44 replies to this topic

[Ptarmigan]

It really isn't a 9µm pixel sensor.  They are binning the 4.5µm pixels you find in the IMX420 and IMX428.  Binning in hardware in some manner, there is a suggestion that they are using some CCD architecture so it could be true hardware binning or they may just be using the kind of firmware you find in the IMX071.

 

Effectively it's an IMX428 which is permanently binned 2x2.

 

IMX428

https://www.sony-sem...J_LQJ_Flyer.pdf

 

IMX432

https://www.sony-sem...J_LQJ_Flyer.pdf

 

I sensitivity in mV are different for IMX428 and IMX432.

 

IMX428 Mono

Sensitivity (F8) 1677 mV

Saturation Signal 1001 mV

 

IMX428 Color

Sensitivity (F8) 2058 mV

Saturation Signal 1001 mV

 

IMX432 Mono

Sensitivity (F8) 4050 mV

Saturation Signal 1001 mV

 

IMX432 Color

Sensitivity (F8) 4910 mV

Saturation Signal 1001 mV

Edited by Ptarmigan, 18 August 2019 - 09:36 PM.

[OleCuss]

IMX428

https://www.sony-sem...J_LQJ_Flyer.pdf

 

IMX432

https://www.sony-sem...J_LQJ_Flyer.pdf

 

I sensitivity in mV are different for IMX428 and IMX432.

 

IMX428 Mono

Sensitivity (F8) 1677 mV

Saturation Signal 1001 mV

 

IMX428 Color

Sensitivity (F8) 2058 mV

Saturation Signal 1001 mV

 

IMX432 Mono

Sensitivity (F8) 4050 mV

Saturation Signal 1001 mV

 

IMX432 Color

Sensitivity (F8) 4910 mV

Saturation Signal 1001 mV

Cool.  Doesn't change the fact that the 9µm pixels are actually binned 4.5µm pixels.  Framos made that quite clear.

 

Now understand, I'm not trying to disrespect the sensor.  Especially if the binning is in-hardware in the manner which CCD sensors do this, you could have at least some reduction in read noise.

[ccs_hello]

Without some important info from the full data sheet (such as the difference of LCG and HCG "conversion ratio", description of various "Quad-Bayer Pixel" modes, measurement protocol, etc.), the discussions on SONY's "sensitivity" value is meaningless.

 

SONY's "sensitivity" value is just an engineering value for camera designers.   

It is neither a figure of merit measurement

nor a true reflection on how sensor is photon "sensitive"

 

Case in point:

1) SONY tunes the sensor output by changing/tuning (the ratio) how the charge (the photo-electrons) got converted into voltage level

  <-- you want the highest conversion ratio image sensor,  get one with 0.8 micro pixel pitch

2) It really has to be tuned such that when Analog_Gain" is zero and in LCG mode, the sensor still has a meaningful dynamic range without

resorting to use (a) external grey optical filter or (b) fast shutter speed to avoid satiation too fast.

[Rickster]

I am thinking that a 9um pixel camera hits a sweet spot, the f10 SCT.  9um gives 0.91 arc sec/px at 2000mm focal length, or 1.45 arc sec/px with a 0.63 reducer/corrector.  That would be great when the skies are great.  If the skies aren't so great, then binning 2x2 to get 18um px would be a nice option.

 

I have a little issue with saying that it is more sensitive than existing STARVIS based cameras.  Granted, it will appear to be more sensitive because of it's large pixels.  I suppose it depends on how one likes to think of sensitivity.  I like to think of it on a square micron basis.  https://www.cloudyni...sensors-coming/

 

The price is definitely not in a sweet spot for most buyers.  It might make sense for some, and it might make sense for Mallincam's business model.  That is a personal decision.

 

The IMX432's main marketing claim is that it is a fast sensor with a global shutter, and therefore good at imaging moving objects.  That would be good for bright objects, like planets.  But for planets, most people would probably want smaller pixels.  I would think that an IMX174, which is also has a global shutter,  with smaller pixels, and a lower price, makes more sense for that application.

 

And lastly, comparing a mono sensor against a color sensor in two different scopes leaves me obfuscated.

[OleCuss]

In Jim's defense, he has not claimed that this was a meticulous comparison between similar or even competing cameras.

 

He was out enjoying the evening and got to enjoy both cameras and showed us what he got with one camera hooked up to one rig and what he got with another camera hooked up to another rig.  One can argue that he equalized the systems as much as he could given what he had to work with.

 

He showed that you can do good stuff with both cameras although they are very different.

 

So you can take it as an observing report.  You can take it as a sort of comparison.  I'd not try to take any great insights from the report (Jim doesn't make great claims in this thread) other than he enjoyed two very different cameras/rigs.

 

I think it was a great report!

[barbarosa]

So you can take it as an observing report.  You can take it as a sort of comparison.  I'd not try to take any great insights from the report (Jim doesn't make great claims in this thread) other than he enjoyed two very different cameras/rigs.

Well said.

 

A good report of a session with two dissimilar cameras and scopes and not one of Jim's meticulous evaluations such as his reports on filters. But at first impression it can be read otherwise. Perhaps Jim should put in a disclaimer "Submitted by Jim the professional engineer" vs "Jim's informed but informal observation."

[A. Viegas]

I think we are all very appreciative of Jim's comparison as it gives us a first light report on a new camera suitable for EAA. I think for me a key question remains if the 9um pixels are just binned 2x as has already been speculated. It's a hard pill to swallow with such low resolution in today's market where we are all accustomed to much larger image files... but the mono sensor is very interesting to me...

I look forward to seeing more tests and examples with this camera. Bring it on!

Al

[Rickster]

I would be interested to know how a Skyraider DS432M binned 1x1 (9um) compares with an ASI183MM Pro binned 3x3 (8.4um).

[Ptarmigan]

Cool.  Doesn't change the fact that the 9µm pixels are actually binned 4.5µm pixels.  Framos made that quite clear.

 

Now understand, I'm not trying to disrespect the sensor.  Especially if the binning is in-hardware in the manner which CCD sensors do this, you could have at least some reduction in read noise.

I understand. I would like to have more information about the sensor then what is presented by Sony.

 

It would be better if the pixels were not binned, but as it is.

 

I was thinking of getting a new camera and the IMX432 came up.

Edited by Ptarmigan, 19 August 2019 - 07:12 PM.

[ccs_hello]

Actually, 2x2 charge domain bin (if it turns out to be true on '432) is okayish.  I.e., only one copy of photo-electron readout step.

It has something to do with SONY's pattern of "cookie cutter" sensor design methodology.  Save money on them and more manufacturing stability.

Also, the microlens itself, when the pixel is getting too large, it's much harder to form a lens with the same curvature.

S did it in the past (in one of the Nikon arly DSLRs), so it is once again, pattern repeating.

[Ptarmigan]

Actually, 2x2 charge domain bin (if it turns out to be true on '432) is okayish.  I.e., only one copy of photo-electron readout step.

It has something to do with SONY's pattern of "cookie cutter" sensor design methodology.  Save money on them and more manufacturing stability.

Also, the microlens itself, when the pixel is getting too large, it's much harder to form a lens with the same curvature.

S did it in the past (in one of the Nikon arly DSLRs), so it is once again, pattern repeating.

Economic realities.

[jimthompson]

I would be interested to know how a Skyraider DS432M binned 1x1 (9um) compares with an ASI183MM Pro binned 3x3 (8.4um).

Funny you should ask Rickster... while I was borrowing the DS432M I also started some filter testing (yes I have many balls in the air!).  Clouds rolled in and I wasn't able to complete my testing.  Almost a week afterwards, when I had clear skies again and was able to redo the filter testing, I no longer had the DS432M so I did the test using my ASI183MM Pro binned 3x3.  You can see my images at the links below.

 

filter testing with DS432M  https://flic.kr/s/aHsmFToQxG

 

filter testing with ASI183MM  https://flic.kr/s/aHsmG3pbNb

 

In a nutshell:  even bin 3, and with the images then resized afterwards by 50%, the ASI183 images were much noisier than those from the DS432M for the exact same scope, and same number of stacked frames and exposure per frame.  Thus my statement that for me the DS432M is a better fit for a monochrome deepsky camera.  In terms of a colour camera, I think I can achieve similar quality images to the DS432C with my ASI294 using slightly longer exposures, so I personally don't have a compelling argument for moving to a colour version of the DS432.

 

Cheers,

 

Jim T.

[DSO_Viewer]

Funny you should ask Rickster... while I was borrowing the DS432M I also started some filter testing (yes I have many balls in the air!).  Clouds rolled in and I wasn't able to complete my testing.  Almost a week afterwards, when I had clear skies again and was able to redo the filter testing, I no longer had the DS432M so I did the test using my ASI183MM Pro binned 3x3.  You can see my images at the links below.

 

filter testing with DS432M  https://flic.kr/s/aHsmFToQxG

 

filter testing with ASI183MM  https://flic.kr/s/aHsmG3pbNb

 

In a nutshell:  even bin 3, and with the images then resized afterwards by 50%, the ASI183 images were much noisier than those from the DS432M for the exact same scope, and same number of stacked frames and exposure per frame.  Thus my statement that for me the DS432M is a better fit for a monochrome deepsky camera.  In terms of a colour camera, I think I can achieve similar quality images to the DS432C with my ASI294 using slightly longer exposures, so I personally don't have a compelling argument for moving to a colour version of the DS432.

 

Cheers,

 

Jim T.

Jim, the ZWO 183 MM uses very small pixels and even software binning at 3 x 3 does not gain much as far as signal to noise goes. The benefits shows when using ccd's and hardware binning. The 183 MM is more geared toward fast scopes with shorter focal lengths. Thank you for your tests though which are very informing.

 

Steve

[Rickster]

Jim,

Thanks Jim. 

 

As I am sure you noticed, cross hatch noise is present in some of the 183 images.  I haven't seen that in previously posted 183 images.  Although I have encountered something similar when resizing images.  Do you know the source in this case?  Was it from the camera?

 

I also noticed some less prominent vertical streaking in some of the 432 images, but those look stacking artifacts to me.

 

Note:  I should be able to answer my own question soon, if my 183 ever arrives (it seems that in stock doesn't necessarily mean in stock).

[Alien Observatory]

Jim,

Thanks Jim. 

 

As I am sure you noticed, cross hatch noise is present in some of the 183 images.  I haven't seen that in previously posted 183 images.  Although I have encountered something similar when resizing images.  Do you know the source in this case?  Was it from the camera?

 

I also noticed some less prominent vertical streaking in some of the 432 images, but those look stacking artifacts to me.

 

Note:  I should be able to answer my own question soon, if my 183 ever arrives (it seems that in stock doesn't necessarily mean in stock).

Yes I have seen the cross hatch pattern in the 183 and 178 Mono cams when using Higher Contrast.  I also see it in the Color Cams when used with Zero Color Saturation (sort of a Mono setting) and very high Contrast.  It is from the Sensor, not from processing of the image... Pat Utah

 

https://www.cloudyni...-ds-scout-60mm/

Edited by Alien Observatory, 20 August 2019 - 10:04 PM.

[ccs_hello]

Not sure what's under the so-called bin setup in previous grid pattern discussions...

 

If it is the like-color bin (actually numerical math), then 

in the case of 2x2 like-color, it really is 4x4 calculation:

 

RGRG   --- > R G

GBGB    

RGRG

GBGB    --- >  G B

 

This would have the effect of "jumping over the neighbors" effect, i.e., grid pattern (16 pxls turn into 4)

 

3x3 is even worse (36 pxls turn into 4) ...

 

RGRGRG   --- > R   G

GBGBGB

RGRGRG

GBGBGB

RGRGRG

GBGBGB   --- > G   B

[mclewis1]

Just reinforces my assumption that the combination of a fast scope, a 432m camera and a narrow filter is going to be a killer EAA setup for folks under poor quality skies.

 

And yeah, the comparison folks will now really want to see is a mono 1600 (mn43230) vs mono 432.

 

Thanks Jim.

[Wildetelescope]

First, I appreciate Jim doing careful work once again. Much appreciated.

I'm kind of intrigued by this but not entirely sure why I would want an IMX432 sensor. (Not that anyone has said I should).

It does look to me from a fairly cursory examination as though the IMX432 had faster acquisition but I'd note that with a 98mm objective it is getting substantially more signal than is the camera using the 80mm objective. I do understand why it likely was not possible to use objectives of the same size and I don't know that all of the difference was due to the light-gathering. And, of course, mono versus OSC may explain some of the difference as well but as noted the filtering along with low gain would have some opposite effect.

What I like least about the IMX432 is that I'd expect to be under-sampled even if using one of my 8" SCTs or my 12" Dobsonian. Get into the 10" SCT and up range and under-sampling should not be an issue.

And I'm really rather puzzled as to why the IMX432 with those nice big pixels has only a 12-bit ADC? I've not found the well-depth as of yet but I'm assuming it is not as deep as I'd expect with those big pixels.


So I find it to be an intriguing sensor/camera but I'm not sure exactly where it would shine in terms of performance in OAP/CAP - whatever form of AP might be preferred.

I suppose with a Corrected SCT like a Meade ACF or a Celestron EdgeHD used at native focal length the under-sampling won't be too bad (maybe not even noticeable) if you are using 8" aperture. Greater aperture would mean no under-sampling but at those focal lengths tracking and atmospheric turbulence are going to be sufficiently problematic that under-sampling probably shouldn't be a concern.

And yeah, I kinda like Mallincam's cooling system. I also like it that their monochrome and OSC versions have the same price!

So if someone can help me out on this one. Where would an IMX432 camera be a great choice over most (maybe all) others? (I'm not saying that it isn't a great choice for something, just that I've not figured it out.)

Machine vision applications. High speed imaging of rapid processing etc. that is what Sony designed these chips for. The 294 chip images looked consistently sharper to me.

Jmd

[XS_Man]

Hi all,

 

Many thanks for this very interesting discussion about IMX432 and Mallincam DS432M !
I desperatly search for a measurement in "real life" of read noise in HCG gain on a
camera like DS432M.

No information coming from Sony. Only informations are given by ZWO/QHYYCD and
FLIR for older Prgius generation and 3rd one for the 4.5 micron familly : 

 

https://www.flir.fr/...nsor-review.pdf

 

https://www.qhyccd.c...=94&id=46&cut=1

 

https://astronomy-im...roduct/asi174mm

 

Nothing for the IMX432 sensor.

About Read noise, to summarize quickly the informations available :

- 1st generation- IMX174 and IMX 249 - 5.9 microns pixels -
  Read noise : nearly 6 e- for LCG and nearly 3 e- in HCG

- 2nd génération - IMX250 and others - 3.45 microns pixels -
  Read noise : ?? e- for LCG and nearly  2.5 e- in HCG (FLIR measurement)

- 3rd génération - IMX420 and others - pixels de 4.5 microns -
  Read noise : nearly 6 e- for LCG and nearly 3 e- in HCG (FLIR measurement)

 

Now, what about IMX432 with 9 microns pixels ?
If the Read noise is the same as for 4.5 microns other sensors, nearly 3e- in HCG
mode, this sensor is really a good candidate both for EEA and DSO short exposure DSO !
For higher values, well... I will forget it I guess...

 

The link given by OleCuss to Framos website is very interesting. So it's a binned
version of IMX428. Why not ? IMX294 is a quad bayer and is a really great sensor (I
have a QHY294, best color camera I never have for short exposure DSO imaging).

If dynamic range is 76 dB for classical 4.5 microns sensors and 86 dB for the
binned 2x2 9 microns pixels with 100 ke- full-well capacity, the theorical value
for read noise is :

 

RN (IMX432) =  100000 / 10exp(86/20) = 5 e- 

Logically should be in LCG.  Is  it true ?

Impossible to say if it's good news at first sight... 
Need to be confirmed by measurement on a real camera in LCG and HCG mode.

 

 And since there is a chance the IMX432 is using the same kind of clunky/noisy binning used with the IMX071 sensor (if you mistakenly turn it on), the IMX428 may actually give you a better SNR than does the IMX432!

Remember that IMX071 is an "old" sensor. It's the hope... We really need measurements on existing Mallincam or Touptek or Explore Scientific cameras  to conclude  !

 

About QE value, all these sensors are between 65% and 70% at 530 nm. So I guess it's
exactly the same for IMX432.

 

Concerning a QHYCCD camera using this sensor, yes they are planning it, Qiu has
answered on QHYCCD forum :

https://www.qhyccd.c...hp?topic=7483.0

 

I hope they will post the read noise curve, to have all answers to our questions !

 

Albéric

Edited by XS_Man, 16 November 2019 - 11:08 AM.

[OleCuss]

Note that while the quotation you cited mentions using the same clunky binning as for the IMX071, it is very possible that it is doing the same kind of hardware binning as did/do the CCD cameras for a reduction in read noise which may be significant (if it is a somewhat noisy sensor) or just a little beneficial (if it is a rather low-noise sensor).

 

It could be an excellent sensor in some ways but I don't see it as something I'm likely to want to get.