Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

ASI294MC Pro - Just for EAA or great for AP and fainter DSOs?

  • Please log in to reply
193 replies to this topic

#1 zalt

zalt

    Sputnik

  • -----
  • topic starter
  • Posts: 47
  • Joined: 13 Nov 2012
  • Loc: Oslo/Jotunheimen, Norway

Posted 01 November 2017 - 07:48 AM

Hi, would love to hear opinions!

 

Spec sounded great, ordered and just received this OSC. https://astronomy-im...94mc-pro-color/

A bit quick, as I misread the dark current test chart! Additionally, the product page said "..dark current of ASI294 is a little bit higher than ASI1600 from out test result", but I don't think so. I see now that the ASI294 dark current is not "a little bit" higher but 7.5 times that of the ASI1600.

Sam also responded that the camera is most suited for EAA.

 

I wanted an extra camera as I had no cooled OSC. To be used for various non-mono stuff. Will be a lot more at a dark site in the future.

One purpose is comets, with C11HD + Hyperstar. (I have sort of solved the Hyperstar/mono/filterwheel conundrum even without moving up to C14, using the ASI1600MMC+EFWmini with a carefully designed vignette. But an OSC will of course be needed for fast moving comets.)

 

ZWO has agreed to take back my unused (due to clouds) ASI294, for a purchase of a 071 or 094.

But they have no pickup-service anymore so exporting would really be a hassle.

No return until I'm absolutely sure.

 

I will of course be able to do longer exposures getting an ASI071 (https://astronomy-im...s/asi071mc-pro/), as the dark current of the 295 is 73 times that of the 071!

 

Then there's the read noise. I don't have much experience with these types of (Sony) sensors, so I don't know how much it means that the HCG does not kick in before gain=120. With a lot more readout noise below 120, I guess that even unity gain is undesirable for imaging DSOs, if I should want to have the otherwise high readout noise gone. (nearly 3 times the 071)

But with HCG the readout noise is superb like the ASI1600MMC.

Well depth and sesnitivity is fantastic, but I can't see these properties outweigh the SNR-issues.

Or do I miss something?



#2 ccs_hello

ccs_hello

    Voyager 1

  • *****
  • Posts: 10050
  • Joined: 03 Jul 2004

Posted 01 November 2017 - 07:58 AM

re: With a lot more readout noise below 120

 

Remember that the scale is in e-, not ADU.

Once you factored in the gain (inverse of the "astronomy-gain"), you'll notice the actual noise relative to the full scale is increasing.

 

P.S. ASI071 is 3.3e- and ASI294 is 7.3e- at gain=0


Edited by ccs_hello, 01 November 2017 - 09:13 AM.


#3 einarin

einarin

    Apollo

  • *****
  • Posts: 1462
  • Joined: 23 Dec 2016

Posted 01 November 2017 - 08:09 AM

Almost did like you and ordered 294MC Pro but then I read the thing about dark current and had second thought.

Just wondering how bad is this - what would the max usable exposure be.

I kind of hate the idea of taking hundreds (or thousands) of frames and stacking them.

Would love a moderate size (say medium format with 6-12mpix and 3-5um pixels) camera with low read and dark noise and decent full well capacity.


  • rsr4 likes this

#4 mikefulb

mikefulb

    Surveyor 1

  • *****
  • Posts: 1588
  • Joined: 17 Apr 2006

Posted 01 November 2017 - 11:10 AM

I'm looking at the graphs - the ASI294MC Pro stops at -10C and a dark current of 0.0465 e-/s/pix.  At -10C the ASI1600 is about 0.01 e-/s/pix.  So a factor of ~5 difference.

 

Why the graph for the 294 doesn't go to cooler temps is a mystery to me.  Maybe at room temperature (25C) they could only get it down to -10C (35C differential)?

 

Clearly the ASI071 is superior with around 0.001 @ -10C.

 

I'm curious how much impact this really has however if you are taking exposures close to the minimum optimal time and considering the 294 has a QE advantage over the 1600 and also has larger pixels. 

 

The 294 does not seem like a camera for those shooting in climates where it doesn't get below say 20C at night!


  • Adun likes this

#5 A. Viegas

A. Viegas

    Mercury-Atlas

  • *****
  • Posts: 2964
  • Joined: 05 Mar 2012
  • Loc: New York City/ CT

Posted 01 November 2017 - 01:05 PM

I have both the 294pro and the 1600mono and i used to have a 1600 color.  The 294pro is head and shoulders way better than the 1600 color it is just so much more sensitive.  I would say its sensitivity is roughly equivalent to my 1600 mono in terms of just eyeballing shooting the same target with the same integration time at unity gain with both cameras... which I did over this weekend.  As for dark noise issue...  first note that in terms of read noise that once you get to unity gain the 294pro read noise is the same as that of the 1600 and is very low.   As for dark current, the 1600 color at 0C  has .03125 vs. 0.07 or so for the 294pro which is not a big deal.  I have cooled my 294pro to -25  and shot this weekend with both cameras at -20C without issue.  I do not know why Sam does not show the reduction in dark current for the 294pro below -10C but I believe that at -20C it would be reduced even more.  This is a great camera for EAA and I think it will be a great OSC for astrophotography too!

Here is a single unbayered frame of M33, converted from .fits to .jpg and resized 80% to fit here.  This was taken at almost full moon with a AT65 (420mm) and no filters, no guiding, 30s single exposure from an orange/redzone.

Al

 

Attached Thumbnails

  • M33_30sec_1x1__frame7_fit_resized80%.jpg

Edited by A. Viegas, 01 November 2017 - 01:08 PM.

  • mikefulb, h2ologg, Stargazer3236 and 3 others like this

#6 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 02:53 PM

I'm looking at the graphs - the ASI294MC Pro stops at -10C and a dark current of 0.0465 e-/s/pix.  At -10C the ASI1600 is about 0.01 e-/s/pix.  So a factor of ~5 difference.

 

Why the graph for the 294 doesn't go to cooler temps is a mystery to me.  Maybe at room temperature (25C) they could only get it down to -10C (35C differential)?

 

Clearly the ASI071 is superior with around 0.001 @ -10C.

 

I'm curious how much impact this really has however if you are taking exposures close to the minimum optimal time and considering the 294 has a QE advantage over the 1600 and also has larger pixels. 

 

The 294 does not seem like a camera for those shooting in climates where it doesn't get below say 20C at night!

Dark current grows with time, period. So it doesn't matter how short your exposures are...you'll get the same dark current in the final integration regardless. If you have 0.05e-/s dark current, and get a single 5 hour exposure, that would be 900e- dark current. If instead you used 100x180s exposures, you would have only 9e- dark current per sub, but you have 100 subs, so it's still 900e- in the end. Dark current is what it is, so how many subs you take or how long they are don't really matter.

 

Read noise, on the other hand, is fixed per exposure. The more exposures you take, the more read noise you end up with in the end. Now with cameras that get down to around 1.5e- or less read noise, this compounding is significantly slower than if you have 3, 5, 9e- read noise. You can stack a lot more subs at 1.5e- read noise and have it hardly matter, but you can't stack as many subs with 3e-, and you need very few, very long subs with 9e- read noise. 

 

The ASI294MC has pretty high dark current. If it can only cool -35C, then that would certainly limit its sensitivity (the amount of camera noise along with Q.E. determine the cameras sensitivity). I am not sure why it wouldn't be able to be cooled as much...ZWO seems to have backed off on their coolers a bit lately, with -35C dT rather than -40-45C. I think the deeper cooling is still necessary with many of these sensors. I've been using the ASI183MM Pro lately, and it does better with some moderately longer exposures...rather than 3-4 minutes like the ASI1600, I'm at around 8 minutes with the IMX183, and I'm even going up to 10 minutes now. The sensor has more read noise than the Panasonic, but it's also got more dark current...and I think in part the dark current is due to glows rather than true dark current. I have been cooling to -20C now that it's winter, and may even try pushing it to -25C, but the cooler can only to -35C. I am going to recommend that they try and beef up the cooler, so it can to -40C, so that the camera can do better during warmer months. 

 

The only caveat with dark current is really the glows. In some cameras, the glows may grow faster than the dark current, and may grow non-linearly with exposure time. So longer exposures may result in worse glows than the same integration time with shorter exposures. The IMX183 is definitely this way. I get a bit more glow with 10 minute subs than with 5 minute subs. Thankfully, noise grows more slowly than signal, so while the total glow signal is a bit higher, the difference in noise is barely measurable so far.



#7 Thirteen

Thirteen

    Gemini

  • *****
  • Posts: 3127
  • Joined: 12 Jul 2013
  • Loc: Milford, Michigan

Posted 01 November 2017 - 03:36 PM

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  


Edited by Thirteen, 01 November 2017 - 03:36 PM.


#8 zalt

zalt

    Sputnik

  • -----
  • topic starter
  • Posts: 47
  • Joined: 13 Nov 2012
  • Loc: Oslo/Jotunheimen, Norway

Posted 01 November 2017 - 03:54 PM

Al, thanks a lot for sharing your findings. Thanks Jon, for setting things straight and sharing "cool experiences".

And thanks to Michael and einarin for sharing bewilderment with me, over what might turn out to be an astrophotography camera gem - suffering concealed marketing.

Yes, how bad can the level of dark current really be if dealt with optimally. And why stop testing/reporting at merely -10°C. (ok, I fear that the dark current isn't getting much lower, but not sure)

 

I read somewhere that the 071 has unity gain of 90, and I simply guess that unity gain is 120 (HCG limit) for the 294. (can't install the Ascom driver right now to check)

It's possible to compare those two cameras at unity gain, but I'm not sure that's the most relevant case for AP. Well capacity (if I spot the graph correctly) is around 15k for the ASI294 at gain=120.

 

Al, what was the gain setting for your sample frame? (not bad with full moon!)

Do you have some samples (FIT/raw would be mostly appreciated) posted/postable, at various gain settings?

 

I did not plan to limit the use of this camera to uniquely f/2 setup (Hyperstar). But if I do, and I got my thinking straight, this should seriously limit the dark current issue.

Further, I will almost never image above 0°C, the average night out during a season has its max temp at approximately -10°C.

 

Very interesting that you also noted the "apparent drop" of the ASI's delta-T, Jon.

Both my 1600MMC units seemed to make -40°C dT easily some nights, others not. If not, even if keeping the fan power down quite a bit, it was like the unit overheated when the cooling process were approaching setpoint, and afterwards it wouldn't go anywhere near, instead finding some equilibrium more like -20°C dT. I'm pretty sure the single difference was no drift at all in the air on the nights with cooling failure. Unit too compact and too small fan? Anyway, I now use setpoint -35°C during 95% of the season, and -30°C in early autumn and late spring when temperatures can be slightly above zero. Sam confirmed that 35°C dT was more like it.

 

As for the 294 at my side - inclined to just keep it.

But would still very much like to hear more from you guys, with ASI294 or similar AP-experiences :)



#9 zalt

zalt

    Sputnik

  • -----
  • topic starter
  • Posts: 47
  • Joined: 13 Nov 2012
  • Loc: Oslo/Jotunheimen, Norway

Posted 01 November 2017 - 04:09 PM

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

Should definitely be noted. As a whole, the reported very high sensitivity will outweigh quite a bit of the dark current I hope.

 

If considerably lower sensor temperature actually improves the dark current, and if I dedicate this camera to a very fast imaging system, then the extra dark current specified might turn out negligible. I hope..



#10 dvalid

dvalid

    Ranger 4

  • -----
  • Posts: 322
  • Joined: 02 Apr 2014
  • Loc: Georgia

Posted 01 November 2017 - 04:27 PM

Dark current grows with time, period. So it doesn't matter how short your exposures are...you'll get the same dark current in the final integration regardless. If you have 0.05e-/s dark current, and get a single 5 hour exposure, that would be 900e- dark current. If instead you used 100x180s exposures, you would have only 9e- dark current per sub, but you have 100 subs, so it's still 900e- in the end. Dark current is what it is, so how many subs you take or how long they are don't really matter.

 

 

Jon, is this true for an amp glow as well?

 

David



#11 atacamastar

atacamastar

    Vostok 1

  • *****
  • Posts: 131
  • Joined: 21 Mar 2017

Posted 01 November 2017 - 04:37 PM

For the non-technical AP'ers amongst us would someone please go out on a limb and list the pros and cons of the 1600MC vs the 294MC for straight AP?

 

Many thanks!


Edited by atacamastar, 01 November 2017 - 04:48 PM.

  • Scott Rose likes this

#12 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 05:19 PM

 

Dark current grows with time, period. So it doesn't matter how short your exposures are...you'll get the same dark current in the final integration regardless. If you have 0.05e-/s dark current, and get a single 5 hour exposure, that would be 900e- dark current. If instead you used 100x180s exposures, you would have only 9e- dark current per sub, but you have 100 subs, so it's still 900e- in the end. Dark current is what it is, so how many subs you take or how long they are don't really matter.

 

 

Jon, is this true for an amp glow as well?

 

David

 

Generally speaking. But read my last paragraph, as I address the caveats with glows there. I consider glow to be a distinct form of dark signal, different from dark current, as it could be thermally related but localized, or it could be NIR related (some of these starburst glows definitely don't seem like heat to me, they look like NIR light projected onto the sensor, possibly diffracted through something or reflected off of something, or both). 



#13 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 05:22 PM

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

This depends on image scale. If you are on the same scope, yes, the bigger pixels will gather more signal per pixel. If the image scales are the same, then the bigger pixels would get the same amount of light as smaller pixels, making sensitivity only relative to QE and camera noise. However on the same scope, the total object signal with either camera would be the same. Depends on how you look at it. With the same total object signal, smaller pixels could be averaged down to create bigger output pixels, which would negate most if not all of the benefits larger pixels have (it can depend on other factors as well, but generally speaking.)



#14 ccs_hello

ccs_hello

    Voyager 1

  • *****
  • Posts: 10050
  • Joined: 03 Jul 2004

Posted 01 November 2017 - 05:30 PM

Some astroimager mfgs use their own bag of tricks to stop/slow down the image sensor SoC internal operations such that the amp-glow (technically, it has more contributing factors than sense-amplifier glow) during exposure period is significantly reduced.  Yes, some tricks can indeed be done on CMOS SoC, just not the same way as CCD imaging system.

I don't want to name the firm(s) due to this thread is for a ZWO product



#15 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 05:32 PM

 

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

Should definitely be noted. As a whole, the reported very high sensitivity will outweigh quite a bit of the dark current I hope.

 

If considerably lower sensor temperature actually improves the dark current, and if I dedicate this camera to a very fast imaging system, then the extra dark current specified might turn out negligible. I hope..

 

If we go with peak Q.E., the difference is about 15% in absolute terms. So for 100 photons, the ASI294 would convert 75 of them, the ASI1600 would convert 60 of them. In terms of photon shot noise, the difference there is about 12% in terms of SNR.

 

On the flip side, you have the dark current. If it takes you 20 minutes to get those 100 photons, you would have ~60e- dark current (plus any amp glow, but that would be non-uniform across the field and difficult to fully quantify so lets ignore it for now) with the ASI294, and about 7.2e- dark current with the ASI1600. In terms of noise, that is an additional 7.75e- dark current noise for the ASI294, and an additional 2.7e- for the ASI1600.

 

In terms of SNR, when you account for all the noise terms (except amp glow), assuming Gain 120 on the 294 (definitely the optimal setting for that camera, apparently very close to unity) and Gain 139 for the 1600 (unity):

 

SNR294 = 75/SQRT(75 + 60 + 1.8^2) = 75/SQRT(135 + 3.24) = 75/SQRT(138.24) = 75/11.76 = 6.4:1

SNR1600 = 60/SQRT(60 + 7.2 + 1.55^2) = 60/SQRT(67.2 + 2.4) = 60/SQRT(69.6) = 60/8.34 = 7.2:1

 

Barring other factors, at the very least the two cameras should perform similarly...I wouldn't bet on any huge advantage with the 294 (the larger pixels MIGHT be a benefit, but you can always downsample the ASI1600 data to improve SNR further as well). The ASI1600 might have a slight advantage in fundamental SNR, but it's not big enough to really tip the scales I don't think. 

 

 

EDIT: 

 

Accounting for differences in pixel size, I think that would normalize the dark current of the larger pixels to ~0.033e-/s on a normalized area basis. That probably helps negate some of the difference in SNR as determined by the formulas above. Additionally, due to the higher dynamic range, you could use fewer longer subs with the ASI294, while you would need to use a few subs to get the same DR with the ASI1600. That probably negates the rest of the difference. So the two cameras should perform on par on the same scope, same skies, etc.


Edited by Jon Rista, 01 November 2017 - 07:22 PM.

  • Jared, zalt and ecorm like this

#16 ccs_hello

ccs_hello

    Voyager 1

  • *****
  • Posts: 10050
  • Joined: 03 Jul 2004

Posted 01 November 2017 - 05:39 PM

< edited to clarify it's @ post 8  on unity gain...      Sorry Jon, you and I wrote at the same period of time >

Be careful comparing ASI1600, which is a 12-bit image sensor SoC (MN34230) with a

14-bit ASI294 using SONY Exmor IMX294.

The unity gain is defined by e- / ADU when it equals to one, while 12-bit's ADU can still be filled with a 14-bit sensor's 4 ADU.

Also, trying not to use a sebsor in HCG mode's result to compare with a sensor which is always in LCG mode.

 

When using ZWO's charts for comparison, you have to either scale up ASI1600 by 4 times (which the sensor itself can't do, but purely by math for theoretcial calculation for comparison reasons)

or intentionally scale down ASI294's by 4 time.  I.e., intentionally crippling, which is unfair


Edited by ccs_hello, 01 November 2017 - 06:12 PM.


#17 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 05:56 PM

I'm generally unconcerned about ADU. I care about electrons, not ADU. In terms of electrons, your true signal, I think the math above demonstrates well enough that you shouldn't be expecting some massive improvement with the ASI294. There may well be benefits, it's larger pixels when used on the same scope could be a benefit. The biggest benefit I see with the 294 is it;s dynamic range...its got a whopping 13 stops at Gain 120 (unity?), which is pretty phenomenal. 

 

Also, keep in mind, the ASI294 appears to be only a color camera. Against an ASI1600MC, I think the playing field is probably fairly level, maybe tipped in favor of the 294 due to larger pixels and higher DR at unity. 

 

But in contrast to an ASI1600MM, I don't know that the ASI294 would necessarily come out on top. 



#18 Thirteen

Thirteen

    Gemini

  • *****
  • Posts: 3127
  • Joined: 12 Jul 2013
  • Loc: Milford, Michigan

Posted 01 November 2017 - 06:56 PM

 

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

This depends on image scale. If you are on the same scope, yes, the bigger pixels will gather more signal per pixel. If the image scales are the same, then the bigger pixels would get the same amount of light as smaller pixels, making sensitivity only relative to QE and camera noise. However on the same scope, the total object signal with either camera would be the same. Depends on how you look at it. With the same total object signal, smaller pixels could be averaged down to create bigger output pixels, which would negate most if not all of the benefits larger pixels have (it can depend on other factors as well, but generally speaking.)

 

Assuming the testing the same camera on the same scope, you should include the area as a consideration for dark current noise.  That's all I'm saying.  I'm not sure if you did or didn't in your calculation, as I haven't compared the numbers myself.     The units for dark current are in pixels, so that needs to be converted to area for a fair comparison.  



#19 Jared

Jared

    Fly Me to the Moon

  • *****
  • Posts: 6051
  • Joined: 11 Oct 2005
  • Loc: Piedmont, California, U.S.

Posted 01 November 2017 - 07:16 PM

 

 

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

Should definitely be noted. As a whole, the reported very high sensitivity will outweigh quite a bit of the dark current I hope.

 

If considerably lower sensor temperature actually improves the dark current, and if I dedicate this camera to a very fast imaging system, then the extra dark current specified might turn out negligible. I hope..

 

If we go with peak Q.E., the difference is about 15% in absolute terms. So for 100 photons, the ASI294 would convert 75 of them, the ASI1600 would convert 60 of them. In terms of photon shot noise, the difference there is about 12% in terms of SNR.

 

On the flip side, you have the dark current. If it takes you 20 minutes to get those 100 photons, you would have ~60e- dark current (plus any amp glow, but that would be non-uniform across the field and difficult to fully quantify so lets ignore it for now) with the ASI294, and about 7.2e- dark current with the ASI1600. In terms of noise, that is an additional 7.75e- dark current noise for the ASI294, and an additional 2.7e- for the ASI1600.

 

In terms of SNR, when you account for all the noise terms (except amp glow), assuming Gain 120 on the 294 (definitely the optimal setting for that camera, apparently very close to unity) and Gain 139 for the 1600 (unity):

 

SNR294 = 75/SQRT(75 + 60 + 1.8^2) = 75/SQRT(135 + 3.24) = 75/SQRT(138.24) = 75/11.76 = 6.4:1

SNR1600 = 60/SQRT(60 + 7.2 + 1.55^2) = 60/SQRT(67.2 + 2.4) = 60/SQRT(69.6) = 60/8.34 = 7.2:1

 

Barring other factors, at the very least the two cameras should perform similarly...I wouldn't bet on any huge advantage with the 294 (the larger pixels MIGHT be a benefit, but you can always downsample the ASI1600 data to improve SNR further as well). The ASI1600 might have a slight advantage in fundamental SNR, but it's not big enough to really tip the scales I don't think. 

 

Jon, I think you may have missed one factor in your formulae above... The original poster is using a very fast scope, so I suspect the sub exposure duration in using either camera would be set to just avoid clipping the brightest area of detail.  Since the 294 has a significantly higher full well capacity at a gain of 120 than the 1600 does at a gain of 139, even though the 194 is somewhat more sensitive you could use sub exposures that were roughly 2.5x as long on the 294.  The contribution from read noise is fairly small in both cases, but using your sample above if you had 2.5x as many exposures, that would take you from 7.2:1 to roughly 7:1 on the 1600.  The two should be even a little closer in performance than you indicated.  Throw in the larger pixels of the 294 and I suspect the per pixel SNR would actually be higher in the 294 than in the 1600 (potentially at the cost of some resolution).

 

For the original poster... I wouldn't worry too much about the additional dark current in the ASI294.  Yes, it is quite a bit higher.  But once you account for the greater quantum efficiency and the greater well depth and the longer exposures the well depth allows without clipping, I suspect the differences in real world performance are going to be negligible.  You didn't choose a bad camera. This is going to be a killer choice with the 11" Hyperstar.  That large full well is going to be great with a fast setup like yours.  


  • LauraMS, zalt and DSO_Viewer like this

#20 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 07:17 PM

 

 

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

This depends on image scale. If you are on the same scope, yes, the bigger pixels will gather more signal per pixel. If the image scales are the same, then the bigger pixels would get the same amount of light as smaller pixels, making sensitivity only relative to QE and camera noise. However on the same scope, the total object signal with either camera would be the same. Depends on how you look at it. With the same total object signal, smaller pixels could be averaged down to create bigger output pixels, which would negate most if not all of the benefits larger pixels have (it can depend on other factors as well, but generally speaking.)

 

Assuming the testing the same camera on the same scope, you should include the area as a consideration for dark current noise.  That's all I'm saying.  I'm not sure if you did or didn't in your calculation, as I haven't compared the numbers myself.     The units for dark current are in pixels, so that needs to be converted to area for a fair comparison.  

 

Ah, this is a fair point. I did not do this. Accounting for area differences, the comparable dark current would be around ~0.033e-/s, rather than ~0.05e-/s. An improvement, for sure, but nowhere  near close to the ~0.006-0.008e-/s I measure with the ASI1600. Granted, I image at -20C with the ASI1600...I don't know how deeply the ASI294 could be cooled. If it really does only cool reliably to about -10C, the difference between ~0.033e-/s vs. 0.008e-/s by unit area is still quite large. I think it probabably kills off some of the fundamental advantage the ASI1600 has, but the two should perform pretty much on par with each other.


Edited by Jon Rista, 01 November 2017 - 07:46 PM.


#21 zalt

zalt

    Sputnik

  • -----
  • topic starter
  • Posts: 47
  • Joined: 13 Nov 2012
  • Loc: Oslo/Jotunheimen, Norway

Posted 01 November 2017 - 07:18 PM

Thanks for the SNR calculations, that comparison was exactly what I hoped for!

(never actually paid too much attention to the formulae, should have tried it out)

 

Right on, the great dynamic range even at unity is what triggered me. It's extremely appealing not to have bright stars burn out easily - at least it should be clearly better.

So it was good to see that the ASI294 didn't lag much behind "a legend" in terms of SNR.

 

I didn't really ask for the ASI1600 in this context, but I guess you guys felt that the two of them should be compared!

I was thinking of the ASI071 (and possibly the ASI094) as the alternative.

 

Still, your calculation is relevant for proving that the 294 won't lag much behind SNR-wise.

Can't reproduce the calculation for the 071 as ZWO says SNR is TBD.

Btw ZWO writes for the 294 that "we suppose the QE peak is more than 75%", but specifies SNR TBD, so 75 is still a moderate guess.



#22 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 07:19 PM

 

 

 

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

Should definitely be noted. As a whole, the reported very high sensitivity will outweigh quite a bit of the dark current I hope.

 

If considerably lower sensor temperature actually improves the dark current, and if I dedicate this camera to a very fast imaging system, then the extra dark current specified might turn out negligible. I hope..

 

If we go with peak Q.E., the difference is about 15% in absolute terms. So for 100 photons, the ASI294 would convert 75 of them, the ASI1600 would convert 60 of them. In terms of photon shot noise, the difference there is about 12% in terms of SNR.

 

On the flip side, you have the dark current. If it takes you 20 minutes to get those 100 photons, you would have ~60e- dark current (plus any amp glow, but that would be non-uniform across the field and difficult to fully quantify so lets ignore it for now) with the ASI294, and about 7.2e- dark current with the ASI1600. In terms of noise, that is an additional 7.75e- dark current noise for the ASI294, and an additional 2.7e- for the ASI1600.

 

In terms of SNR, when you account for all the noise terms (except amp glow), assuming Gain 120 on the 294 (definitely the optimal setting for that camera, apparently very close to unity) and Gain 139 for the 1600 (unity):

 

SNR294 = 75/SQRT(75 + 60 + 1.8^2) = 75/SQRT(135 + 3.24) = 75/SQRT(138.24) = 75/11.76 = 6.4:1

SNR1600 = 60/SQRT(60 + 7.2 + 1.55^2) = 60/SQRT(67.2 + 2.4) = 60/SQRT(69.6) = 60/8.34 = 7.2:1

 

Barring other factors, at the very least the two cameras should perform similarly...I wouldn't bet on any huge advantage with the 294 (the larger pixels MIGHT be a benefit, but you can always downsample the ASI1600 data to improve SNR further as well). The ASI1600 might have a slight advantage in fundamental SNR, but it's not big enough to really tip the scales I don't think. 

 

Jon, I think you may have missed one factor in your formulae above... The original poster is using a very fast scope, so I suspect the sub exposure duration in using either camera would be set to just avoid clipping the brightest area of detail.  Since the 294 has a significantly higher full well capacity at a gain of 120 than the 1600 does at a gain of 139, even though the 194 is somewhat more sensitive you could use sub exposures that were roughly 2.5x as long on the 294.  The contribution from read noise is fairly small in both cases, but using your sample above if you had 2.5x as many exposures, that would take you from 7.2:1 to roughly 7:1 on the 1600.  The two should be even a little closer in performance than you indicated.  Throw in the larger pixels of the 294 and I suspect the per pixel SNR would actually be higher in the 294 than in the 1600 (potentially at the cost of some resolution).

I agree. I thought somewhere that I noted that i thought the DR was the ASI294's greatest advantage. It's definitely a big advantage...13 stops of DR with under 2e- read noise is huge. That is half a stop or more higher than even some of the better CCD cameras out there, barring some in the $10k+ range.


  • Jared likes this

#23 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 07:33 PM

Thanks for the SNR calculations, that comparison was exactly what I hoped for!

(never actually paid too much attention to the formulae, should have tried it out)

 

Right on, the great dynamic range even at unity is what triggered me. It's extremely appealing not to have bright stars burn out easily - at least it should be clearly better.

So it was good to see that the ASI294 didn't lag much behind "a legend" in terms of SNR.

 

I didn't really ask for the ASI1600 in this context, but I guess you guys felt that the two of them should be compared!

I was thinking of the ASI071 (and possibly the ASI094) as the alternative.

 

Still, your calculation is relevant for proving that the 294 won't lag much behind SNR-wise.

Can't reproduce the calculation for the 071 as ZWO says SNR is TBD.

Btw ZWO writes for the 294 that "we suppose the QE peak is more than 75%", but specifies SNR TBD, so 75 is still a moderate guess.

I figured the ASI1600 because it was mentioned several times in the OP, and also has the same sensor size. Makes for a good comparison on the same scope. \

 

It's tougher to compare sensors with much larger areas. The ASI094 has it's own distinct advantage due to being a HUGE sensor in the grand scheme of things. You have several options with a sensor that large. You can either image much larger areas of the sky at once without mosaicing, or you can use a longer, larger scope that gathers more light with the same FoV. The latter, on a normalized basis (when the larger 094 image is scaled down to the same dimensions as the 294), the larger sensor would kick the crap out of the smaller one. ;P Same object, same FoV, but you ultimately gather more light in total for the same output area.

 

Alternatively, you could avoid downscaling, and simply render the final image much larger with the 094 than with the 294. If you like to print large, there are few things that beat a big, full frame sensor (and I speak from experience there, I imaged with a rather noisy Canon 5D III for years...most of those images, at least the ones from a dark site, are still my best. As much as I like the ASI1600, it has limited me in some ways, and while narrow band imaging has allowed me to image from my heavily light polluted back yard (a HUGE advantage for mono cameras), I miss those massive FoVs and brilliant full color images the full frame sensor and dark site allowed me to get.)

 

Comparing the ASI294 to the ASI071 is probably a much better comparison. The latter has a slightly larger sensor, but nothing like an APS-H or FF sensor. The characteristics of the 294 are quite impressive put up against the 071. I would be you could do 2x1 mosaics covering a larger FoV with efficiency close to that of the 071, thanks to the huge dynamic range and Q.E. of the 294. 

 

One thing about OSC cameras. I think they are great for dark site imaging. You can get phenomenal results at a dark site even with a very noisy camera, so the higher dark current of the 294 is probably not a huge deal for dark site OSC imaging. OSC imaging in light polluted zones is a different story, IMO. Yes, it's possible, but it is far from ideal. You will generally need to spend many times more effort to get one good image with OSC in LP. You can try to mitigate that cost with LP filters, which are not without caveats. I think when it comes to LP imaging, there isn't really any contest between OSC and mono. Go mono. Mono+LRGB is better, just more efficient in general and getting as many photons as fast as possible is the best way to overcome LP. You also have the option for Mono+NB, or even Mono+LRGB+NB. If you are LP bound, I highly recommend considering mono instead of OSC. (in which case, comparing the ASI1600MM Pro with the ASI294MC Pro makes a lot of sense.) 


  • zalt likes this

#24 Jared

Jared

    Fly Me to the Moon

  • *****
  • Posts: 6051
  • Joined: 11 Oct 2005
  • Loc: Piedmont, California, U.S.

Posted 01 November 2017 - 07:43 PM

 

 

 

 

I think it should also be noted that the pixels on the ASI294 are 50% larger in area when compared to the ASI1600.   This means it will acquire signal faster on a per pixel basis.  So, when looking at the noise side of the equation, you should also consider the signal side.  I know it won't bring the dark current to parity, but it does offset it some.  

Should definitely be noted. As a whole, the reported very high sensitivity will outweigh quite a bit of the dark current I hope.

 

If considerably lower sensor temperature actually improves the dark current, and if I dedicate this camera to a very fast imaging system, then the extra dark current specified might turn out negligible. I hope..

 

If we go with peak Q.E., the difference is about 15% in absolute terms. So for 100 photons, the ASI294 would convert 75 of them, the ASI1600 would convert 60 of them. In terms of photon shot noise, the difference there is about 12% in terms of SNR.

 

On the flip side, you have the dark current. If it takes you 20 minutes to get those 100 photons, you would have ~60e- dark current (plus any amp glow, but that would be non-uniform across the field and difficult to fully quantify so lets ignore it for now) with the ASI294, and about 7.2e- dark current with the ASI1600. In terms of noise, that is an additional 7.75e- dark current noise for the ASI294, and an additional 2.7e- for the ASI1600.

 

In terms of SNR, when you account for all the noise terms (except amp glow), assuming Gain 120 on the 294 (definitely the optimal setting for that camera, apparently very close to unity) and Gain 139 for the 1600 (unity):

 

SNR294 = 75/SQRT(75 + 60 + 1.8^2) = 75/SQRT(135 + 3.24) = 75/SQRT(138.24) = 75/11.76 = 6.4:1

SNR1600 = 60/SQRT(60 + 7.2 + 1.55^2) = 60/SQRT(67.2 + 2.4) = 60/SQRT(69.6) = 60/8.34 = 7.2:1

 

Barring other factors, at the very least the two cameras should perform similarly...I wouldn't bet on any huge advantage with the 294 (the larger pixels MIGHT be a benefit, but you can always downsample the ASI1600 data to improve SNR further as well). The ASI1600 might have a slight advantage in fundamental SNR, but it's not big enough to really tip the scales I don't think. 

 

Jon, I think you may have missed one factor in your formulae above... The original poster is using a very fast scope, so I suspect the sub exposure duration in using either camera would be set to just avoid clipping the brightest area of detail.  Since the 294 has a significantly higher full well capacity at a gain of 120 than the 1600 does at a gain of 139, even though the 194 is somewhat more sensitive you could use sub exposures that were roughly 2.5x as long on the 294.  The contribution from read noise is fairly small in both cases, but using your sample above if you had 2.5x as many exposures, that would take you from 7.2:1 to roughly 7:1 on the 1600.  The two should be even a little closer in performance than you indicated.  Throw in the larger pixels of the 294 and I suspect the per pixel SNR would actually be higher in the 294 than in the 1600 (potentially at the cost of some resolution).

I agree. I thought somewhere that I noted that i thought the DR was the ASI294's greatest advantage. It's definitely a big advantage...13 stops of DR with under 2e- read noise is huge. That is half a stop or more higher than even some of the better CCD cameras out there, barring some in the $10k+ range.

 

Yup.  My 16200 based camera has only 12.3 stops of dynamic range, and it costs a great deal more and has significantly higher read noise.


  • zalt likes this

#25 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 21884
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 November 2017 - 07:48 PM

I agree. I thought somewhere that I noted that i thought the DR was the ASI294's greatest advantage. It's definitely a big advantage...13 stops of DR with under 2e- read noise is huge. That is half a stop or more higher than even some of the better CCD cameras out there, barring some in the $10k+ range.

Yup.  My 16200 based camera has only 12.3 stops of dynamic range, and it costs a great deal more and has significantly higher read noise.


Aye. It's pretty impressive, in general, what some of these CMOS cameras can do. There was a thread in the CCD forum where people were discussing comparisons of large frame OSC CMOS vs. mono CCD, and which would come out on top. With such high DR and low read noise, the OSC CMOS cameras are starting to become much more attractive options than older, higher noise mono CCD cameras. Despite the efficiency of mono+LRGB, I am starting to wonder if one of the IMX094 cameras would still be a better option. I guess if you had dark skies, probably. Might be a tossup in a light polluted zone.
  • zalt likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics