CCD running out of time

[Peter Ward]

With OnSemi announcing the wind-up of their CCD foundry, those who have been thinking about a large format CCD camera might want to get a wriggle on.

Last orders look to be March 2020....

I just placed an order for a SBIG STX16200 to supplement my STX16803.
If this does not bring on the rain...I don't know what will

[Xeteth (David)]

I'm so close to pulling the trigger on the QHY16200A... like real close. Unless anyone can suggest an alternative?

[Peter Ward]

Originally Posted by Xeteth

I'm so close to pulling the trigger on the QHY16200A... like real close. Unless anyone can suggest an alternative?

Very droll.... But I did wake to the sound to rain this morning.

[Atmos (Colin)]

My heart is more set on the ASI6200MM/QHY600M at the moment but I did long look at the KAF-16200 sensor.

[Peter Ward]

Originally Posted by Atmos

My heart is more set on the ASI6200MM/QHY600M at the moment but I did long look at the KAF-16200 sensor.

Yep, I looked at the QHY600M as well. But having tinkered with the QHY183
I came to the conclusion high image homogeneity is not there yet with CMOS.

This is not to say it won't happen, but I figured with both Sony and now OnSemi ending CCD production, the remaining players such as E2V will simply be too expensive for amateur applications...

[Atmos (Colin)]

Originally Posted by Peter Ward

Yep, I looked at the QHY600M as well. But having tinkered with the QHY183
I came to the conclusion high image homogeneity is not there yet with CMOS.

This is not to say it won't happen, but I figured with both Sony and now OnSemi ending CCD production, the remaining players such as E2V will simply be too expensive for amateur applications...

You’re referring to amp glow? Some of the CMOS chips suffer from it pretty bad like the IMX183. My understanding is that the higher end sensors made for the DSLR market and not security/video cameras don’t suffer the bright star bursts.

It is nice to finally have some larger mono CMOS sensors but damn those pixels are tiny

[gregbradley]

The QHY600 is an interesting camera but do you really want to process a stack of 24 x 10min 102mb files?

CCDStack demands lots of memory. With 32mb files from the 16803 sensor and 64gb RAM I think it still won't do more than about 30 files at once and it takes a while.

Super high QE, full frame size, the ability to bin 2x2 and still have a large amount of mb and no amp glow as reported by Martin though are appealing.

The 42.4mb Sony backside illuminated full frame sensor would have been a nice step for the astro market. Its the sensor used in the A7r2 and A7r3 Sony mirrorless cameras.

The Sony A7riv that uses that same 61mp sensor has 1 stop worse noise performance at high ISO than the 42.4mp A7r3 sensor.

Sony has some nice full frame sensors not yet released so expect more.

The Sony A7Siii when it gets released some time this year could be a nice sensor if they release it to QHY and others.

Greg.

[lazjen (Chris)]

Greg, try using something like Pixinsight - should have no problems stacking that number of files of that size.

[Peter Ward]

Originally Posted by lazjen

Greg, try using something like Pixinsight - should have no problems stacking that number of files of that size.

I use PixInsight frequently these days. Most recent was a test 40 frame registration and stack from a Canon EOS-Ra. It took a good half hour....hardly a cracking pace!

I agree with Richard Wright at Sky & Telescope. For scientific work, due their non-linearity, pixel to pixel non-uniformity and amplifier glow CMOS rate a poor second to CCD's.

I suspect the demise of CCD's will be lamented by the amateur scientific community....as well as cash-strapped tertiary institution Astronomy programs..

For imaging, CMOS data can be cleaned up without much fuss...and the results to be fair can be very good. I'll no doubt acquire a shiny new sCMOS camera in time...but not just yet.

[lazjen (Chris)]

What are the size of those files - 30 Mb each? If so, may I suggest you look at upgrading your computer? If you're using Windows with Pixinsight you can be losing up to 50% of potential speed/power vs a Linux setup. It should take far less than half an hour calibrate, register and stack that many images.

[Slawomir (Suavi)]

Originally Posted by lazjen

What are the size of those files - 30 Mb each? If so, may I suggest you look at upgrading your computer? If you're using Windows with Pixinsight you can be losing up to 50% of potential speed/power vs a Linux setup. It should take far less than half an hour calibrate, register and stack that many images.

I totally agree. My a few years old PC did a calibration, registering and stacking of nearly 400 x 18MB lights in much much less than 30 minutes, also in PI.

[Peter Ward]

Originally Posted by Slawomir

I totally agree. My a few years old PC did a calibration, registering and stacking of nearly 400 x 18MB lights in much much less than 30 minutes, also in PI.

There is no shortage of PC's at my place, but each .CR3 RAW file from the EOS Ra are just shy of 40Mb in size. So 40 of them is 1.6Gb of data...but in BIG chunks.

My older Win i7 hex core, 32Gb RAM 3.2 Ghz machine does the just the PI registration in 15mins.

My new (2019) i7 quad core Mac running Catalina, 4.3 Ghz clock with 40Gb of RAM Registered the same set in 13 minutes (12:56 to be exact)

Stacking times were about 8 and 6 minutes respectively, so sure, not a half hour for the Win machine (23 mins) and 19 mins for the new Mac.

The 6720 x 4480 pixel raw files I'd suggest are the bottle-neck.

I'd be interested to hear from anyone running the 60Mp Sony chip to see how long it takes to register 30-40 frames.

[Slawomir (Suavi)]

Originally Posted by Peter Ward

There is no shortage of PC's at my place, but each .CR3 RAW file from the EOS Ra are just shy of 40Mb in size. So 40 of them is 1.6Gb of data...but in BIG chunks.

My older Win i7 hex core, 32Gb RAM 3.2 Ghz machine does the just the PI registration in 15mins.

My new (2019) i7 quad core Mac running Catalina, 4.3 Ghz clock with 40Gb of RAM Registered the same set in 13 minutes (12:56 to be exact)

Stacking times were about 8 and 6 minutes respectively, so sure, not a half hour for the Win machine (23 mins) and 19 mins for the new Mac.

The 6720 x 4480 pixel raw files I'd suggest are the bottle-neck.

I'd be interested to hear from anyone running the 60Mp Sony chip to see how long it takes to register 30-40 frames.

Then I’m really glad I only have an old school 9mp CCD that generates reasonable 18MB files!

[Camelopardalis (Dunk)]

Originally Posted by Peter Ward

There is no shortage of PC's at my place, but each .CR3 RAW file from the EOS Ra are just shy of 40Mb in size. So 40 of them is 1.6Gb of data...but in BIG chunks.

It’s even worse than that, because PI will debayer those such that each channel has per pixel representation, and unless you have compression enabled, the result xisf files will be much larger

I use a solid state scratch “disk” with my (much older) Mac and it speeds things along a bit, especially as I have a tendency to go with (often) hundreds of short-ish subs. Drizzling becomes a click-and-watch-a-movie affair

[Atmos (Colin)]

It took me 2.5 hours to drizzle integrate 284 exposures from my ASI094 (36.3 MP OSC). Stacking normally can be done MUCH faster though.

[Peter Ward]

Originally Posted by Atmos

It took me 2.5 hours to drizzle integrate 284 exposures from my ASI094 (36.3 MP OSC). Stacking normally can be done MUCH faster though.

Interesting.

I just ran a PI registration process, without generating drizzle data
and the 40 frame stack took 10 minutes. As Dunk pointed out the .xif files
from the earlier camera raw data were huge. 177Mb each.

Processing 7Gig of data in a quarter hour is no so bad after all!

[StuTodd]

Originally Posted by Peter Ward

I agree with Richard Wright at Sky & Telescope. For scientific work, due their non-linearity, pixel to pixel non-uniformity and amplifier glow CMOS rate a poor second to CCD's.

I suspect the demise of CCD's will be lamented by the amateur scientific community...

The best statement of this thread Peter.

This is the absolute worst case for the amateur who isn't interested in taking "pretty pictures".

With clouds, smoke, shiftwork and this sad news, I'm at the brink of giving it up

[Peter Ward]

Originally Posted by StuTodd

The best statement of this thread Peter.

This is the absolute worst case for the amateur who isn't interested in taking "pretty pictures".

With clouds, smoke, shiftwork and this sad news, I'm at the brink of giving it up

It will be a little while longer before the pool of Sony CCD sensors also dries up
(though their well depths tend to be shallow), but it you want bigger pixels with deeper wells OnSemi were the affordable gold standard.

[lazjen (Chris)]

For amateurs, what science can't be done by the CMOS cameras that CCD can do?

[Peter Ward]

Originally Posted by lazjen

For amateurs, what science can't be done by the CMOS cameras that CCD can do?

Millimagnitude photometry, high resolution spectroscopy....plus ultra deep work where you need 30 minute sub exposures just to barely get a signal.

P.S. http://www.astrosurf.com/buil/CMOSvsCCD/index.html

Is an interesting link....interference fringes seem quite problematic....but the good news is CMOS performance is improving.