Sub exposure tables for ASI071 and possibly QHY168C

[Shiraz]

hi

the new CMOS cameras allow inversegain/read noise to be varied, which means that there is no single measure of the most efficient sub length (here "efficient" means that the best possible image resolution and SNR will be obtained in the overall shortest possible time) - what sorts of sub lengths make sense then?.

The following tables have been prepared for two colour CMOS cameras using a fairly detailed astrograph model to suggest some starting sub lengths for a variety of environmental conditions and camera gain settings. To use, measure or estimate your sky brightness (eg use an SQM or a sky brightness atlas) and choose the table that best represents your sky. Then look at the column that best describes your scope Fnumber. Look down that column and choose the sub length that corresponds to your chosen gain setting for the camera - use lower gain for best dynamic range and higher gain if you want short subs.

There is also a column that shows the expected ADU value above bias for your gain setting. When you take a sub under the chosen sky and with the chosen gain, you should see an ADU value in the background sky that is the sum of this expected number plus the average camera bias value. If you measure a background sky value much higher that the expected value+bias, reduce your sub length and vice versa.

working through an example, if your sky brightness is 19, and you choose a gain of 150(asi) with an f8 system, the suggested sub length is 50 seconds. The background sky should be 450ADU above bias, so, if your camera produces bias images with an average value of (say) 350ADU, then you should see an ADU value of 800 (= 450 + 350) in the clear background sky regions of your lights. If you want to use gain or Fnumber values between those in the tables, guessing on the basis of nearby values should be good enough.

The published data for the cameras is a bit sparse, so some rather heroic assumptions had to be made on things such as quantum efficiency. However, the suggested sub lengths should not be too far wrong and will give a starting point that can be refined. The data were calculated for the ASI071, but equivalent gain points for the QHY168C were estimated by cross referencing (they are in grey). QHY claim slightly lower read noise at an equivalent gain than ASI, so shorter subs may be possible with that camera - as a starting point though, the suggestions could be OK for both cameras.

I do not have either camera, so cannot test the predictions. If you use the tables, I would be very grateful for any feedback on how effective the suggested sub lengths are. also, if you find any obvious errors, please say.

thanks for looking. cheers Ray

[glend]

Thanks for this Ray. I will check these figures against your ASI1600 chart for comparative purposes. I keep your original 1600mm-c chart laminated and on the wall in my observatory.

[kens]

Nice work Ray. Sorry for being a pedant but the column labelled ZWO presumably should be QHY. ZWO is the company that makes the ASI range of camers.

[Shiraz]

thanks very much Ken - fixed

note to self - don't post anything at 1.30 am :)

[Shiraz]

thanks very much Glen. This chip has a less dramatic range of read noise than the 1600 and it is also colour, so has much lower broadband QE than the 1600. Was a bit hesitant to post this stuff, since I do not have a camera to test the predictions or measure the effective QE etc, but the model is good and have been as careful as I can with assumptions, so hopefully it is still useful. Cheers

[ChrisV]

Thanks, this is greatly appreciated but will take a while to absorb. Like Glen, I might have to laminate, frame it and put it up on the wall!
Where can I read about how this is calculated? Sounds like a good bedtime read.

[Shiraz]

You can download and play with the model itself Chris - it has notes that explain what the various bits do. http://www.iceinspace.com.au/forum/showthread.php?t=151000

the only problem is that your spreadsheet software will try to protect you from what it regards as unknown software - I cannot do anything about that except to assure that it is a locked Excel spreadsheet prepared on an up-to-date Win10 machine. If you don't want to risk it, here is a screen image of the thing in operation - it shows the structure and most comments without needing to run the software itself http://www.astrobin.com/full/308223/0/?nc=Shiraz&real=&mod=

You have an 071, could I please ask what sort of background sky values you get in your images? cheers

[ChrisV]

I got these values on an 8" F5 newt with the ASI071 at unity gain (90) and offset of 20 running at -5C.

I've gone through a few images and the better ones (with darker backgrounds) have a background sky value of about 1300 for 6min subs. I think I worked it our properly - opened the fits file in fitswork and read off the minimum value on the histogram.

At this gain/offset - the offset/bias frame had a mean value of ~80 ADU (+/- 10).

[glend]

Chris, just my opinion, but six minute subs seem excessive for a OSC camera shooting in a city, even with a LP filter. Narrowband would be fine provided your not over-exposing. What sort of sky values are you seeing at 60 sec?

[Shiraz]

thanks that is great. At gain 90, the expected ADU is 290, so you could get down to that by reducing the exposures by a factor of 4 (to ~1.5 minutes) if you wanted, without doing any damage to the SNR. That would give better dynamic range and possibly better resolution.

sorry to be a pest, but what is your Bortle scale sky brightness? https://en.wikipedia.org/wiki/Bortle_scale.

thanks again cheers Ray

[ChrisV]

Neve done short subs with low gain. I've only done 30-60sec subs with high gain (250-300) for single shot live(ish) viewing.

I was always worried that I'd lose dim info. I'll try tonight !

[ChrisV]

Ray, you are far from being a pest. This is interesting and informative!
I just did some unity gain subs. Background sky is
~300ADU @ 60sec exposure
~600ADU @ 120sec
~1200ADU @ 240sec

I'm guessing my sky Bortle scale is about 5 by that description. https://www.lightpollutionmap.info says I'm a yellow zone, SQM 20.4-20.9. And we have a few nasty sodium lights in front of our house, although the backyard is shaded from it. I'd like to take them out!

[Shiraz]

Excellent Chris.

From your Bortle estimation and the LP map, your sky is probably a little darker than mag20. That means that, if the table assumptions are OK, you should get a bit less sky signal than would be predicted by the mag20 table.

Now the mag20 table, at gain 90 and f5, predicts that you should get to 370ADU (290ADU + bias) in 55 seconds. When you did your test, you got to 300ADU in 60 seconds (assume that includes bias?) - which is a bit less than the sky signal predicted by the mag20 table.

ie, your sky should have produced a little less signal than that predicted by the mag20 table - and that is what you found when you measured it. It looks like the assumptions were not too bad and the tabular data is at least in the right ballpark :thumbsup: Thanks very much for doing that testing.

So, can somewhat more confidently suggest that you could drop your subs to around 1- 1.5 minutes if you wish, without losing any SNR/sensitivity - you will gain extra dynamic range and possibly better resolution by doing so. Or you could continue to use longer subs if you want a lower image processing load and are happy with the dynamic range and resolution - you could consider gain 0 for long subs.

very grateful. cheers Ray

[glend]

Chris did you use that global light pollution map to zoom right down to your street to get the 2017 data on your location. It displays in the traditional lp colours and you can get other info as well. Unless you live on the fringes of Sydney, you are likely into yellow or orange range. Check here:

https://www.lightpollutionmap.info/#zoom=11&lat=-3989126&lon=16825660&layers=B0TFFFFFFT

You can drag the map around to your area, expand or contract as required. Perhaps you have check this already. The problem with sodium lights is the sky glow contribution, however they can be pretty effectively filtered out by a good light pollution filter. The new STC Multispectrum filter seems to be pretty good,

https://www.cyclopsoptics.com/filters/stc-astro-multispectra-filter/

Check out the spectrum bandpass chart by scrolling down. BTW HKD converts to $264 AUD, but there are other lower cost filters.

[ChrisV]

Thanks Ray. So given those parameters what should I be using for an offset - so far I've had offset at about 80-90 ADU. Is that enough ?

I might try a bit of helix Neb tonight!! Or maybe redo an object I've recently done for comparison.

All this is starting to sink into my thick skull.

[ChrisV]

Glen. If I click where I live its says Bortle 5 SQM 19.72 (using ATLAS2015), or yellow 5.23 (using VIIRS2017). But the values change a lot if I go up or down my street - the park at the northern end is 20.10, the station at the other end is 19.65. I've got a 2" UHC filter that I haven't used for a while - might try it out again.

[Shiraz]

that's even better for checking the model assumptions.

The tables are based on generic scope assumptions that are close enough for most purposes, but re-did at gain 90 for your Newtonian scope and a sky of 19.72. the model says that, with bias, you should see 361 ADU in 1 minute. You recorded ~ 300 ADU in 1 minute. If your figure includes the bias, then the model is overestimating by 20% or, if your figure does not include bias, then the model is underestimating by <10%. Either way, in view of the wide range of possible unknowns, this is a good enough indication that the model assumptions are doing a reasonable job of capturing the behaviour of the camera.

You could confidently drop your gain90 exposures to 1 minute if you wanted to. The offset is required to stop read noise driving the signal below zero - an offset of 20 (= 80ADU) should do that for all reasonable camera settings.

interesting and good fun. thanks again. cheers ray

[ChrisV]

Ray

I've started on the Helix Neb following your advice. So far I've got 90min of 1.5min subs. Its starting to look good, actually better than the 3hr of 6min subs I did a week ago.

So thanks for the advice about sub lengths. I'll get some more on it next month before putting it up for public comment ...

Looks like Helix needs a lot of data to bring any detail - well at least in my hands it does.

Chris

[Shiraz]

phew. that's good to know. yes, it's big but dim - lovely object, but very few photons.

looking forward to your image. regards Ray

[h0ughy]

thanks for ChrisV pointing me to this thread. i just got a ASI071 and together with a brand new RASA 11" i am looking forward to some photon destruction. i have a hell of a lot to learn, and learn to drive a ZWO ASI071

[Shiraz]

looking forward to some images David. With such a fast scope, suggest setting the gain at 0 and taking subs at around a minute (or less depending on sky). No point in using higher gain unless you are aiming for some very short-sub lucky imaging. Cheers Ray

[h0ughy]

Might do some unguided subs and test it

[h0ughy]

http://www.iceinspace.com.au/forum/showthread.php?t=166787 took your advice - one minute exposures x 5 gain 0

[Shiraz]

thanks very much for the info Dave. gosh that looks amazing for such a short total exposure. Looking forward to you getting well and posting something taken over an hour or two. Cheers Ray

[h0ughy]

OK well using your table as a guide it's a sixty seconds shot which in poor sky conditions would be equivalent to a five minutes shot at gain 0. So therefore 5min worth would be the same as 30min at f7 roughly or am I not in the ballpark :shrug:

[Shiraz]

Not quite sure that I understand the question Dave, but for certain, your scope is definitely very fast.

Take for example the mag 21 table at gain 0. Your scope will need a bit over 35 seconds and an f7 scope will need 7 minutes per sub to get to the same result. At f2.2, your scope puts photons into the pixels at about 10x the rate of an f7 scope.