What's the longest exposure you use personally?

[pipnina]

I've done a lot of imaging at 10 and 15 minute exposures recently, but I'm curious as to what people tend to think is most reasonable in their setups when going deep. I attempted SH2-113 + 114 recently in H-alpha, and 9 hours of integration still only got me a very grainy image of the object. I also noticed that my minimum / average ADU in my sub frames was only about 60 minimum and 310 average on a 16-bit camera, during a full moon!

Admittedly some of this is me not realising my camera's bias was randomly getting set to 0 instead of the driver's boot-up default of 768. But a bias frame at 0 gain would likely have a minimum value of around -150 or so if the camera could record negatives, and in high gain mode my RisingCam 571 has at least 4 ADU per electron. So in each 10 minute H-alpha sub I was probably only getting 100 photons per pixel or so (on average) during a full moon.

So now I am wondering if 10 minutes is plain too little for faint targets in narrowband, even with an appropriate gain setting the number of photon ADU per pixel is lower than my read noise peaks and troughs. Could it be that I need 20 or even 30 minutes or more for some targets to see the best results?

I do see a lot of very good images on here though with even as low as 300 second exposures in narrowband, so maybe it's not all it seems.

For me, beyond 30 minutes I think my fear of guiding blips, cloud tufts and poor-fitting of the imaging session with the onset of twilight are too great, personally.

[Elp]

I used to think doing 120-300s per sub for narrowband was the way to go but any guiding blip and the subs ruined, so now I tend to go short but many, usually 60s, the more you have the better the noise averages out. Saying that when I did the iris nebula last did over 20h and it still seemed grainy, my most recent m16 was just over 2h and is very noisy but I think due to the short time was more productive. Local conditions matter a lot.

[scotty38]

I must admit to not seeing anyone, anywhere doing anything near 10 minutes let alone 20 or even 30 minutes on the 571 sensor. Those CCD length exposures seem long gone and as Elp says exposures of 5mins and less seem to be plenty long enough on the latest cmos sensors for even the faintest targets.

I'd say you need to have a good look at something else in your setup to be honest.

[iantaylor2uk]

You can go shorter if you put the gain up. This also has the benefit of getting you lower noise. If I'm imaging with a UV/IR filter I usually use unity gain (90 on my ZWO 071) and exposures of 30 or 60 seconds, If I'm using an L-Enhance filter I up the gain to 200 (max gain on the camera is 240) and use exposures of 180 secs. 

[DaveS]

I run 900 sec subs for NB with my G3 16200, while RGB is run with a first order G2v calibration, resulting in 600, 600, and 1050 sec exposures for R, G, and B.

[Fegato]

Obviously focal ratio and filter bandwidth plays into this. With dual narrowband (10nm) at F/2.2 I max at 180s, but will often shoot 60s.  At F/7 and dual narrowband it can be up to 900s, although as above, this feels a bit over the top so normally I max at 600s. If you've got a slow system and very narrow filters, you may want to go longer. You certainly need to make sure you are swamping the read noise sufficiently anyway.

re: gain, I set this for maximum dynamic range (i.e. with ZWO cooled cameras, where the HCG mode kicks in), and I'm not really inclined to increase gain to reduce exposure time. I suppose conditions or poor guiding for other reasons might make this a reasonable choice.

[ONIKKINEN]

Imaging with a Newtonian under usually a bit of wind i try to keep exposures as short as possible. 60s will do for OSC just fine under SQM 20.8 skies, but its quite close to where i would want to increase the exposure.

I have switched comacorrectors from F/4.2 to F/5 so probably will have to aim maybe for 120s subs from now on. If the Moon is out then 60s will still be enough with the Rising Cam 571.

[symmetal]

I expose to the point where the sky background noise swamps the camera read noise by a specified factor, and so making the read noise contribution of the total noise in the image insignificant. At this point I start another exposure as exposing beyond this duration has no advantage in S/N and just overexposes the stars more. I used to swamp the read noise by a factor of 3 but will use a factor of 5 in future to reduce the number of subs taken. Advice from vlaiv taken here. 🙂 At factor 3 the read noise contribution to the image is fairly insignificant while at factor 5 it's very insignificant and can be ignored.

It depends on your sky darkness but for read noise swamping by a factor 3, I expose 1 min for L and 3min for RG and B with the ASI6200 and ASI2600 with an f6.3 scope. However for narrowband imaging I would need very long exposures to significantly swamp the read noise. I'd need around 90 mins at Ha and over 3 hours for OIII and SII. This is in bortle 3 skies. If you have more light pollution that these optimum exposure values will be lower.

So, unless you have high light pollution, for narrowband, it's still beneficial to expose for as long as you are able, before mount or guiding issues, or the chance of a plane etc. flying through ruining the image becomes too great. I therefore use 900s exposures for narrowband. Well below the optimum duration for adequate read noise swamping, so read noise is still an issue for narrowband imaging, even with the latest HCG low read noise cameras.

I don't know what sensor your camera has pipnina, as RisingCam's AliExpress page doesn't seem to give model numbers, just specs. If you have graphs of Read Noise and gain in e-/ADU as ZWO and others provide I can give you a sky background ADU value to aim for, for the above Read Noise swamping factors. If not, just specifying the sensor used, I can provide values for Zwo cameras with the same sensor which should be close. 😀

Alan

[The Lazy Astronomer]

30 - 60 seconds with LRGB @ unity gain, 3 minutes Ha/Sii/Oiii @ 200 gain

The narrowband exposure is set as the minimum required such that the sky background swamps the read noise by a certain factor (I can't recall what the factor is right now).

My minimum required exposure for LRGB is <10s, so the sub length is set such that I don't end up with many thousands of frames to have to stack.

[pipnina]

52 minutes ago, symmetal said:

I expose to the point where the sky background noise swamps the camera read noise by a specified factor, and so making the read noise contribution of the total noise in the image insignificant. At this point I start another exposure as exposing beyond this duration has no advantage in S/N and just overexposes the stars more. I used to swamp the read noise by a factor of 3 but will use a factor of 5 in future to reduce the number of subs taken. Advice from vlaiv taken here. 🙂 At factor 3 the read noise contribution to the image is fairly insignificant while at factor 5 it's very insignificant and can be ignored.

It depends on your sky darkness but for read noise swamping by a factor 3, I expose 1 min for L and 3min for RG and B with the ASI6200 and ASI2600 with an f6.3 scope. However for narrowband imaging I would need very long exposures to significantly swamp the read noise. I'd need around 90 mins at Ha and over 3 hours for OIII and SII. This is in bortle 3 skies. If you have more light pollution that these optimum exposure values will be lower.

So, unless you have high light pollution, for narrowband, it's still beneficial to expose for as long as you are able, before mount or guiding issues, or the chance of a plane etc. flying through ruining the image becomes too great. I therefore use 900s exposures for narrowband. Well below the optimum duration for adequate read noise swamping, so read noise is still an issue for narrowband imaging, even with the latest HCG low read noise cameras.

I don't know what sensor your camera has pipnina, as RisingCam's AliExpress page doesn't seem to give model numbers, just specs. If you have graphs of Read Noise and gain in e-/ADU as ZWO and others provide I can give you a sky background ADU value to aim for, for the above Read Noise swamping factors. If not, just specifying the sensor used, I can provide values for Zwo cameras with the same sensor which should be close. 😀

Alan

This is the graph I saw when buying the camera initially. It uses a Sony IMX 571 sensor.

[pipnina]

17 minutes ago, The Lazy Astronomer said:

30 - 60 seconds with LRGB @ unity gain, 3 minutes Ha/Sii/Oiii @ 200 gain

The narrowband exposure is set as the minimum required such that the sky background swamps the read noise by a certain factor (I can't recall what the factor is right now).

My minimum required exposure for LRGB is <10s, so the sub length is set such that I don't end up with many thousands of frames to have to stack.

Haha. Yes with those 50mb FITS files the hard drive does fill up fast at 1min subs.

I did about 25 hours or so of the leo triplet back in march-april ish time. All in 1 min subs because my guidecam hadn't arrived yet. I had 1500 images or something and it took pixinsight about 5 hours to stack them all! And left me with a 73 GB project folder to archive! And that's after I deleted the pixinsight registered and calibrated files, which brought it up to almost 400GB!

Edited August 17, 2022 by pipnina

[ONIKKINEN]

@pipnina If you are imaging with Gain 100 and HCG mode on you have either 0.9e or 1.2e of read noise, depending on if the obscure "low noise mode" is on. Lets assume its not as some software dont have a toggle for this and i have no idea what it does anyway.

In order to swamp read noise x5 your median signal for the background needs to be read noise x5 squared in electrons.

So (1,2x5)^2 which comes out to 36e. The e/ADU conversion rate at 100 gain and HCG mode is 0.25e/ADU so you need 36x4=144 ADUs of median signal to swamp read noise x5. Youll find that under most conditions this is fulfilled easily with your fast newtonian.

[symmetal]

Your graphs are a bit confusing pipnina. There are no units on the vertical scale, and the gain in e-/ADU has no actual values. Here's the graphs for the ASI2600 (same sensor) below.

Your read noise is quoted below 1.0 for all your gain ranges and could be assumed to be in electrons (e-) but this seems a rather low figure. The ASI2600 read noise varies from around 1 to 1.5 electrons once the HCG mode is enabled. This implies that HCG mode is on all the time for your camera and so implies your gain is 10dB (100 in ZWO 0.1dB gain units) when your gain is set to 100 linear units (top row) which is really no gain (ie. a linear gain of 1) which has been called 0.0 dB (bottom row), incorrectly I believe. The IMX571 data sheet quotes gain can be set between 0 and 37dB and the HCG mode comes on at gain 10dB and above.

Your full well of around 11,000 at quoted 0.0dB gain coresponds to the full well of the 2600 at actual gain 10dB, which again implies the HCG mode is on all the time on your camera.

Your e-/ADU gain graph has no value stated, and at your 0.0dB gain setting (really 10.0dB) is actually around 0.25 e-/ADU on the 2600.

Using the ASI2600 figures for your minimum gain setting of 100 or 0.0dB which I assume you are using.

Sky background ADU = (read_noise * swamp_factor) ^ 2 / gain + camera bias ADU

= (1.45 * 5) ^ 2 / 0.245 + bias ADU

= 214 ADU + bias ADU.

The bias ADU is required as that's added to every frame you take and more importantly it includes the camera offset value used. You'll find for images at low camera gain the offset contributes most of the ADU value in your image. Just take the mean ADU value of a bias frame or your master bias if you have one as your bias ADU value. My ASI2600 offset is set at 50 and it adds 10ADU per offset unit. My bias ADU is actually 503 ADU which is 500 ADU from the offset and 3 ADU from the camera electronics, so my Sky background ADU value is 214 + 503 = 717 ADU.

It's easier to have the bias added on as that's the value displayed on any capture software when you hover your mouse over the image, and you can quickly read the Sky background value.

@ONIKKINEN has quoted the value before the offset (bias) is added on and has used different values interpreted from your graphs which may be more correct. 😉 Use whichever you want as a guide. 😊

Hope this helps pipnina.

Alan

[ONIKKINEN]

33 minutes ago, symmetal said:

Your graphs are a bit confusing pipnina. There are no units on the vertical scale, and the gain in e-/ADU has no actual values. Here's the graphs for the ASI2600 (same sensor) below.

Your read noise is quoted below 1.0 for all your gain ranges and could be assumed to be in electrons (e-) but this seems a rather low figure. The ASI2600 read noise varies from around 1 to 1.5 electrons once the HCG mode is enabled. This implies that HCG mode is on all the time for your camera and so implies your gain is 10dB (100 in ZWO 0.1dB gain units) when your gain is set to 100 linear units (top row) which is really no gain (ie. a linear gain of 1) which has been called 0.0 dB (bottom row), incorrectly I believe. The IMX571 data sheet quotes gain can be set between 0 and 37dB and the HCG mode comes on at gain 10dB and above.

Your full well of around 11,000 at quoted 0.0dB gain coresponds to the full well of the 2600 at actual gain 10dB, which again implies the HCG mode is on all the time on your camera.

Your e-/ADU gain graph has no value stated, and at your 0.0dB gain setting (really 10.0dB) is actually around 0.25 e-/ADU on the 2600.

Using the ASI2600 figures for your minimum gain setting of 100 or 0.0dB which I assume you are using.

Sky background ADU = (read_noise * swamp_factor) ^ 2 / gain + camera bias ADU

= (1.45 * 5) ^ 2 / 0.245 + bias ADU

= 214 ADU + bias ADU.

The bias ADU is required as that's added to every frame you take and more importantly it includes the camera offset value used. You'll find for images at low camera gain the offset contributes most of the ADU value in your image. Just take the mean ADU value of a bias frame or your master bias if you have one as your bias ADU value. My ASI2600 offset is set at 50 and it adds 10ADU per offset unit. My bias ADU is actually 503 ADU which is 500 ADU from the offset and 3 ADU from the camera electronics, so my Sky background ADU value is 214 + 503 = 717 ADU.

It's easier to have the bias added on as that's the value displayed on any capture software when you hover your mouse over the image, and you can quickly read the Sky background value.

@ONIKKINEN has quoted the value before the offset (bias) is added on and has used different values interpreted from your graphs which may be more correct. 😉 Use whichever you want as a guide. 😊

Hope this helps pipnina.

Alan

The ToupTek made cameras differ significantly in specs with ZWO and QHY versions, so the numbers dont really translate between the 3 all that well. All versions of the camera perform well, but for some reason the ToupTek ones really do have lower readnoise than the others. Its not a big difference for most imagers in most common conditions, but with narrowband it could be a big deal whether one has to expose for 20 or 50 electrons to swamp read noise. Especially important if one has trouble managing the very long exposures needed and has to throw out subs because of it.

Here is an actual sensor analysis from mine, which is the same as OPs camera but just OSC, with HCG mode and low noise mode on:

Actually i remembered a bit wrong, the read noise without the low noise thing is apparently 1e instead of 1,2e. Here is an analysis (Not mine, from a CN thread) with the low noise mode disabled:

I did do this analysis run myself too, but seem to have lost the file/screenshot for it and cant be bothered to do it now. For some reason i thought it was 1,2e, but this claims otherwise. Framerates in sharpcap are abysmal and the analysis takes well over an hour so take whichever number you want to believe, both are low.

With 1e of read noise an exposure swamps it 5x with just 25 electrons, or 100ADUs (above offset, from background). With 0.89e just 80 ADUs above offset is enough.

I am suspicious of the low noise thing though, as when its enabled the framerate drops to exactly half of what it is without the mode. That leads me to think that perhaps there is some kind of internal calibration going on that does some dark current/hot pixel subtraction for every frame which is not really that helpful for astrophotographers that will calibrate their data anyway. That feature would have to turn off with long exposures so that the user is not left with 10 minutes of extra waiting time after a 10 minute sub and internal calibration with the low noise mode. This would not be apparent in sharpcap sensor analysis as all the data it gathers is with quite short exposures so it could very well turn itself off after a certain length of exposure is reached.

Not a camera or software engineer so not going to try and figure out more what is going on there, but i am wary of the low noise thing as long as i dont see a real explanation for it.

Edited August 17, 2022 by ONIKKINEN
typos

[symmetal]

@ONIKKINEN Thanks for your info on the ToupTek version cameras. I wasn't aware they were that different. I'll do the Sharpcap analysis on the 2600.

Can you turn off the HCG mode on your camera at any gain setting then? This seems contrary to what I read about the sensor which I admit were more reviews rather than actual facts. With HCG off, gain below 100 on the Zwo) the read noise is over 3 electrons. The low noise option you have seems to be taking a dark frame for subtraction after each light, like some DSLRs can do, which as you say isn't what you want if you're calibrating anyway.

Even at 0.9 electrons read noise I wouldn't be able to swamp it unless I exposed for over an hour at Ha. What sky background ADU values do you get for your usual narrowband exposures?

Alan

Edited August 18, 2022 by symmetal

[ONIKKINEN]

11 hours ago, symmetal said:

@ONIKKINEN Thanks for your info on the ToupTek version cameras. I wasn't aware they were that different. I'll do the Sharpcap analysis on the 2600.

Can you turn off the HCG mode on your camera at any gain setting then? This seems contrary to what I read about the sensor which I admit were more reviews rather than actual facts. With HCG off, gain below 100 on the Zwo) the read noise is over 3 electrons. The low noise option you have seems to be taking a dark frame for subtraction after each light, like some DSLRs can do, which as you say isn't what you want if you're calibrating anyway.

Even at 0.9 electrons read noise I wouldn't be able to swamp it unless I exposed for over an hour at Ha. What sky background ADU values do you get for your usual narrowband exposures?

Alan

Ah, i have not done any narrowband with mine. But at the darksite of 20.8 sqm i frequent i suspect narrowband would still require at least 30 minutes to swamp x5. I was swamping read noise x5 with a 350nm bandpass UV/IR filter in less than 30 seconds, so a narrowband filter with 100th the bandpass would require 100x longer exposures. So silly length still. It would probably be wise to aim for x2 or if possible x3 read noise swamping with narrowband.

Swamping x3 with 0.9e read noise requires only 7,3e of median background signal. That should be doable, but would probably still be 10 minutes.

[iantaylor2uk]

31 minutes ago, ONIKKINEN said:

Ah, i have not done any narrowband with mine. But at the darksite of 20.8 sqm i frequent i suspect narrowband would still require at least 30 minutes to swamp x5. I was swamping read noise x5 with a 350nm bandpass UV/IR filter in less than 30 seconds, so a narrowband filter with 100th the bandpass would require 100x longer exposures. So silly length still. It would probably be wise to aim for x2 or if possible x3 read noise swamping with narrowband.

Swamping x3 with 0.9e read noise requires only 7,3e of median background signal. That should be doable, but would probably still be 10 minutes.

If you use the spreadsheets that are available for estimating exposure lengths for ZWO cameras, you find that you should go around 3x longer for a filter like the L-Enhance, compared to a simple UV/IR filter. If it is a specific narrowband filter you would need to go longer, that's true.