ASI 1600 vs ASI 2600

[Rodd]

This comparison has been beaten to death - I am not resurecting it.  I have a specific question.  I am in a position to upgrade some equipment--moderate cost--so no new mount or OTA.  I am thinking of upgrading my ZWO ASI 1600 camera.  I can't use the 6200 with my 36 mm filters--so that is out.  The 2600, however, on paper is a much better camera (back illuminated, 16 bit, larger sensor, high efficiency).  My question is, am I likely to notice much of a difference shooting with the 2600.  In short--is it worth it? Amp glow does not bother me as it calibrates out.  I tend to collect many subs, so the shortfalls of 12 bit with respect to dynamic range are mitigated.  The 1600 is clean..  It works (the microlensing is a pain--but if that is the only difference I am not sure its wporth it at this time.

How much of a difference will I see.  If its marginal, or if the difference can be made up for by collecting more subs with the ASI 1600. maybe the money is better spent elsewhere--like a new laptop, or whatever.  

Thanks,

rodd

[vlaiv]

Main difference is in two things:

1. speed of capture

2. absence of microlens artifacts

Speed of capture - this comes in two distinct "flavors":

1a - higher QE. We can say that ASI2600 has up to 50% higher QE than ASI1600

In 10 hours with ASI2600 - you'll collect as much signal as in 15h with ASI1600

1b - size of sensor - this is not something that you should really consider because you won't be changing your scopes, but larger sensor is faster sensor because it can be paired with bigger scope.

In fact - this is something that you might find interesting with your C11 - when you bin x2 or x3 - you get small pixel count in final image with ASI1600. Bin x3 will give you image that is about 1600x1200 if I'm not mistaken. With ASI2600 - you'll get larger image with same bin factor

Second point is self explanatory.

You say that speed is not that important - then it is really down to microlens - would you spend your money to get rid of them.

[jjosefsen]

1 hour ago, vlaiv said:

Main difference is in two things:

1. speed of capture

2. absence of microlens artifacts

Speed of capture - this comes in two distinct "flavors":

1a - higher QE. We can say that ASI2600 has up to 50% higher QE than ASI1600

In 10 hours with ASI2600 - you'll collect as much signal as in 15h with ASI1600

1b - size of sensor - this is not something that you should really consider because you won't be changing your scopes, but larger sensor is faster sensor because it can be paired with bigger scope.

In fact - this is something that you might find interesting with your C11 - when you bin x2 or x3 - you get small pixel count in final image with ASI1600. Bin x3 will give you image that is about 1600x1200 if I'm not mistaken. With ASI2600 - you'll get larger image with same bin factor

Second point is self explanatory.

You say that speed is not that important - then it is really down to microlens - would you spend your money to get rid of them.

Isn't the IMX571 in the 2600 also a much "cleaner" sensor than the, at times "finicky" panasonic sensor in the 1600?

I have seen a few raw frame examples, and they look super smooth!! I know there are many factors playing in here..

[Rodd]

2 hours ago, vlaiv said:

Main difference is in two things:

1. speed of capture

2. absence of microlens artifacts

Speed of capture - this comes in two distinct "flavors":

1a - higher QE. We can say that ASI2600 has up to 50% higher QE than ASI1600

In 10 hours with ASI2600 - you'll collect as much signal as in 15h with ASI1600

1b - size of sensor - this is not something that you should really consider because you won't be changing your scopes, but larger sensor is faster sensor because it can be paired with bigger scope.

In fact - this is something that you might find interesting with your C11 - when you bin x2 or x3 - you get small pixel count in final image with ASI1600. Bin x3 will give you image that is about 1600x1200 if I'm not mistaken. With ASI2600 - you'll get larger image with same bin factor

Second point is self explanatory.

You say that speed is not that important - then it is really down to microlens - would you spend your money to get rid of them.

The question is is the data better. Faster is good, but it takes me weeks sometimes months to finish an image. So saving a few hours here and there won’t really be noticed.  When the sky is bad, no amount of data makes a difference.  What I am looking for is better data. The micro lens issue is important to me, as is the larger sensor.  Is the only benefit of higher efficiency, back illumination, and FWC covered by speed alone?  It’s 16 bit and supports hardware binning. Do all of these superlatives only translate to speed?

[Adam J]

45 minutes ago, Rodd said:

The question is is the data better. Faster is good, but it takes me weeks sometimes months to finish an image. So saving a few hours here and there won’t really be noticed.  When the sky is bad, no amount of data makes a difference.  What I am looking for is better data. The micro lens issue is important to me, as is the larger sensor.  Is the only benefit of higher efficiency, back illumination, and FWC covered by speed alone?  It’s 16 bit and supports hardware binning. Do all of these superlatives only translate to speed?

You get what you pay for in this case the ASI1600mm pro is still a good sensor at its price point in my opinion, but of course the 2600 is better and given the price delta you would really hope so. I think you will get better images from it in the end of the day Rob, you just need to decide if you want to pay for them. Also its not just the camera, in many cases unless you already have 36mm filters then you need to get the filters too and wow thats a big outlay. I run AD 1.25inch filters and I cant imaging paying for 36mm AD narroband filters. Worth remembering that with the lottery with some cheaper filters you could wipe out half your gain by getting a lower peak transmisson filter 90% vs 70% = 25% loss in effective QE and thats not unknown.

Edit: this is exactly what I am talking about:

Micro lens problem is a factor here thoug in my opinion and speed will translate into image quality one way or another.

Expensive camera + cheap filter may perform less well than a cheaper camera and expensive filter and you may never know.

Adam

Edited December 22, 2021 by Adam J

[vlaiv]

11 hours ago, jjosefsen said:

Isn't the IMX571 in the 2600 also a much "cleaner" sensor than the, at times "finicky" panasonic sensor in the 1600?

Once you calibrate the data - well, with both cameras you are left with

1) signal

2) noise

Signal is the same in both cameras. If that was different - well, one of the cameras would be seriously flawed

Noise is characterized by two things - magnitude and shape. Magnitude is really not that important here, for "quality of the data" - it relates to speed, or how much time you need to spend to lower magnitude of the noise below certain threshold.

This leaves us with "shape" of the noise, and hopefully that shape is - random.

Panasonic sensor does have a feature that makes noise less than truly random. It has telegraph type noise, but with dithering that is spread around and "drowned" in regular random noise. I haven't seen much other issues with that sensor.

For IMX571 - I simply don't know as I haven't handled one yet, but I do believe noise to random enough.

What I'm saying above - both sensors are good and capable of producing equal quality data given enough time (Panasonic obviously needing more).

 

[vlaiv]

10 hours ago, Rodd said:

What I am looking for is better data.

Not really sure what that means.

Imagine you have instruments that measures height of something. One gives you fast readout (few milliseconds) and is precise down to 1/10th of millimeter in its reading, while other gives you slower readout - you need to wait 0.2s and is precise down to 1/5mm

Will one give you better data then the other?

No - data is what it is.

In image - it depends on the sky, mount, telescope and all that factors - but camera just records what light gets to sensor. It is not responsible for "quality" of that light hitting it.

Both cameras will produce same quality data given enough time (except for micro lensing - which is true artifact of measurement present in Panasonic camera and no in the other).

[Rodd]

1 hour ago, vlaiv said:

Not really sure what that means.

Imagine you have instruments that measures height of something. One gives you fast readout (few milliseconds) and is precise down to 1/10th of millimeter in its reading, while other gives you slower readout - you need to wait 0.2s and is precise down to 1/5mm

Will one give you better data then the other?

No - data is what it is.

In image - it depends on the sky, mount, telescope and all that factors - but camera just records what light gets to sensor. It is not responsible for "quality" of that light hitting it.

Both cameras will produce same quality data given enough time (except for micro lensing - which is true artifact of measurement present in Panasonic camera and no in the other).

Kind of dampens my enthusiasm.  How about dynamic range. 16 bit VS 12 bit, and fwc, those impact data. Wont better numbers here yield better data?

[Rodd]

11 hours ago, Adam J said:

You get what you pay for in this case the ASI1600mm pro is still a good sensor at its price point in my opinion, but of course the 2600 is better and given the price delta you would really hope so. I think you will get better images from it in the end of the day Rob, you just need to decide if you want to pay for them. Also its not just the camera, in many cases unless you already have 36mm filters then you need to get the filters too and wow thats a big outlay. I run AD 1.25inch filters and I cant imaging paying for 36mm AD narroband filters. Worth remembering that with the lottery with some cheaper filters you could wipe out half your gain by getting a lower peak transmisson filter 90% vs 70% = 25% loss in effective QE and thats not unknown.

Edit: this is exactly what I am talking about:

Micro lens problem is a factor here thoug in my opinion and speed will translate into image quality one way or another.

Expensive camera + cheap filter may perform less well than a cheaper camera and expensive filter and you may never know.

Adam

I do shoot with 36 mm filters. Ironically, that is what is preventing me from getting the 6200–the cost of 2” filters!

[Rodd]

2 hours ago, vlaiv said:

ignal is the same in both cameras. If that was different - well, one of the cameras would be seriously flawed

So a 91% efficient camera with back illumination won't produce more signal than a 60% efficient front illuminated camera?  I thought that was the point of it.  

[Adam J]

1 hour ago, Rodd said:

So a 91% efficient camera with back illumination won't produce more signal than a 60% efficient front illuminated camera?  I thought that was the point of it.  

I think vlaiv was being a little pedantic in that the signal is the same but the QE is different. In any case yes it will make a difference as said above you will achieve the same image quality in 2/3 the total exposure. 

[vlaiv]

4 hours ago, Rodd said:

Kind of dampens my enthusiasm.  How about dynamic range. 16 bit VS 12 bit, and fwc, those impact data. Wont better numbers here yield better data?

None of those things matter as all that is controlled by stacking and number of subs and their individual duration.

3 hours ago, Rodd said:

So a 91% efficient camera with back illumination won't produce more signal than a 60% efficient front illuminated camera?  I thought that was the point of it.  

It will produce better SNR - but signal will be the same with both cameras.

Signal is say - 10 photons / second / pixel.

You can't make a number better. You can only determine it with more precision - say 10 +/- 0.1 photons or 10 +/- 0.001.

Second measurement has better precision - or we in imaging say - it has higher SNR.

What 91% vs 60% allows you to do is hit your target SNR in less time. That is what I wrote above - ASI2600 is faster camera, and you said you are not interested in faster camera as you can throw more time into image.

[vlaiv]

1 hour ago, Adam J said:

I think vlaiv was being a little pedantic in that the signal is the same but the QE is different. In any case yes it will make a difference as said above you will achieve the same image quality in 2/3 the total exposure. 

Yes, signal is the same - it is SNR in given amount of time that differs. Want same SNR - then image for more time with camera that has lower QE (or change some other parameters of imaging setup - like aperture size or transparency or whatever ).

[Rodd]

2 hours ago, vlaiv said:

Yes, signal is the same - it is SNR in given amount of time that differs. Want same SNR - then image for more time with camera that has lower QE (or change some other parameters of imaging setup - like aperture size or transparency or whatever ).

Well--efficiency and back illumination can be seperated, no?  Sensors can be 90% efficient using front illumination (I forget the model).  So a camera that is 91% efficient (meaning 91% of the electrons that hit the sensor get counted)  will provide more signal if it is back illuminated as more photons will be counted--so 91% of, say 1,000,000 is greater than 91% of 500,000.

[vlaiv]

1 minute ago, Rodd said:

Well--efficiency and back illumination can be seperated, no?  Sensors can be 90% efficient using front illumination (I forget the model).  So a camera that is 91% efficient (meaning 91% of the electrons that hit the sensor get counted)  will provide more signal if it is back illuminated as more photons will be counted--so 91% of, say 1,000,000 is greater than 91% of 500,000.

No.

QE is quoted in general - regardless of what is done to sensor to make it so. 91% means as you say - number of recorded photon hits.

This relates to photons hitting sensor - regardless if they are detected "in front" or "in the back" of silicon substrate.

You shine 100 photons and out of those 91 will get caught by sensor (on average of course, photon statistics and all).

With back illuminated sensor technology - it is just easier to get to 91% that is all.

[Rodd]

8 minutes ago, vlaiv said:

No.

QE is quoted in general - regardless of what is done to sensor to make it so. 91% means as you say - number of recorded photon hits.

This relates to photons hitting sensor - regardless if they are detected "in front" or "in the back" of silicon substrate.

You shine 100 photons and out of those 91 will get caught by sensor (on average of course, photon statistics and all).

With back illuminated sensor technology - it is just easier to get to 91% that is all.

But not all photons that hit a sensor are counted.  So you can have a front illuminated sensor and another fromnt illuminated sensor with different efficiencies.  So a sensor that is more efficient (for what ever reason other than illumination type) must be bumped up if it then is equipped with front illumination

[vlaiv]

3 minutes ago, Rodd said:

But not all photons that hit a sensor are counted.  So you can have a front illuminated sensor and another fromnt illuminated sensor with different efficiencies.  So a sensor that is more efficient (for what ever reason other than illumination type) must be bumped up if it then is equipped with front illumination

Sensor "illumination type" which is just fancy term for sensor construction does not add additional efficiency.

Complete efficiency - due to whatever feature of sensor is all accounted for in QE.

Back/front illumination is noting special that modifies QE or sensitivity of the sensor - it is just type of construction.

It's like having 6" SCT and 6" Newtonian - and saying - but 6" SCT must have larger aperture because it is SCT. No - aperture size is given for both scopes and is independent of their construction.

Back vs front - just means "order" of elements in sensor itself - where photo diode goes and where metal wiring goes - is it front (facing the front of sensor) or back - at the back of the sensor.

 

[Rodd]

7 minutes ago, vlaiv said:

Back vs front - just means "order" of elements in sensor itself - where photo diode goes and where metal wiring goes - is it front (facing the front of sensor) or back - at the back of the sensor.

I know--more photons will hit the sensor in back illuminated sensor as is evident in your diagram.  My asi 1600 and my STT 8300 are both front illuminated, but the ASI 1600 is more efficient.  There is one front illuminated sensor--can't recall teh camera, that had a very high efficincy.   If that camera was made back illuminated, it would be even more efficient

[vlaiv]

1 minute ago, Rodd said:

 If that camera was made back illuminated, it would be even more efficient

That is probably so, but any BSI/FSI that has been done is already incorporated into QE - so no need to account for it again - just compare QE and that is all you need to know.

[Rodd]

9 minutes ago, vlaiv said:

That is probably so, but any BSI/FSI that has been done is already incorporated into QE - so no need to account for it again - just compare QE and that is all you need to know.

OK--The other think about high efficiency that hasn't really been touched on is when a target is so dim and the sky is poor, that no amount of exposure from my camera will get SNR past a certain point.  For example, I have an Ha stack of NGC 1491 composed of 41 300 sec subs and it is in dire need of more signal.  It is very faint.  I also have a stack of the same target with 225 300 sec subs--taken over thr same general time period, and it looks identical!  1,000 300 sec subs would no doubt appear the same.  The stack is very clean--so I do not think noise is teh issue.  Some targets are just dim.  Would a 91% efficient camera do better in this case?  Would I be able to pick up more of the Ha emissions?

Edited December 22, 2021 by Rodd

[vlaiv]

9 minutes ago, Rodd said:

The other think about high efficiency that hasn't really been touched on is when a target is so dim and the sky is poor, that no amount of exposure from my camera will get SNR past a certain point.

This is simply not correct.

It really does not matter how poor are your skies and how dim is the target.

Whenever you quadruple your imaging time you double your SNR.

If you take single sub and have certain SNR - then to reach any SNR you like - you need to double that first SNR enough times. Simple as that.

What higher QE gives you - is better SNR for that first single sub - but rest of the story is the same no matter how good the camera is - quadruple time to double SNR (or to be precise - for any time you multiply total imaging time - x5 more, x100 more x10000 more - SNR improvement is square root of that - so sqrt(5), sqrt(100) sqrt(10000) - that is why x4 more imaging time gives you x2 SNR improvement as sqrt(4)=2 )

 

[Rodd]

5 minutes ago, vlaiv said:

This is simply not correct.

It really does not matter how poor are your skies and how dim is the target.

Whenever you quadruple your imaging time you double your SNR.

If you take single sub and have certain SNR - then to reach any SNR you like - you need to double that first SNR enough times. Simple as that.

What higher QE gives you - is better SNR for that first single sub - but rest of the story is the same no matter how good the camera is - quadruple time to double SNR (or to be precise - for any time you multiply total imaging time - x5 more, x100 more x10000 more - SNR improvement is square root of that - so sqrt(5), sqrt(100) sqrt(10000) - that is why x4 more imaging time gives you x2 SNR improvement as sqrt(4)=2 )

 

I know the formula--but my 41 sub stack looks the same as my 225 (it was actually 266 becuase I know it was over 22 hours) sub stack--that is 4 hours vs 22 hours and I can see no difference.  After stacking 275 subs, there is not much noise--AND not much signal

[Rodd]

10 minutes ago, vlaiv said:

It really does not matter how poor are your skies and how dim is the target.

I dont agree with this--if the signal is below the sky fog limit, no amount of doubling will increase signal.  This is only true for dim targets that are below teh sky fog threshold

[vlaiv]

2 minutes ago, Rodd said:

After stacking 275 subs, there is not much noise--AND not much signal

Signal can be amplified very simply - just multiply pixel values with a constant and you'll get stronger signal.

Say you have 0.000001 signal - well, multiply that with 1,000,000,000 and you have 1000 signal now - very strong!

 

 

[vlaiv]

Just now, Rodd said:

I dont agree with this--if the signal is below the sky fog limit, no amount of doubling will increase signal.  This is only true for dim targets that are below teh sky fog threshold

What is sky fog limit?

Here is image of SQM 26 signal taken in SQM 18.5 skies - yes, that is 7 magnitudes fainter signal then the sky - yet captured without issues.

Rest of the faint outer parts of galaxy is about SQM24 - so still about 5 magnitudes fainter than sky.

It works for any signal and noise - no matter how bright the sky is - it is down to SNR.