QE curves imx462m

Use this section to discuss "standard" Baader/Coronado/ Lunt SolarView/ Daystar, etc… filters, cameras and scopes. No mods, just questions/ answers and reviews.
Post Reply
Dennis
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 1668
Joined: Sat Apr 27, 2019 6:44 pm
Location: Germany
Has thanked: 2974 times
Been thanked: 2082 times

QE curves imx462m

Post by Dennis »

Hi,
i have been looking at the QE curves of the imx 462mm chip at the zwo and player one website..
https://astronomy-imaging-camera.com/product/asi462mm/

Can anybody help me make sense of these two graphs, one with the relative and one with the absolute QE numbers. Somehow i struggle to connect those in this case. The curves seem to behave very different from each other, is it just me or is Sonys imx chips QE performance often confusing?

On the first image the relative QE at 400nm is about 64% which should result in 57% absolute QE, right? On the second image it says it is ~87% at 400nm..
Attachments
1.PNG
1.PNG (162.02 KiB) Viewed 2312 times


Triband C9.25
H-a: Quark Chromosphere with BF mod
WL: Antlia 500nm/ 3nm, 393 nm/ 3nm
Ca-K: homebrew (includes 2x 1.5A filters, thanks Apollo)
Player One Apollo Max + Mars MII
Barlows:
-2x Gerd Düring 2.7x
-2x TMB 1.8x
christian viladrich
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 2150
Joined: Sun Jun 14, 2015 4:46 pm
Location: France
Has thanked: 1 time
Been thanked: 2712 times
Contact:

Re: QE curves imx462m

Post by christian viladrich »

Hi Dennis,

Quantum efficiency is about photons and photoelectrons : number of photoelectrons creates in the photosite/ number of photons received by the photosite
QE of 50% means that for 100 photons received by the photosite, 50 electrons are created in the photosite.

Relative response is about energy. Number of photoelectrons created in the photosite for a given amount of radiant energy.
On other words, it takes into account the fact that the blue photons have more energy than red photons.
More here :
https://scientificimaging.com/knowledge ... fficiency/

In any case, relative response is a meaningless for us, and quite confusing too.

It would be much better if Sony provided absolute quantum efficiency curves.


Christian Viladrich
Co-author of "Planetary Astronomy"
http://planetary-astronomy.com/
Editor of "Solar Astronomy"
http://www.astronomiesolaire.com/
Dennis
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 1668
Joined: Sat Apr 27, 2019 6:44 pm
Location: Germany
Has thanked: 2974 times
Been thanked: 2082 times

Re: QE curves imx462m

Post by Dennis »

Thanks very much for the clarification Christian.


Triband C9.25
H-a: Quark Chromosphere with BF mod
WL: Antlia 500nm/ 3nm, 393 nm/ 3nm
Ca-K: homebrew (includes 2x 1.5A filters, thanks Apollo)
Player One Apollo Max + Mars MII
Barlows:
-2x Gerd Düring 2.7x
-2x TMB 1.8x
User avatar
BGazing
Ohhhhhh My!
Ohhhhhh My!
Posts: 197
Joined: Mon Apr 11, 2022 6:53 pm
Has thanked: 289 times
Been thanked: 351 times

Re: QE curves imx462m

Post by BGazing »

Yes - but - what is 'relative response' about? Is it relative QE? If it were, it would distribute similarly to the absolute QE curve...


Lunt 80MT DS, 533/1600/462
christian viladrich
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 2150
Joined: Sun Jun 14, 2015 4:46 pm
Location: France
Has thanked: 1 time
Been thanked: 2712 times
Contact:

Re: QE curves imx462m

Post by christian viladrich »

QE is about how many photons you need to create an electron in the photosite (in fact it is the reverse of this ratio).

Relative response is about how much energy is needed to create an electron in the photosite (in fact it is the reverse of this ratio).

https://scientificimaging.com/knowledge ... fficiency/


Christian Viladrich
Co-author of "Planetary Astronomy"
http://planetary-astronomy.com/
Editor of "Solar Astronomy"
http://www.astronomiesolaire.com/
User avatar
BGazing
Ohhhhhh My!
Ohhhhhh My!
Posts: 197
Joined: Mon Apr 11, 2022 6:53 pm
Has thanked: 289 times
Been thanked: 351 times

Re: QE curves imx462m

Post by BGazing »

christian viladrich wrote: Sun Jul 09, 2023 2:04 pm QE is about how many photons you need to create an electron in the photosite (in fact it is the reverse of this ratio).

Relative response is about how much energy is needed to create an electron in the photosite (in fact it is the reverse of this ratio).

https://scientificimaging.com/knowledge ... fficiency/
Thank you, NOW I have gotten it, I think.
Relative response is, indeed, meaningless and I do not understand why it is published.
462 seems to peak at 500nm on that QE graph. Still decent in IR, very good in blue and probably UV and 80plus at 656nm.
Which is why I found it strange when I asked PO support about exposure time and why, at critical sampling, I get same exposure time in 462 like in 1600. They said 'pixel size', but that should not matter if both are at critical sampling, right?


Lunt 80MT DS, 533/1600/462
christian viladrich
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 2150
Joined: Sun Jun 14, 2015 4:46 pm
Location: France
Has thanked: 1 time
Been thanked: 2712 times
Contact:

Re: QE curves imx462m

Post by christian viladrich »

christian viladrich wrote: Sun Jul 09, 2023 2:04 pm They said 'pixel size', but that should not matter if both are at critical sampling, right?
Yes, you are 100% right !!

BTW, I've check the difference of "sensitivity" between the IMX290 and the IMX 462 in Ca K. Both have 2.9 micron pixels.

Accounting for the difference in full well capacity, the difference of "sensivity" is about 40% in favor of the IMX462. In other words, for the same amount of e- in the phosites, the exposure time is 1.4x shorter with the IMX 462...which is nice :-)


Christian Viladrich
Co-author of "Planetary Astronomy"
http://planetary-astronomy.com/
Editor of "Solar Astronomy"
http://www.astronomiesolaire.com/
User avatar
BGazing
Ohhhhhh My!
Ohhhhhh My!
Posts: 197
Joined: Mon Apr 11, 2022 6:53 pm
Has thanked: 289 times
Been thanked: 351 times

Re: QE curves imx462m

Post by BGazing »

christian viladrich wrote: Mon Jul 10, 2023 12:06 pm
christian viladrich wrote: Sun Jul 09, 2023 2:04 pm They said 'pixel size', but that should not matter if both are at critical sampling, right?
Yes, you are 100% right !!

BTW, I've check the difference of "sensitivity" between the IMX290 and the IMX 462 in Ca K. Both have 2.9 micron pixels.

Accounting for the difference in full well capacity, the difference of "sensivity" is about 40% in favor of the IMX462. In other words, for the same amount of e- in the phosites, the exposure time is 1.4x shorter with the IMX 462...which is nice :-)
Thank you! He repeated that, which is surprising - they are experienced sellers (and quite dilligent and prompt in their replies).
That is why I am a bit confused when it comes to 656nm...1600MM has 20k full well, 462 only 12k, but that should not matter for exposure, right?

He also mentioned something about having to reset tilter plate to 0 for planetary - which was a bit strange. I thought that for such small sensor reasonable amounts of tilt would not matter for solar focus across the full sensor, let alone for planetary.


Lunt 80MT DS, 533/1600/462
christian viladrich
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 2150
Joined: Sun Jun 14, 2015 4:46 pm
Location: France
Has thanked: 1 time
Been thanked: 2712 times
Contact:

Re: QE curves imx462m

Post by christian viladrich »

BGazing wrote: Mon Jul 10, 2023 1:10 pm
That is why I am a bit confused when it comes to 656nm...1600MM has 20k full well, 462 only 12k, but that should not matter for exposure, right?

He also mentioned something about having to reset tilter plate to 0 for planetary - which was a bit strange. I thought that for such small sensor reasonable amounts of tilt would not matter for solar focus across the full sensor, let alone for planetary.
Full well capacity matters, if you compare two sensors with the same QE and FW1 = 20 k while FW2 = 12k, the camera 1 will need 20/12 = 1.7x more exposure to reach saturation.
Another way to look at it : if you keep the same exposure time, the histogram of camera 1 will be at 12/20 = 60% max histogram while camera 2 will be at 100% max histogram.

Max tilt is set by the field of depth (with depends of f-ratio and wavelength). If max tilt is OK pour solar imaging, then it is OK for planetary imaging.
Conversely, if it is not OK for planetary imaging, then it is not for solar imaging ;-)


Christian Viladrich
Co-author of "Planetary Astronomy"
http://planetary-astronomy.com/
Editor of "Solar Astronomy"
http://www.astronomiesolaire.com/
User avatar
BGazing
Ohhhhhh My!
Ohhhhhh My!
Posts: 197
Joined: Mon Apr 11, 2022 6:53 pm
Has thanked: 289 times
Been thanked: 351 times

Re: QE curves imx462m

Post by BGazing »

christian viladrich wrote: Mon Jul 10, 2023 2:03 pm
BGazing wrote: Mon Jul 10, 2023 1:10 pm
That is why I am a bit confused when it comes to 656nm...1600MM has 20k full well, 462 only 12k, but that should not matter for exposure, right?

He also mentioned something about having to reset tilter plate to 0 for planetary - which was a bit strange. I thought that for such small sensor reasonable amounts of tilt would not matter for solar focus across the full sensor, let alone for planetary.
Full well capacity matters, if you compare two sensors with the same QE and FW1 = 20 k while FW2 = 12k, the camera 1 will need 20/12 = 1.7x more exposure to reach saturation.
Another way to look at it : if you keep the same exposure time, the histogram of camera 1 will be at 12/20 = 60% max histogram while camera 2 will be at 100% max histogram.

Max tilt is set by the field of depth (with depends of f-ratio and wavelength). If max tilt is OK pour solar imaging, then it is OK for planetary imaging.
Conversely, if it is not OK for planetary imaging, then it is not for solar imaging ;-)
Thank you!
Yes, I understood that bit about full well, and also that you need to select less frames because of S/N if histo is equal.
But 1600 vs 462, 1600 has histo which is harder to fill, plus 1600 is less sensitive so 462 should have double the advantage...and I do not see that in practice. Really really odd, I cannot explain.


Lunt 80MT DS, 533/1600/462
Dennis
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 1668
Joined: Sat Apr 27, 2019 6:44 pm
Location: Germany
Has thanked: 2974 times
Been thanked: 2082 times

Re: QE curves imx462m

Post by Dennis »

christian viladrich wrote: Mon Jul 10, 2023 2:03 pm
BGazing wrote: Mon Jul 10, 2023 1:10 pm
That is why I am a bit confused when it comes to 656nm...1600MM has 20k full well, 462 only 12k, but that should not matter for exposure, right?

He also mentioned something about having to reset tilter plate to 0 for planetary - which was a bit strange. I thought that for such small sensor reasonable amounts of tilt would not matter for solar focus across the full sensor, let alone for planetary.
Full well capacity matters, if you compare two sensors with the same QE and FW1 = 20 k while FW2 = 12k, the camera 1 will need 20/12 = 1.7x more exposure to reach saturation.
Another way to look at it : if you keep the same exposure time, the histogram of camera 1 will be at 12/20 = 60% max histogram while camera 2 will be at 100% max histogram.
I would assume if both sensors have the same pixel size and exposure time and qe, but different full well capacity:
There would be no difference in s/n and the resulting processed image would be of equal quality if histogram stretched afterward? Or is it of advantage to have the histogram higher right away?


Triband C9.25
H-a: Quark Chromosphere with BF mod
WL: Antlia 500nm/ 3nm, 393 nm/ 3nm
Ca-K: homebrew (includes 2x 1.5A filters, thanks Apollo)
Player One Apollo Max + Mars MII
Barlows:
-2x Gerd Düring 2.7x
-2x TMB 1.8x
christian viladrich
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 2150
Joined: Sun Jun 14, 2015 4:46 pm
Location: France
Has thanked: 1 time
Been thanked: 2712 times
Contact:

Re: QE curves imx462m

Post by christian viladrich »

BGazing wrote: Mon Jul 10, 2023 3:19 pm I would assume if both sensors have the same pixel size and exposure time and qe, but different full well capacity:
There would be no difference in s/n and the resulting processed image would be of equal quality if histogram stretched afterward? Or is it of advantage to have the histogram higher right away?
No. This is not correct.

Let's start again with two cameras having 20 k (for camera 1) and 12 k full well capacities (for camera 2). For sake of simplification, let's assume QE = 100%.

1) The maximum S/N is :
- for camera 1 : sqrt (20000) = 141 or 7.1 bits,
- for camera 2 : sqrt (12000) = 110 or 6.8 bits.
This maximum S/N is reached when the gains of the cameras is at its lowest value, and histogram at 100%. The photosite will have 20 000 e- for camera 1, and 12 000 electrons for camera 2.

In order to reach full capacity of photosite in both camera, you will need 1.7x more exposure in camera 1 compared to camera 2 (since QE = 100% in both camera).

So, in order to have the same S/N in the stacked frames, and if camera 1 is used closed to saturation (= 20 000 e- in photosite), you will need 1.7x more frames in the stack (=20 k/12 k) for camera 2.

2) Now, if we have the same exposure for camera 1 and 2 (and no saturation of camera 2), then the S/N is exactly the same for both cameras since the S/N is the square root of the number of e- in the photosite.

3) The S/N of camera 1 becomes greater than camera 2 only once the exposure time saturates camera 2 (while camera 1 is not yet saturated since its full well capacity is larger).


Christian Viladrich
Co-author of "Planetary Astronomy"
http://planetary-astronomy.com/
Editor of "Solar Astronomy"
http://www.astronomiesolaire.com/
Dennis
Way More Fun to Share It!!
Way More Fun to Share It!!
Posts: 1668
Joined: Sat Apr 27, 2019 6:44 pm
Location: Germany
Has thanked: 2974 times
Been thanked: 2082 times

Re: QE curves imx462m

Post by Dennis »

christian viladrich wrote: Wed Jul 12, 2023 1:16 pm
BGazing wrote: Mon Jul 10, 2023 3:19 pm I would assume if both sensors have the same pixel size and exposure time and qe, but different full well capacity:
There would be no difference in s/n and the resulting processed image would be of equal quality if histogram stretched afterward? Or is it of advantage to have the histogram higher right away?
No. This is not correct.

Let's start again with two cameras having 20 k (for camera 1) and 12 k full well capacities (for camera 2). For sake of simplification, let's assume QE = 100%.

1) The maximum S/N is :
- for camera 1 : sqrt (20000) = 141 or 7.1 bits,
- for camera 2 : sqrt (12000) = 110 or 6.8 bits.
This maximum S/N is reached when the gains of the cameras is at its lowest value, and histogram at 100%. The photosite will have 20 000 e- for camera 1, and 12 000 electrons for camera 2.

In order to reach full capacity of photosite in both camera, you will need 1.7x more exposure in camera 1 compared to camera 2 (since QE = 100% in both camera).

So, in order to have the same S/N in the stacked frames, and if camera 1 is used closed to saturation (= 20 000 e- in photosite), you will need 1.7x more frames in the stack (=20 k/12 k) for camera 2.

2) Now, if we have the same exposure for camera 1 and 2 (and no saturation of camera 2), then the S/N is exactly the same for both cameras since the S/N is the square root of the number of e- in the photosite.

3) The S/N of camera 1 becomes greater than camera 2 only once the exposure time saturates camera 2 (while camera 1 is not yet saturated since its full well capacity is larger).

Thanks for the explanation. Yes, i was thinking in the direcion of case 2, which means once can only profit from higher full well capacity if one increases the exposure time which is not always possible or wanted.

I think this leads to the alternative of either larger pixels to collect photons faster without the need to increase exposure time (imx432). Or the newer back illumination technology of pixels, which increase the photon collection speed (like the imx533).

The s/n of the imx462, despite its high QE remains relatively low, and maybe thats what caused the dissapointment.


Triband C9.25
H-a: Quark Chromosphere with BF mod
WL: Antlia 500nm/ 3nm, 393 nm/ 3nm
Ca-K: homebrew (includes 2x 1.5A filters, thanks Apollo)
Player One Apollo Max + Mars MII
Barlows:
-2x Gerd Düring 2.7x
-2x TMB 1.8x
User avatar
BGazing
Ohhhhhh My!
Ohhhhhh My!
Posts: 197
Joined: Mon Apr 11, 2022 6:53 pm
Has thanked: 289 times
Been thanked: 351 times

Re: QE curves imx462m

Post by BGazing »

Dennis wrote: Wed Jul 12, 2023 2:57 pm The s/n of the imx462, despite its high QE remains relatively low, and maybe thats what caused the dissapointment.
I don't have time to post the pictures, they are not disappointing even in iffy seeing :).
My take from the above is that S/N of the deeper well camera is potentially higher IF you stack the same number of frames of the same histogram fill. Otherwise, smaller well camera can compensate by stacking more - if it is a faster camera, it evens out, probably.


Lunt 80MT DS, 533/1600/462
Post Reply