patch of empty sky to measure value)

[The Lazy Astronomer]

Aside from what's been mentioned above, the other consideration to make is storage space and stacking time. Probably not such an issue with the relatively small files from the 533 though, unless you end up going really short. 

[vlaiv]

Very good point from @The Lazy Astronomer above - what's been said is lower bound (and not a strict one). If you feel comfortable with going with longer subs - by all means.

There will be less of them to store and manage if you go with longer subs. Sometimes it is better to have more data to work with and in case of accident (like bumping the mount or someone shining a torch towards the telescope) - there is less data to waste if you go with shorter subs (it is better to discard 1 minute than 5 minutes of your imaging time). Satellites and passing planes are handled by sigma rejection, so you should not count those as "discards".