Gain/ISO

[Gavin1234]

Anyone know if it’s possible to calculate the right amount of gain for the particular exposure time you’re trying to shoot? Or is it just trial and error? Seems time consuming to wait for each shot to finish before each adjustment.

If so any suggestions on where I should start? I’m using gain of 180 out of the max 240 for 5 min exposures at the moment which works well. I want to do 15 or 20 min exposures which show as a bright white screen at 180. My camera is a ASI071MC.

[xelasnave]

I am in a similar situation and am trying various settings to get a feel.
alex

[kens (Ken)]

You want to set your exposure according to the read noise on your camera. Aim for an exposure where the histogram peak corresponds to between 5x and 10x the read-noise squared.
You can look up the read noise on the ZWO website - it varies with gain.
You have to convert the read noise to the numbers you see on the histogram using the gain in e-/ADU, offset and number of bits in the camera.
For example: for the ASI071 at unity gain (94) the read noise is 2.6e-
So 10 x RN^2 = 68
Offset is either 20 or 65 (depending which driver) - lets say its 65
And it is a 14 bit camera

So you first convert the 68 to ADU by dividing by the gain in e-/ADU which at unity is 1.0
So you have 68 ADU
Now add offset of 65 which gives 133
And multiply by 4 to convert to 16 bit equivalent which gives 530

So expose till your histogram peak is around 530. Its not highly prescriptive. But longer exposures offer little extra benefit over more subs.

[Gavin1234]

Thanks Ken. I’ll go through my numbers during the week.

[Stonius (Markus)]

Originally Posted by kens

You can look up the read noise on the ZWO website - it varies with gain.

I always find this confusing. for the 1600, RN is listed as 1.2e @30dB gain.

I assume this doesn't mean 'Gain setting 30' because according to the graphs in the manual, the gain units are in units of 0.1dB. So dividing by 10 can I assume that 300 Gain is 30dB and the magical point at which read noise is referenced to 1.2e?

As mentioned, it varies by gain, but I'm not sure how to calculate it for *other gains. I'd rather calculate it exactly rather than read values off those hokey little graphs but I don't know how.

Markus

[Stonius (Markus)]

Originally Posted by Stonius

I always find this confusing. for the 1600, RN is listed as 1.2e @30dB gain.

I assume this doesn't mean 'Gain setting 30' because according to the graphs in the manual, the gain units are in units of 0.1dB. So dividing by 10 can I assume that 300 Gain is 30dB and the magical point at which read noise is referenced to 1.2e?

As mentioned, it varies by gain, but I'm not sure how to calculate it for *other gains. I'd rather calculate it exactly rather than read values off those hokey little graphs but I don't know how.

Markus

Working through your example and just going by the graphs, I get 141 at unity (offset 21) and 368 for Gain 200 (offset 50) for a ZWO1600.

On my test images of 47 Tuc I was getting the background around 2800 (16bit values) on Luminance subs of 30 secs at Gain 75, -20 degrees Celsius. I can't remember the offset I think it was 21. Maybe that's all wrong. The brightest parts of the image are at around 32,000. Statistics show the brightest pixels at 65504, just 31 shy of clipping at 65535. I think this was the basis on which I was exposing, TBH - not clipping, rather than skyglow)

It was a heavily light polluted area. Maybe I should have been using gain 200 with shorter exposures.

Markus

[ChrisV (Chris)]

Shiraz made an excellent spreadsheet to calculate the optimal exposure time to smother read noise for a bunch of cameras such as the asi071. For my outer Sydney suburban skies, unity gain with an 8"F5 newt it's about 90-120sec. I vaguely remember that it gives me sky values of about 10-20xRN^2; following the calculations mentioned below. If you make exposure time longer there's no appreciable increase in SNR, it just makes saturation of stars worse.

Here's the link to his thread and another about it:
http://www.iceinspace.com.au/forum/s...d.php?t=151000
hhttp://www.iceinspace.com.au/forum/showthread.php?t=160444

Edit: made a mistake it gives 5-6xRN^2 at 90-120sec subs

[kens (Ken)]

Originally Posted by Stonius

Working through your example and just going by the graphs, I get 141 at unity (offset 21) and 368 for Gain 200 (offset 50) for a ZWO1600.

On my test images of 47 Tuc I was getting the background around 2800 (16bit values) on Luminance subs of 30 secs at Gain 75, -20 degrees Celsius. I can't remember the offset I think it was 21. Maybe that's all wrong. The brightest parts of the image are at around 32,000. Statistics show the brightest pixels at 65504, just 31 shy of clipping at 65535. I think this was the basis on which I was exposing, TBH - not clipping, rather than skyglow)

It was a heavily light polluted area. Maybe I should have been using gain 200 with shorter exposures.

Markus

The 1600 has a 12bit ADC so you need to multiply by 16 rather than 4 so your target values should be 4x higher.
Also for the 1600: RN=0.482*Gain (in e-/ADU) + 1.178
At unity with an offset of 21 I get a target value of 777 as the RN is 1.7
A histogram peak at 2800 is quite high so you would get more clipping than if you targetted 700 as per the calculation for 70/21
Don't go higher in gain. 30s subs are already short and you still need the integration time.

[xelasnave]

Hi Ken
I have yet to get my head around the numbers, my brain has been run hard for a while now working out stuff that I find new and very difficult which is a reflection of my age, perpetual tiredness due to clear skies and a huge work load, so in that context my I ask the following.
For my set up I am going for 2 minute subs could you suggest a range for gain so I have a ball park setting area until I can make up a prompt sheet for in the field.
It may sound funny but what do you mean when you say off set?
I have been using gain around 250 which gets me something but after the Pink concert which will see me gone until Monday night I will only have a couple of days left before I have to go back to Sydney so I want to get as close to what I should be doing with what I have planned.
I captured a range of subs on the Prawn Nebula last night and want to get a few more tonight but as I have to pack up early ( everything put away rather than leave it out while away) because of an early start I want to get the best I can while the Moon is less bright and than next week...I may manage only two hours and I am thinking to get more R B G and luminance in that order and next week get more narrow band.
So I am looking for a guide to what gain for tonight.
Alex

[Stonius (Markus)]

Originally Posted by kens

The 1600 has a 12bit ADC so you need to multiply by 16 rather than 4 so your target values should be 4x higher.
Also for the 1600: RN=0.482*Gain (in e-/ADU) + 1.178
At unity with an offset of 21 I get a target value of 777 as the RN is 1.7
A histogram peak at 2800 is quite high so you would get more clipping than if you targetted 700 as per the calculation for 70/21
Don't go higher in gain. 30s subs are already short and you still need the integration time.

Okay, Ken, that makes more sense, Thanks.

So is the general idea to keep the read noise separate from the sky fog, exposing just enough that you keep the rest of the dynamic range free for signal?

Cheers

Markus