1189 replies to this topic

[Jon Rista]

 

... IMO, the IMX183 is best paired with an f/4 system around 1000mm for optimal image scale to maximize the resolution benefits on small objects. 

With a reducer, my RC would come in roughly at 1333mm @ f/5.3. That sounds like it would be a reasonable match, correct? The FOV would actually be *slightly* larger (mostly wider) than my 8300 on my RC without the reducer. 

 

Having said that, the 1600 is nearly the same size as the 8300, so maybe that would be a better option at this focal length.

 

At that focal length, the ASI1600 might be a better match. Your image scale would be 0.37"/px with the IMX183, which if you have excellent seeing would be awesome, but if you have more around 2" you would end up sampling over 5x. That isn't necessarily a problem, per-se...and in fact, if you sample that well heavier deconvolution should be a breeze. It might hurt exposures a bit, but you can always crank up the gain and reduce read noise a bit as well...so again, not necessarily a problem.

 

If you have better seeing, say 1.5", then the ASI183 would be sampling at 4x (upper end of ideal by my book), would still deconvolve very well, and in no way do smaller pixels ever "hurt" resolution...so you should be able to get some amazing results.

 

The ASI1600 would have an image scale of 0.59"/px, which is just about ideal. You would be sampling 2" seeing at 3.3x, also ideal. So for average seeing, the ASI1600 would probably be a better option. The better your seeing gets beyond 2", though, the more the 2.4 micron pixels of the ASI183 would show their value. 

Edited by Jon Rista, 22 January 2018 - 08:32 PM.

[DaveB]

Thanks Jon. "New England" and "seeing" don't belong in the same sentence unless "bad" or "poor" is also in that sentence...

 

So, I would agree that the 1600 is probably a better fit for me and my setup.

[Jon Rista]

Thanks Jon. "New England" and "seeing" don't belong in the same sentence unless "bad" or "poor" is also in that sentence...

 

So, I would agree that the 1600 is probably a better fit for me and my setup.

Ah yes. If you are in New England area, then the ASI1600. Even that might be a bit overkill if your seeing is closer to 3" most of the time, but again the better sampled you are, the easier it is to deconvolve, so you should be able to recover a decent amount if information even with poorer seeing. 

[StevenBellavia]

Can I ask a question regarding the Sony 178 versus the 183?

 

Isn't the 178 better for DSO imaging, because at 0.5 e/ADU gain, it still has 12 stops DR, and about 1.9e read noise, compared to the 183, which at the same 0.5e/ADU would be 10 stops DR, with about 1.8e read noise?

 

They both have 15,000e FWC, so isn't the 0.5 e/ADU the same gain for both (just divided up less finely)?

 

So you could go longer exposures with the 178, and would need less of them.

 

Yes the 178 is a smaller chip, but I am thinking of using it on a small scope at low focal length, so the field of view is actually not terrible (see FOV simulation, below which is the 178 on an 80mm scope at f/4.8), and it costs less too.

 

Or are the 2 stops DR not worth the loss of FOV?  Just take more frames with the 183 to make up the 2 stops? I am guessing that at a dark sky site, the 178 does better, but with typical sites, with more noise on both, there isn't much difference, for about the same total integration time, after stacking. (?)

Attached Thumbnails

• 
• 
• 

[Jon Rista]

You wouldn't really be able to get much longer exposures. And you certainly wouldn't really need them.

 

The main reason you can't with these cameras is the amp glow, not the bit depth. Without any amp glow, then a long enough exposure at a low gain, even with 12 bits, would bury the quantization error in shot noise. BECAUSE of the amp glows, which accelerate the longer you expose (particularly with Sony cameras), long exposures aren't particularly viable (on most of these CMOS cameras, once you get up to 15 minute subs or longer, the glows are often brighter than the object with narrow band imaging).

 

Another issue here is the FWC. A 15ke- FWC is only at minimum gain. At a conversion ratio of 0.5e-/ADU, you have around 8200e- or so, with about 12 stops of DR. I use gain 53 on the ASI183, which gives a similar FWC. It has more read noise, which brings the DR down to about 11.82 stops. So the ASI178 does have an advantage here....but it is not a large one. 

 

So I don't think the ASI178 would be any more capable of very long exposures. Not at Gain 50 anyway. The ASI178 will get it's full dynamic range, which is over 12.5 stops, at minimum gain. You would need longer exposures at that gain, so you would be pushing the amp glow to the limits...probably with 20 minute subs, which is IMO a less optimal use case than using gain 50. 

 

The FoV with the same pixel size is the big differentiator here, and the key advantage the IMX183 has. The dynamic range is similar between the two cameras at Gain 50...11.8-12 stops. But the IMX183 has a much larger FoV. That makes it more capable for a wider range of objects. And, FTR, with narrow band at Gain 53, I use 10 minute subs (about as long as I care to go, even though I've also used and had success with 15 minute subs). A 10 minute sub isn't "short" by any means. ;P An IMX178 would be able to use around 5 minute subs at Gain 50 thanks to lower quantization noise (and thus lower total read noise), which are moderate length, however there certainly wouldn't be anything stopping you from using 10 minute subs if you wanted to. 

[StevenBellavia]

Thank you for that.

 

Okay, so are you ready for the next dumb question?

(Just say "uncle" when you've had enough):

 

Do all CMOS sensors have amp glow?

 

My Canon EOS SL1, currently my primary imaging camera, (until I eventually pull the trigger and buy an astronomy camera), is very noisy.

I wish I had not got that one.  But I didn't do my research, and saw that it was the smallest and lightest DSLR, and I figured all the Canon CMOS sensors had to be about the same.  Then I sent it out and had it astro-modified.  So I've been kind of stuck with it.

 

It is a CMOS sensor, and below is a typical 8-minute dark frame at ISO 1600, at 10C (it was cold out that night, and the camera has no cooling).  But there is no amp-glow?  Is it so noisy that the amp-glow is buried behind the noise?  On the lower right are the image stats (out of 16bit), with a mean of 8186 (min of 5844).

 

Also attached is a dark frame from an un-cooled QHY178 (that I returned - I didn't like that it didn't provide the sensor temperature).

That is 120 seconds, Gain 15 out of 41, offset 75 (of 1023), and I am guessing around the same temperature (as it was cold that night too).

It has a mean of 2375, (min of 216),  and you can see the amp glow in the upper right.

 

So does my DSLR CMOS have no amp glow?  8 minutes is a long sub for me (I was doing H-alpha with it, to capture the spaghetti nebula, despite it being an OSC)

 

Steve

Attached Thumbnails

• 
• 

Edited by StevenBellavia, 31 January 2018 - 07:28 PM.

[james7ca]

I think all of the surveillance and industrial CMOS cameras have amp glow, it's only some of the sensors that were designed for use in DSLR and mirrorless consumer cameras that have little if any amp glow (but you may even see it there if you look closely enough). The Panasonic sensor used in the ASI1600 and similar QHY163 was used in consumer cameras, but it has a small amount of amp glow. Also, the Sony IMX071 sensor (ZWO ASI071MC and QHY168C) was used in the NEX-5 series of APS-C cameras and I know that it has very little amp glow (based upon results with my NEX-5N and NEX-5R, I expect that the cooled cameras from ZWO and QHY are similar).

Edited by james7ca, 31 January 2018 - 07:42 PM.

[Jon Rista]

No, not all CMOS sensors have amp glow. Most of the DSLR ones either do not, or have such minimal amp glow that it isn't any worse than you get with some CCD cameras that have bias shading.

 

I think it is mostly these CMOS sensors that have an integrated system on chip that have the glow problems. These tiny processors can produce a lot of heat, and usually it is generated in the very short amount of time the sensor is read out (millions of pixels all processed in a fraction of a second.) There also seem to be some NIR emission sources that can cause starburst glows. It may be that the source of such glows could be blocked with a physical box around the sensor, I don't know for sure.

 

It also seems that these glows, to one degree or another, can be managed by programming the sensors with FPGA firmware and drivers, which is how some of the ASI1600 glows and some of the ASI183/QHY183 glows were eliminated or minimized.

[PhysicsStudent]

I meant to post this on Saturday, but I got pretty sick. Anyway first light with the asi183mm-c pro. First narrowband image ever. 20 five minute subs before the clouds rolled in. Astrodon 5nm H-alpha.

Attached Thumbnails

• 

[Jon Rista]

Congrats on the first pic!

 

Just curious...what gain?

[PhysicsStudent]

Congrats on the first pic!

 

Just curious...what gain?

Thanks Jon! Based on reading what you've wrote previously in this thread, I shot at gain 53.

Edited by PhysicsStudent, 01 February 2018 - 09:11 PM.

[Jon Rista]

Gocha. That is with the f/4.9 scope? If you want to use 5 minute subs with that scope, then you might want to consider unity gain, which is Gain 111. I think that would be a better match, with lower read noise. If you want to use Gain 53, you will probably need 10-15 minute subs for best results. 

[PhysicsStudent]

Yes, the f/4.9. This was really just kind of a test, since I had never done narrowband before. If you think that longer integrations will be better at that gain, I will certainly try that. Your knowledge, and willingness to share it on these threads is something that us newbies truly appreciate.

[Jon Rista]

I would either use 5 minutes at Gain 111, or 10 minutes at Gain 53. Up to you to decide which you can handle, and either should work just as well. 

[PhysicsStudent]

What would you say for LRGB as far as the gain is concerned?

[Jon Rista]

Gain 0 or Gain 53. If you have really bright skies, Gain 0. You can actually still benefit from the deeper well for stars, even though you don't get better DR. This is because the well allows longer exposures before clipping the stars, but the DR affects every level in the image (not just the clipping point), so despite the fact that you are packing 15,000e- into 4096 levels, having 15,000e- is still useful for bright skies and LRGB to avoid clipping. It should allow exposures about twice as long as Gain 53. Dithering will become more important at Gain 0, as that is where the horizontal banding is the worst. With enough dithering (every 2-3 frames, unless you are integrating a good deal more than 500), and stacking plenty of frames, horizontal banding will be averaged out.

Edited by Jon Rista, 02 February 2018 - 05:27 PM.

[Astrocava]

So, if I understand, will the 1600  be better for lucky imaging?

[Jon Rista]

So, if I understand, will the 1600  be better for lucky imaging?

Possibly. But I can't say for sure.

 

I have not had enough time with the ASI183MM Pro in a high speed video context to know for sure. When I was testing (I've taken a bit of a hiatus mostly due to weather the last couple of months) SharpCap did not work with the ASI183. It probably works now, so I can give it a try. The ASI183 doesn't get down to quite as low read noise as the ASI1600...1.5e- vs. 1.13e-. It is not really that large of a difference, and if you are thinking more along the lines of resolving double stars or doing SSO imaging, then the difference is not going to matter.

 

I don't quite know exactly how the ROI-based frame rate of the IMX183 compares to the ASI1600. If you can get ~100fps with an ROI with the ASI183, then it should be just as capable as the ASI1600. It has different dark signal characteristics as well, and the ASI183 does not seem to have RTS noise like the ASI1600. So that might be a bonus for the IMX183 based cameras when it comes to lucky imaging. 

 

Anyway...it's an area that needs more testing. 

[evan9162]

You don't have to wonder, ZWO publishes the frame rates for different ROIs for both cameras:

 

ASI1600

10Bit ADC

4656×3520 23fps

3840×2160 36.2fps

1920×1680 57.7fps

1280×960 73.6fps

640×480 125.7fps

320×240 192.4fps

12Bit ADC

4656×3520 14.7fps

3840×2160 23.1fps

1920×1680 32.9fps

1280×960 47.1fps

640×480 80.4fps

320×240 124.4fps

ASI183:

10Bit ADC

5496×3672 19fps

3840×2160 41.04fps

1920×1080 80.10fps

1280×720 117.30fps

640×480 169.92fps

320×240 308.17fps

12bit ADC

5496×3672 19fps

3840×2160 36.12fps

1920×1080 70.48fps

1280×720 103.23fps

640×480 149.53fps

320×240 271.19fps

The 183 is significantly faster than the 1600, especially with 12-bit data.

[Jon Rista]

Well there you go. Sounds like the IMX183 is a pretty good candidate for lucky imaging.

[wz5mm]

I guess I'm a little slow on the uptake.

 

Without reading 17 pages, what is "lucky imaging"?

[james7ca]

I guess I'm a little slow on the uptake.

 

Without reading 17 pages, what is "lucky imaging"?

To help combat the blurring effects of the atmosphere you use a very large number of short exposures (ranging from a VERY few seconds or better yet just small fractions of a second). This is the standard technique used for planetary photography but you can attempt similar techniques with the brighter DSOs. In the latter case this has been termed "lucky tracking" since you don't need guiding or at least not very good guiding and some don't think that the term "lucky imaging" should be used when the exposure times are more than just a fraction of a second.

 

Here is a thread on using such techniques on DSOs:

 

  https://www.cloudyni...s/#entry7430927

 

And here is a link to Emil Kraaikamp's website where you can find more examples of "lucky imaging" (both for planetary and DSOs). Emil is the author of AutoStakkert! which is pretty much the standard tool for selecting and stacking images that have been done with lucky imaging.

 

  http://www.astrokraai.nl/?fromcn=1

Edited by james7ca, 04 February 2018 - 06:27 PM.

[Jon Rista]

High speed imaging with fraction of a second long subs (usually around 10ms or so), which allows you to freeze the speckle pattern of stars or resolve fine planetary details in the moment where seeing gets good. You then cull a majority of the subs, keep only those that have the highest quality characteristics. You stack those high quality subs, and can usually resolve diffraction limited details. The technique works better with good seeing than bad, but it can be used in bad seeing to separate double and multi star systems at the dawes limit of the scope. 

[evan9162]

Well there you go. Sounds like the IMX183 is a pretty good candidate for lucky imaging.

The 183 will be a killer lunar imaging camera.  

[calypsob]

No, not all CMOS sensors have amp glow. Most of the DSLR ones either do not, or have such minimal amp glow that it isn't any worse than you get with some CCD cameras that have bias shading.

 

I think it is mostly these CMOS sensors that have an integrated system on chip that have the glow problems. These tiny processors can produce a lot of heat, and usually it is generated in the very short amount of time the sensor is read out (millions of pixels all processed in a fraction of a second.) There also seem to be some NIR emission sources that can cause starburst glows. It may be that the source of such glows could be blocked with a physical box around the sensor, I don't know for sure.

 

It also seems that these glows, to one degree or another, can be managed by programming the sensors with FPGA firmware and drivers, which is how some of the ASI1600 glows and some of the ASI183/QHY183 glows were eliminated or minimized.

It may be that the source of such glows could be blocked with a physical box around the sensor, I don't know for sure.

 

This would be an interesting 3D print experiment!