ASI 553MC pro?

[Thommy]

Working from home those days makes me reconsider future astro projects😎

The 553mc camera (IMX 533) seems very nice - deep full-well (almost 20,000 at gain 100), reasonable read noise (1,5 e at gain 100) and 14 bit converter. Further, the QE is close to 80%. I'm almost convinced it's time to go back to OSC from mono.  Almost ...

I suppose the referenced QE is not taking the colour layer into account? Have you seen any references on what the "true" QE is? Do any of you know of any plans to incorporate a similar chip but mono into a astro-camera?

What are the experiences so far?

/Thommy 

[Adam J]

1 hour ago, Thommy said:

Working from home those days makes me reconsider future astro projects😎

The 553mc camera (IMX 533) seems very nice - deep full-well (almost 20,000 at gain 100), reasonable read noise (1,5 e at gain 100) and 14 bit converter. Further, the QE is close to 80%. I'm almost convinced it's time to go back to OSC from mono.  Almost ...

I suppose the referenced QE is not taking the colour layer into account? Have you seen any references on what the "true" QE is? Do any of you know of any plans to incorporate a similar chip but mono into a astro-camera?

What are the experiences so far?

/Thommy 

The 80% is just the peak in green it will not be that high across the entire spectrum.

[Thommy]

57 minutes ago, Adam J said:

The 80% is just the peak in green it will not be that high across the entire spectrum.

That's right. They always announce the peak value. I have seen the relative efficiency across the spectrum somewhere, and it looks quite flat, which is also positive. But how much is absorbed by the filters? 10-20%?

[newbie alert]

Square sensor and rectangular pixels seems odd

[Adam J]

1 hour ago, Thommy said:

That's right. They always announce the peak value. I have seen the relative efficiency across the spectrum somewhere, and it looks quite flat, which is also positive. But how much is absorbed by the filters? 10-20%?

I think you are looking at it the wrong way. That QE is just for light of that frequency so that amount of 530 odd nm light that is converted to electrons. But there then there is fall off either side of that peak frequency, so yes the real figure averaged across the green is probably about 10-20% less if you stick to the same band as you would get for a typical set of LRGB filters. 

Red 64%

Green 64%

Blue 53%

But we also have to consider that there are only 25% red pixels 25% blue Pixels and 50% Green pixels. 

So we get the following absolute figures (approximately):

Red = 0.64 x 0.25 = 0.16

Green = 0.64 x 0.5 = 0.32

Blue = 0.53 x 0.25 = 0.1325

Compare that to mono in the form of an ASI183mm pro (Peak =83%):

Red = 0.664

Green = 0.8

Blue = 0.71

Lum = (0.664+0.8+0.71) / 3 = 0.725

But all pixels are active and you can also use luminescence so we get the following absolute efficiencies:

Red = 0.664 x 1 = 0.664

Green = 0.664 x 1 = 0.8

Blue = 0.71 x 1 = 0.71

Lum = 0.725 x 1 = 0.725

Now the OSC does have one advantage, all subs will collect all channels at the same time. However typically Lum will collect date at three times the rate of RGB so lets see how that leaves us over 120 subs assuming a 50/50 split between Lum and RGB for the mono camera. 

Starting with the OSC we get:

Red = 0.16 x 120 = 19.2

Green = 0.32 x 120 = 38.4

Blue = 0.1325 x 120 = 15.9

TOTAL OSC = 73.5

Next for the Mono Camera using LRGB:

Red = 0.664 x 20 = 13.28

Green = 0.8 x 20 = 16

Blue = 0.71 x 20 = 14.2

Lum = 0.725 x 60 x 3 = 130.5

TOTAL Mono (LRGB) = 173.98

Next the Mono without the luminescence filter.

Red = 0.664 x 40 = 26.56

Green = 0.8 x 40 = 32

Blue = 0.725 x 40 = 29

Total for Mono (RGB) = 87.56

 

So in conclusion Mono LRGB vastly outperforms OSC and Mono RGB still outperforms OSC in terms of signal collected .

But this is not the full story, you will note that the filters are better balanced and there is no excess Green in the Mono RGB that you get in the OSC RGB. 

Another thing is that you will get a noticeable increase in resolution with the Mono camera over the OSC camera due to the bayer matrix on the OSC.

 

Hence the answer is if you want best performance for imaging you still need a mono camera. 

Adam 

 

 

 

Edited May 20, 2020 by Adam J

[Adam J]

33 minutes ago, newbie alert said:

Square sensor and rectangular pixels seems odd

They are not rectangular pixels.

[newbie alert]

5 hours ago, Adam J said:

They are not rectangular pixels.

I was looking up camera's for a friend over the weekend, one of them being the 553.. no idea what I was looking at as neither has retangular pixels.. sure I see 3.76x2.4  which  is odd, no idea now what I was looking at..

[Adam J]

1 hour ago, newbie alert said:

I was looking up camera's for a friend over the weekend, one of them being the 553.. no idea what I was looking at as neither has retangular pixels.. sure I see 3.76x2.4  which  is odd, no idea now what I was looking at..

I dont know of any camera with non square pixels. So if you find one let me know :)

[tooth_dr]

1 minute ago, Adam J said:

I dont know of any camera with non square pixels. So if you find one let me know

SX Lodestar.

[Adam J]

10 minutes ago, tooth_dr said:

SX Lodestar.

That must be the V1 then because looking at the v2 those are square.

[tooth_dr]

2 minutes ago, Adam J said:

That must be the V1 then because looking at the v2 those are square.

It's both

[tooth_dr]

At least now you cant say you dont know any cameras with non-square pixels 🤷‍♂️ 

[Adam J]

2 minutes ago, tooth_dr said:

It's both

8.2 x 8.4.....its less rectangular than it is just not perfectly square only 2.4% differnece, but ok.

[tooth_dr]

...

Edited May 20, 2020 by tooth_dr

[Adam J]

29 minutes ago, tooth_dr said:

It's not square.

The interesting thing being that it can only be displayed as if it was square once its on a computer monitor. Incidently I see your image next to mine in the astrobin top pics. Nice.

Edited May 20, 2020 by Adam J

[tooth_dr]

15 minutes ago, Adam J said:

The interesting thing being that it can only be displayed as if it was square once its on a computer monitor. Incidently I see your image next to mine in the astrobin top pics. Nice.

Thanks for that heads up, I'll stop being pedantic 😆

[Thommy]

On 20/05/2020 at 14:28, Adam J said:

Red 64%

Green 64%

Blue 53%

But we also have to consider that there are only 25% red pixels 25% blue Pixels and 50% Green pixels. 

So we get the following absolute figures (approximately):

Red = 0.64 x 0.25 = 0.16

Green = 0.64 x 0.5 = 0.32

Blue = 0.53 x 0.25 = 0.1325

Adam,

Thanks for the comprehensive explanation. My first thought was that the quoted calculation makes sense. Now I'm not so sure.  Take the pixels covered by a red filter, let's call it "red pixels" for now. For every 2x2 pixels there is one red pixel. By multiplying with the percentage of red pixels, you assume that the red light hitting this red pixel is spread over the 2x2 "superpixel", and that the remaining three pixels are not active to red light. Is this a good approximation?  I would say that it is a problem for the resolution, but not for the efficiency. As I understand it, but to be honest this is not my field of expertice, a de-bayer interpolates the values for pixels between the red pixels, but takes into account that blue pixels are a bit sensitive to red light as well.

With regard to your discussion on square pixels, incidently my current camera shares chip with Lodestar 2. Normally, I don't really see this, but some platesolvers are really sensitive to rectangular pixels!

/Thommy 

[Adam J]

3 hours ago, Thommy said:

Adam,

Thanks for the comprehensive explanation. My first thought was that the quoted calculation makes sense. Now I'm not so sure.  Take the pixels covered by a red filter, let's call it "red pixels" for now. For every 2x2 pixels there is one red pixel. By multiplying with the percentage of red pixels, you assume that the red light hitting this red pixel is spread over the 2x2 "superpixel", and that the remaining three pixels are not active to red light. Is this a good approximation?  I would say that it is a problem for the resolution, but not for the efficiency. As I understand it, but to be honest this is not my field of expertice, a de-bayer interpolates the values for pixels between the red pixels, but takes into account that blue pixels are a bit sensitive to red light as well.

With regard to your discussion on square pixels, incidently my current camera shares chip with Lodestar 2. Normally, I don't really see this, but some platesolvers are really sensitive to rectangular pixels!

/Thommy 

I know what you mean, but if you use it in super pixel mode you are taking a 50% resolution hit. The other problem is that I could use the mono 2x2 and in effect I am as the same resolution as the OSC but I have doubled my signal to noise ratio, which you did not do by using the OSC in super pixel mode. I can also play other games with that as I will be able to apply quite a bit of noise reduction before things start looking fuzzy but if I am already sampled at half the rate of the Mono when using OSC noise reduction will not be as effective.

SO what I am saying is that if you are using the OSC in super pixel mode then the direct comparison is not to the mono in 1x1 bin its to the mono in 2x2 bin. If you are not using the OSC in super pixel mode then you have corse noise structure and end up having to apply more noise reduction than in the mono image and you end up back in the same place.

In terms of colour bleed between pixels in OSC due to the less sharp cut off, that is why you will always struggle to get the more vivid differentiated core (yellow) and spiral arms (blue) in galaxy images. Now you can argue that the OSC might be closer to what we would see....but is that better from an artistic standpoint.

Hope that helps.

Adam

Edited May 25, 2020 by Adam J

[Adam J]

3 hours ago, Thommy said:

Adam,

Thanks for the comprehensive explanation. My first thought was that the quoted calculation makes sense. Now I'm not so sure.  Take the pixels covered by a red filter, let's call it "red pixels" for now. For every 2x2 pixels there is one red pixel. By multiplying with the percentage of red pixels, you assume that the red light hitting this red pixel is spread over the 2x2 "superpixel", and that the remaining three pixels are not active to red light. Is this a good approximation?  I would say that it is a problem for the resolution, but not for the efficiency. As I understand it, but to be honest this is not my field of expertice, a de-bayer interpolates the values for pixels between the red pixels, but takes into account that blue pixels are a bit sensitive to red light as well.

With regard to your discussion on square pixels, incidently my current camera shares chip with Lodestar 2. Normally, I don't really see this, but some platesolvers are really sensitive to rectangular pixels!

/Thommy 

Also if you dont like that argument then how about this one. OIII sits exactly in the dip in sensitivity between the Blue and Green Pixels and so you have around a 50% hit to sensitivity when collecting that wavelength with most OSC sensors.

Dont get me wrong, I would still consider a OSC myself for a mobile rig so that I dont have to cart as much kit about and to make it lighter. But I would do that accepting the trade off is significantly less sensitivity.

Adam

Edited May 25, 2020 by Adam J

[Thommy]

I tend to agree with you.

Do anyone know if a new CMOS based  mono camera similar to 553MC (pixelsize, QE, chipsize, full well, read noise and bit depth) is on the rise? Any rumours? 

 

[Adam J]

28 minutes ago, Thommy said:

I tend to agree with you.

Do anyone know if a new CMOS based  mono camera similar to 553MC (pixelsize, QE, chipsize, full well, read noise and bit depth) is on the rise? Any rumours? 

 

Nothing immediatly comparable.

You either get slightly less QE and a slightly larger sensor but a near identical pixel size or you get a slightly smaller pixel lower sensitivity (but still better than osc).

So something like the ASI1600MM Pro, or the Moravian C2-12000a if you want similar pixel size.

or

If you want similar price, the ASI183mm pro.

Personally on balance I would give the Moravian C2-12000a a shot if I did not already own a ASI1600MM Pro. Its expensive but its from a premium brand and should not have such bad microlens patterns (although you will still get something).

https://www.365astronomy.com/moravian-instruments-c2-12000a-monochrome-cmos-astrophoto-camera-with-sony-imx253-sensor-for-12-bit-long-exposure.html

Personally I am not a fan of the 183, for me those pixels are just a little too small.

I think FLO do have an account with Moravian but for some reason I dont understand they dont list them on their site.

Adam

Edited May 26, 2020 by Adam J

[CCD-Freak]

I had an AASI-1600MC-Cool and sold it to get the ASI-533MC-Pro and I like the square sensor as it frames most objects better with less wasted "real estate".   I have only had the 533 out a couple times due to the abysmal weather here.   I chose it for the improved specs and little or no amp glow.   Here is a test image of M3 60 x 10 seconds with no Darks or Flats applied.  The image was made with an Ultima 8 SCT at F10 ( 2000 mm FL )  I really like how the star colors came out.   I wish they would produce a mono version with this sensor.

Looking forward to getting it out to the dark site for a proper test with my better scopes.

John Love
CCD-Freak
WD5IKX

Edited May 27, 2020 by CCD-Freak

[Thommy]

Dear CCD-Freak, or any other owner of a 533 camera - would it be possible to link to some raw files (i.e. non-de-bayered) taken with this camera? I think it could be useful for me to play around with them. 

/Thommy 

[CCD-Freak]

I finally got a chance to image from one of my favorite dark sites (Bortle 2).  While ironing out computer and equipment issues I managed to get some images.

The M17 image is only 16 minutes total !!! (8 x 2m) and NGC281 was a total of 60 minutes (30 x 2m) Gain=100 at -5°C  with an 8" F3.8 imaging Newt.  I am looking forward to my expedition to western New Mexico this fall.  The ASI-533MCP is proving to be a sweet camera.

 

John Love
CCD-Freak
WD5IKX

 

 

 

[2STAR]

I love these conversations even though I dont have a clue what its all about !  🤣

eric