1189 replies to this topic

[dx_ron]

If you watch this part of Robin Glover's talk, you can see that cooling the IMX183 below even 10C makes very little difference in noise. I chose -10C for all my imaging because (so far) the camera has been capable of maintaining that during summer with the cooler running at no more than 50-60% and it is a few degrees colder than the lowest I am likely to be out imaging (it does get colder here, but I'm a wimp).

[bobzeq25]

Bob and Chris,

 

Now you've got me curious about cooling and its affect on images  What would be the reason for not cooling down to as low as you are able to? Of course you want to keep it the same temperature for all your frames, but if you are capable of cooling to say -20C what would be the reason of not cooling it to that? I realize that, especially with these sensors, the improvement in SNR you get going from -10C to -20C is pretty small, but I don't see why you wouldn't still take that small gain.

A few reasons.

 

Mostly is that (for most people in most situations), cooling a modern CMOS camera to -20C makes absolutely no difference to your images, compared to cooling to 0C.  You get _no_ value from it, at all.  We're not talking about a small gain.  _No_ gain at all.

 

The underlying reason is that it's completely buried in other forms of noise.  The rms equation for total noise comes into play here.  Say you have two sources of noise.  Noise A has a value of 1.0.  Noise B a value of .05.   Looks like a 5% increase?  Not even _remotely_ close.  Total noise is:

 

The square root of (1 squared plus .05 squared).  1.001.  The increase is not 5%, it's one tenth of 1 percent.  You'll never see it.

 

All this is well described (with nice graphs) in this educational video.

 

https://www.youtube....h?v=3RH93UvP358

 

Other point.  You never want to cool below the value you can always reach (makes life hard), or a value that takes 90% of your cooling capacity to reach (makes the temperature likely to be erratic).

 

Beginners (and some experienced imagers) routinely overcool modern CMOS cameras, thinking that there'll be some value.  There isn't.

 

People routinely worry too much about the wrong things in DSO AP, and too little about the important things.  Like the value of a _really_ good mount (they start at $5000), well autoguided.  You don't have to spend $5000 on a mount.  But, it's not wasted.  There are gains to be had there.

 

I'm no smarter than you.  But my bookshelf (which has on it both editions of Chris' excellent book), can beat up your bookshelf.  <smile>

Edited by bobzeq25, 28 June 2021 - 01:15 PM.

[joeytroy]

Just watched the whole video and also the gain video that was cut short. Interesting points he was talking about with bringing down the sub exposure to smaller times based on light pollution and noise and based on OSC/Mono. Now time to mess with the calculator http://tools.sharpcap.co.uk/

[Chris W]

A word of caution with very short exposures.  I know that the small read noise of CMOS and the amp glow of long exposures encourage using short exposures. One thing to consider, however, is how much effective exposure you get over a fixed period. When you take the interval between exposures into account (download time + dithering + settle time and any capture overheads) the efficiency, and hence the final image SNR, does deteriorate. With short exposures and high sky background flux , the effect of the read noise is not significant, but at darker sites, it does start to make a more noticeable difference (depending on how short you go).

 

Using Stan Moore's equations and making a little spreadsheet that works out the overall SNR for different cameras, sky fluxes and exposures over a fixed period is illuminating. For my semi-rural location which has moderate LP,  I would typically image NB exposures for 15-20 min and LRGB between 2-5 minutes with the QSI683 (1x1 binning). I have found the sweet spot for my QHY CMOS use shorter exposures, NB around 10 minutes (usually at a moderate gain) and LRGB at 2-5 mins at low to unity gain, depending on the brightest stars. These are not fixed values; I just use the sensor datasheet values for the different gains, the mean value of a dark exposure and a sky background value to work out the sky flux and then I can evaluate alternative exposure plans.

[Keudn]

A few reasons.

 

Mostly is that (for most people in most situations), cooling a modern CMOS camera to -20C makes absolutely no difference to your images, compared to cooling to 0C.  You get _no_ value from it, at all.  We're not talking about a small gain.  _No_ gain at all.

 

The underlying reason is that it's completely buried in other forms of noise.  The rms equation for total noise comes into play here.  Say you have two sources of noise.  Noise A has a value of 1.0.  Noise B a value of .05.   Looks like a 5% increase?  Not even _remotely_ close.  Total noise is:

 

The square root of (1 squared plus .05 squared).  1.001.  The increase is not 5%, it's one tenth of 1 percent.  You'll never see it.

 

All this is well described (with nice graphs) in this educational video.

 

https://www.youtube....h?v=3RH93UvP358

 

Other point.  You never want to cool below the value you can always reach (makes life hard), or a value that takes 90% of your cooling capacity to reach (makes the temperature likely to be erratic).

 

Beginners (and some experienced imagers) routinely overcool modern CMOS cameras, thinking that there'll be some value.  There isn't.

 

People routinely worry too much about the wrong things in DSO AP, and too little about the important things.  Like the value of a _really_ good mount (they start at $5000), well autoguided.  You don't have to spend $5000 on a mount.  But, it's not wasted.  There are gains to be had there.

 

I'm no smarter than you.  But my bookshelf (which has on it both editions of Chris' excellent book), can beat up your bookshelf.  <smile>

All good points Bob! Right now I'm cooling my camera to -10C because where I'm at in the summer its not possible to reach -20C all the time 

[Chris W]

yeah - folks in Canada might just reach +15C right now, if they are lucky

[dx_ron]

Question about the USB Traffic speed setting. INDI defaults it to 30. Does changing it make any difference? There's a qhyccd facebook post about reducing amp glow (on a *different* qhy camera) by setting traffic speed to 0. Does that apply to the 183?

 

If it does make a difference, and I change it - I assume new darks are in order, correct?

[bobzeq25]

FWIW, my experience.

 

I've never changed the default USB traffic setting on my 183s (I have both).

 

Using carefully done bias, flats, and darks, I have never had an amp glow issue.  I use PixInsight to calibrate the lights, do the individual steps instead of using a script.

 

Never.  <smile>

Edited by bobzeq25, 26 July 2021 - 09:55 PM.

[dx_ron]

Thanks. It is unlikely, but possible, that the INDI default value is different from the native QHY value.

 

Anyway, my amp glow calibrates out nicely. I'm just getting to a level of familiarity with the system where every once in a while I notice a parameter I haven't learned about yet and wonder what it actually does.

[SilverLitz]

Though I do not have the experience of Bob, the ASI183mm-Pro has been my main camera for 2+ years (though I now also have a QHY268M, which will be the Big Dog).  I also go with the defaults on traffic; I see that as an option for my 268, but I do not recall seeing it as a user option to change on my 183.

 

I have not had issues with amp glow not calibrating out using PI.  There was a case with residual amp glow when I was using DSS.  This bad case with DSS was when the Master Dark was in a different tab than my lights; e.g. R,G,B in different tabs but with them all having the same exposure, the dark was only in the R tab.  When I had all the calibration masters in the same tab as the lights, DSS calibrated fine.

Edited by SilverLitz, 27 July 2021 - 08:15 AM.

[Keudn]

Hey all quick question,

 

I am heading out to a dark site (I usually shoot from my apartment in town) this next week and the forecast is showing mostly clear skies with maybe some small clouds popping up basically the whole week. I've taken this camera to a dark site one other time, and I found with my LRGB filters I needed to shoot roughly 6-8 minutes to get enough light at unity gain.

 

My question is, if I'm worried about clouds ruining frames when they are 6-8 minutes (plus satellites which I struggled with last time) and not having enough good frames after its all said and done, should I change up my gain? I know Jon likes shooting at 53 gain for dark sites or with narrowband, would it be a good idea to change it up and shoot at 53 for this trip? I haven't messed with changing my gain away from unity yet and I'm a bit nervous about changing it now and potentially making a mistake and ruining the data from this trip because of it. Its as simple as change the setting in software, and then make sure I have calibration frames at the same gain as well, right?

[bigeastro]

Not answering all your questions but I do have one comment.  Don't worry about the satellites.    The rejection routines in pixinsight will resolve the issue and remove the trails.   However, clouds are a problem for sure.

[dx_ron]

  I also go with the defaults on traffic; I see that as an option for my 268, but I do not recall seeing it as a user option to change on my 183.

 

 

Last night turned into a testing night (thanks to a mount issue that I was able to fix this morning). So I played with the USB traffic setting (you can set it in the INDI control panel). The manual says you can lower the value to increase frame rates in video mode, so I was hoping it might reduce the download time for saving images. It doesn't, so I set it back to 30 and won't worry about it any more.

[Derektion]

Hey Jon, I wonder if I can ask for your opinion. Right now I am using an asi294mc pro paired with the William Optics Z73. This combination does produce under sampling from what i have read. I have been looking and changing the camera to the asi2600mm pro. But someone suggested the 183 would be a good choice as well. They both would work, but the specs on the 2600 are much better. Eventually I plan on moving up to possible an esprit 100, or similar. Which camera would you suggest I look at ?

[jonnybravo0311]

While I realize you were soliciting a response from Mr Rista, I'll offer up my opinion in case he does not respond to your query. Besides, my name is Jon, so I can feign ignorance and assume you were asking me

 

My opinion is that people make far too big a deal out of the sampling numbers spit back from a calculator. Yes, the 183 has smaller pixels and according to the calculators it is a great match with the Z73. That's only a single factor in the "what camera should I purchase" decision making process. The 2600MM Pro is a superior product in every way. Will it lead to under-sampling? Sure... and if that's all you care about, then get the 183.

 

Now, I posted this in another thread, but I feel it relevant here. I image with the 294MM Pro. It has its default mode and its unlocked mode. With my setup, that's a 2.5"/px image scale in the "default" mode and a 1.25"/px scale in the "unlocked" mode. I regularly image with the default mode, which is under-sampled. If I zoom way in, then I can see the stars a blocky. However, when I'm looking at an image sized to fit a screen (my 27" monitor, my iPad, my iPhone, whatever)... well... it becomes extremely difficult to discern that level of detail. Example images... First one of the Veil complex:

 

 

100% crop of Pickering's Triangle from that image:

 

 

Here's one of the North America and Pelican:

 

 

The Veil is under-sampled at 2.5"/px. The North America and Pelican is well sampled at 1.25"/px. If I hadn't written this, would you have been able to tell from just looking?

 

My point is to stop worrying about what a calculator says. Plenty of fantastic images have been shot in what calculators state are quite under-sampled scale. Look instead at the system as a whole and make your decision based on all factors, not just a single one of the many