•

# My ideal subs exposure time is 1.7 seconds?

10 replies to this topic

### #1 Juggernaut

Juggernaut

Viking 1

• topic starter
• Posts: 509
• Joined: 17 Feb 2013

Posted 08 December 2023 - 10:14 AM

Hi,

Can someone confirm my calculations. Seems nuts.

I'm looking at 5% noise (C factor of 10)

My camera ASI585 has read noise of 0.8e (color camera) ®

I am in Bortle 5.

My scope is f/4.7 (so assuming f/5 from the chart I have). (P = 3.7)

So based on calculation (optimum sub exposure length = C x (R^2/P)) my ideal exposure time for my subs is: 10 x (0.8^2/3.7) = 1.7 seconds????

Am I doing something wrong or is that the correct information?

Thanks!

Paul

### #2 sharkmelley

sharkmelley

Fly Me to the Moon

• Posts: 7,264
• Joined: 19 Feb 2013
• Loc: UK

Posted 08 December 2023 - 10:19 AM

Do you have a reference for that equation?  Without a reference it's not at all clear what it means.

### #3 bobzeq25

bobzeq25

ISS

• Posts: 34,151
• Joined: 27 Oct 2014

Posted 08 December 2023 - 11:04 AM

Glover says F5 Bortle 5 color camera is 51, for 2.5e read noise.  It's linear for read noise, so that says 17 seconds for you.  Probably close if swamping read noise is your criteria.  You maybe could go a bit longer if saturated pixels are your measure, say 30 seconds (depends on target).  You don't need more than 30 seconds.

I'd say 15-30 is "close enough".  At that point subexposure is not the issue, it's total imaging time.

Worry about subexposure less.  Total imaging time drives the train.  Shoot more subs.  <smile>

Edited by bobzeq25, 08 December 2023 - 11:06 AM.

### #4 unimatrix0

unimatrix0

Fly Me to the Moon

• Posts: 6,205
• Joined: 03 Jan 2021

Posted 08 December 2023 - 11:34 AM

I really don't think we can boil it down to a "works-for-all" - formula.

Doing a lot of Sharpcap exposure evaluation,  it always needs the current sky brightness measurement and camera sensor analysis data the determine the optimal sub exposure length.

That means I get varying number of sub exposure recommendation on the same target at different nights,  or different targets at the same night.

After doing that for 2 years mostly out of curiosity, I already know the ballpark number I should use

### #5 sharkmelley

sharkmelley

Fly Me to the Moon

• Posts: 7,264
• Joined: 19 Feb 2013
• Loc: UK

Posted 08 December 2023 - 12:05 PM

Your ASI 585 has pixel size of 2.9 microns
Weakest colour channel is blue with an average QE of 40% so let's base the calculations on that

So plug Bortle 5, f/4.7, 2.9microns, QE 40% into the Sharpcap Sky Background calculator: http://www.tools.sharpcap.co.uk/

It gives 0.61e/pixel/sec

You want to swap your background sky count to swamp read noise (0.8e) squared by a factor of 10:

Exposure time = 10*(0.8^2)/0.61 = 10.5sec

### #6 smiller

smiller

Mercury-Atlas

• Posts: 2,993
• Joined: 27 Oct 2018
• Loc: Vancouver Washington (not BC!)

Posted 08 December 2023 - 12:12 PM

Your ASI 585 has pixel size of 2.9 microns
Weakest colour channel is blue with an average QE of 40% so let's base the calculations on that

I’m really interested in understanding the average quantum efficiencies of these CMOS cameras, so can you point me to your data source for this?

The reason I ask is the ASI website shows a higher number for the average quantum efficiency of the blue channel so I’m not sure if that’s relative efficiency and I need to use another source or if that’s absolute quantum efficiency.

### #7 sharkmelley

sharkmelley

Fly Me to the Moon

• Posts: 7,264
• Joined: 19 Feb 2013
• Loc: UK

Posted 08 December 2023 - 12:26 PM

I’m really interested in understanding the average quantum efficiencies of these CMOS cameras, so can you point me to your data source for this?

The reason I ask is the ASI website shows a higher number for the average quantum efficiency of the blue channel so I’m not sure if that’s relative efficiency and I need to use another source or if that’s absolute quantum efficiency.

I looked at the QE graph here:

https://astronomy-im...oduct/asi585mc/

which peaks at 80% in blue.  I reckon the average QE for blue is somewhere in the range 40-50% if you consider that 550nm is the upper end of the blue transmission (from the graph).

### #8 smiller

smiller

Mercury-Atlas

• Posts: 2,993
• Joined: 27 Oct 2018
• Loc: Vancouver Washington (not BC!)

Posted 08 December 2023 - 12:34 PM

I looked at the QE graph here:

https://astronomy-im...oduct/asi585mc/

which peaks at 80% in blue.  I reckon the average QE for blue is somewhere in the range 40-50% if you consider that 550nm is the upper end of the blue transmission (from the graph).

Are you looking at blue sensitivity over the entire spectrum range or only within the “blue“ tange of about 400 to about 500 nm?

To the best of my understanding scaling the numbers Robin Glover uses in his video is based on the quantum efficiency of the color cameras within their respective color spectrum range for each color sensor.  Then for color cameras he multiplies his mono number by 3 to account for color sensors each capturing about 1/3 of the light, which we all know is just a rough estimate, but should be good enough for a first order approximation.

Edited by smiller, 08 December 2023 - 12:44 PM.

### #9 imtl

imtl

Cosmos

• Posts: 7,859
• Joined: 07 Jun 2016
• Loc: Down in a hole

Posted 08 December 2023 - 12:37 PM

The readout noise depends on the gain you are using. It's not a flat 0.8e-rms. If you are using it at HCG then you are at 1.5e-rms.

Assuming this formula (which is at best not complete) then at PEAK of Blue which is 80% efficient according to the website:

exposure time=10*(1.5^2)/1.22=18.44sec

If you go for 40% efficiency like Mark wrote (I think it is more or less correct as an average QE over the range) then you will be getting

exposure time=36.88sec

### #10 sharkmelley

sharkmelley

Fly Me to the Moon

• Posts: 7,264
• Joined: 19 Feb 2013
• Loc: UK

Posted 08 December 2023 - 12:48 PM

Are you looking at blue sensitivity over the entire spectrum range or only within the “blue“ tange of about 400 to about 500 nm?

To the best of my understanding scaling the numbers Robin Glover uses in his video is based on the quantum efficiency of the color cameras within their respective color spectrum range for each color sensor.

I was assuming 550nm.  However, to be consistent with the Sharpcap sky background calculator it's probably best to average over a 100nm range of 400-500nm which is more like 70% giving a rate of 1.07e/pixel/sec

### #11 smiller

smiller

Mercury-Atlas

• Posts: 2,993
• Joined: 27 Oct 2018
• Loc: Vancouver Washington (not BC!)

Posted 08 December 2023 - 01:01 PM

OK, I’ve used the Robin Glover tables as a baseline for my calculations many times, so here is my full calculation based on Juggernaut’s configuration and situation.  He’s talking using short exposures on a Goto Dob, very similar to mine, and I started with a similar camera, so I’ll configure it as I would if I was in his shoes:

Camera gain: 300 (gives the best combination of low read noise at ~1.1 and a large enough well depth to not be blowing out stars with Juggernaut’s likely range of exposures times from 5-10 seconds).

OK, from Robin Glover’s table for a mono CMOS camera and Bortle 5 and an F/5 scope:

17 seconds starting number.  Now let’s put in all the modifiers.

1) Juggernaut’s F-ratio of F4.7 delivering slightly more light gathering power per pixel:

17 * (4.7/5)^2 = 15.02 seconds

2) Juggernaut’s use of 300 gain/1.1e read noise versus the 2.5e in the benchmark:

15.02 * (1.1/2.5)^2 = 2.91 seconds.

3) Juggernaut’s use of a camera with ~~70% quantum efficiency (just to use sharkmelley’s last number which is probably in the ballpark)  versus the 50% in the benchmark:

2.91 * (50/70) = 2.077 seconds

4) Juggernaut’s 2.9 um pixels versus the benchmark of 3.75um pixels, so less light gathered per pixel:

2.077 * (3.75/2.9)^2 = 3.47 seconds

5) Juggernaut using a color camera, so only about 1/3 the light per pixel vs a mono camera of similar technology:

3.47 * 3 = 10.4 seconds

So I think if Juggernaut wants to keep read noise to about 5% of the total stack noise on this camera and scope at 300 gain, in B5 skies shooting broadband, he needs exposures times of roughly 10 seconds.

If he has to use 5 seconds, then his stack noise will have an additional 10% noise due to the read noise and to get back to the amount of noise that someone with very long exposures experiences with said setup, he would need to use up to (1.1)^2 = 121% of the imaging time, or about 21% more imaging time.

Hopefully I didn’t miss anything big here.

if anyone is interested in the fundamental equations used to generate the table they can watch Robin Glover’s video and he goes through all the equations and you can actually build a spreadsheet that generates the same results.

My only issue is Robin Glover‘s use of the term “optimal sub length“ when he really should have said “sub length required for 5% read noise”, because when doing astrophotography the optimal sub length involves balancing many different trade-offs and he doesn’t go into those so that just is an unfortunate term that he used.

Assuming my calculations are close to correct, here are the big hitters for impacting this number:

1) What your actual sky darkness is.  B5 has a pretty big range, so it can be 1/3 shorter or 50% longer depending on what your actual sky scale is.

2) If the moon is out or you are shooting low in the sky:  The Bortle scale is generally a good night straight up from what I understand.  Actually shooting conditions are often less dark than that so exposure times can be shorter.

3) If you get that Nexus reducer you mentioned once before.  That lets about 78% more light per pixel, so minimum exposure time is 1/1.78 as long.

4) If you get a larger pixel camera

5) If you get a dual narrowband filter, the number will increase dramatically:  very roughly proportional to 1/(% of light that gets through to each sensor relative to broadband).   For the IDAS NBZ at about 12nm bandpass per color channel versus about 100nm, it's about 1/(12.5%) or 8x longer.  For the L-Ultimate at 3nm per channel vs 100nm, it's 33x longer.  These are crude calculations, you can do a more precise calculation using the actual spectral sensitivity of the camera, filter bandpass %, etc...

Edited by smiller, 08 December 2023 - 01:30 PM.

## Recent Topics

 Cloudy Nights LLC Cloudy Nights Sponsor: Astronomics