Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

A review of the QHY163M camera from the perspective of a DSLR user

CMOS dslr equipment astrophotography imaging
  • Please log in to reply
36 replies to this topic

#1 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 05 October 2017 - 06:08 AM

After two years of DSO astrophotography with my Canon 7D Mark II, I decided it was time to upgrade and I recently purchased a QHY163M. The first commandment of astronomy (Thou shalt not buy a new astro-toy) was immediately enforced and the wrath of gods materialized as a thick layer of clouds, so i decided to run a full suite of (standard and non-standard) tests on my new camera to find out its real specs (the first light only came two weeks later).

 

There are many other reviews about the QHY163M and the closely related ASI1600MM-C (same sensor), but I tried to be original and to put myself in the perspective of a user upgrading from a DSLR (a very likely upgrade path for this camera), to answer some of the questions DSLR users are more likely to ask.

 

Just in case anyone is interested, I'm sharing here a short summary, while the full review (30 pages) can be downloaded from my website, in English (PDF) and Italian (PDF) (review page address: http://www.alessiobeltrame.com/qhy163m)

 

If you used to shoot DSO images with a DSLR and you're accustomed to think in terms of stops (mainly ISO stops), as opposed to gains and offsets:

  • to some extent, you will feel at home with the QHY163M
  • think of gain just like you think of ISO speed, but...
  • in a DSLR, ISO speed is proportional to the amplification, but the gain of the QHY163 is not so, therefore...
  • ADD 60 to the gain the get the equivalent of an extra ISO stop (e.g. from 240 to 300); in a DSLR you would DOUBLE the ISO (e.g.from 800 to 1600)
  • SUBTRACT 60 from gain to obtain the equivalent of a reduction of an ISO stop
  • at gain 0, the QHY163M behaves much like a DSLR at ISO 200; for the above rule, gain=60 is comparable to ISO 400, gain=120 to ISO 800, gain=180 (very close to the DSO preset in ASCOM driver) to ISO 1600, and so on
  • the DSO preset (gain=174) in the ASCOM driver is a very good compromise between read noise (1.3 electrons only) and dynamic range (11 stops) - basically, it corresponds to ISO 1600 on a DSLR, a good starting point for many Canon cameras, so...
  • if your used to expose for 3 minutes at ISO 1600 with your DSLR, you can keep the same exposure duration with the QHY163M at gain=180 - every other condition (focal ratio, filters, etc.) being unchanged. Used to work at ISO 800? No problem, just start with a gain of 180 - 60 = 120 (note: use this rule only as a starting point - the optimal exposure length for the QHY163M is likely to be different, depending on the readout noise of your DSLR, the working temperature and the filters you plan to use). 
  • Regarding the offset, just set it at 40% to 50% the value of gain. Doesn't need to be precise, just check that the histogram is detached from the axis.
  • at lowest gain you get the best dynamic range at the expense of readout noise, at highest gain you get the lowest readout noise, at the expense of dynamic range. It's just like your DSLR

As regards main specifications, my measurements confirm they are in line with manufacturer's datasheet:

  • excellent linearity throughout the entire dynamic range
  • excellent readout noise, less than 4 electrons at higher gain, just over one electron in typical usage conditions
  • excellent dynamic range, up to 12.5 stops at lowest gain, 11 stops in typical  usage conditions
  • full well capacity of 19800 electrons at lowest gain
  • very low fixed pattern noise (checked with FFT)

Regarding thermal noise, I measured what follows:

  • dark current is approx. 0,006 e-/s at -20°C
  • dark current doubles for each increment of 8.50°C in temperature
  • amp glow starts to be noticeable with exposures of some minutes - for short exposures, like 30 seconds, you don't even need darks.
  • in any case, the effects amp glow can be canceled out through a good calibration of light frames

WARNING: due to amp glow, with this camera you should never use dark frame scaling (Optimization option in PixInsight ImageCalibration process - the guys at Pleiades are working on that, see https://pixinsight.c...g66770#msg66770). As pointed out many times elsewhere on CN, just use darks taken at the same temperature, exposure time, gain and offset. Using darks taken at different temperature may result in under-correction or over-correction.

 

My overall evaluation: the QHY163M is an excellent camera with an outstanding price/performance ratio, which you may also use for planetary work and/or in unconventional ways (due to the very low read noise, you're not forced to go with the very long exposures that are required by CCD cameras, particularly in narrow band imaging). Beside that, the use of a CMOS sensor could be an attractive feature for DSLR users searching for an upgrade. At the very least, the learning curve should be easier to climb.

 

If you find any error or flaw in my review or my procedures, please let me know. Hoping it may be useful to somebody.

 

Alessio


  • gonnessia, buckeyestargazer, rathijit and 5 others like this

#2 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 05 October 2017 - 07:38 AM

Very nice summary here (I haven't read the full review).

 

I haven't looked very hard yet, but is there a table somewhere that shows the changes in read noise as gain is increased/decreased?  I was a little surprised by this:

"the DSO preset (gain=174) in the ASCOM driver is a very good compromise between read noise (1.3 electrons only) and dynamic range (11 stops)"

I've been working at gain 50 for LRGB imaging in order to maximize dynamic range, but now I'm wondering if the trade-off is worth it.  If I could see what the read noise is at various gain settings that would help me decide what gain to use.



#3 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1,734
  • Joined: 05 Mar 2009
  • Loc: Germany

Posted 05 October 2017 - 07:44 AM

wow, very comprehensive!!! great job. BTW I did not read it carefully nor completely (just browsed through) but what caught my attention was mentioning biases in calibration process. I would never use them with this sensor, maybe you used them for short flats. As you can see here, in my yet another issue with 163m:

https://www.cloudyni...libration-only/

^ please could you comment there that you see the same what I do see (that my camera is faulty)

?

my issue is just with simple dark calibration as you can see above in my problem description. Thanks if you give some feedback from your side :D

good job!

BTW after my experience I'd (from now on) only recommend ZWO ASI's :D I have had many QHY cameras in the past and 70% of them went back to China to repair (I have insane bad luck on anything I purchase).



#4 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1,734
  • Joined: 05 Mar 2009
  • Loc: Germany

Posted 05 October 2017 - 07:48 AM

Joel:

QHY163M_V2_1.jpg

QHY163M_V2_2.png

QHY163M_V2_3.png

QHY163M_V2_4.png

source: http://qhyccd.com/QHY163.html


  • EMG likes this

#5 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 05 October 2017 - 08:07 AM

Thanks Pavel, I just didn't scroll down far enough on the QHY website (I didn't realize there was still real info underneath the sample images).  

 

So now my question is, what is an acceptable level of dynamic range?  12 sounds better than 11...



#6 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 05 October 2017 - 09:01 AM

wow, very comprehensive!!! great job. BTW I did not read it carefully nor completely (just browsed through) but what caught my attention was mentioning biases in calibration process. I would never use them with this sensor, maybe you used them for short flats. As you can see here, in my yet another issue with 163m:

https://www.cloudyni...libration-only/

^ please could you comment there that you see the same what I do see (that my camera is faulty)

?

my issue is just with simple dark calibration as you can see above in my problem description. Thanks if you give some feedback from your side laugh.gif

good job!

BTW after my experience I'd (from now on) only recommend ZWO ASI's laugh.gif I have had many QHY cameras in the past and 70% of them went back to China to repair (I have insane bad luck on anything I purchase).

My calibration procedure was essentially the same described by Vicent Peris on PixInsight website (https://www.pixinsig...ames/index.html), with some exception. If I rembember well:

  1. Bias frame integration into a Master Bias
  2. Dark frame integration into a Master Dark
  3. Flat frame calibration with Master Bias and Master Dark (this is the part I'm less certain about)
  4. Flat frame integration into a Master Flat
  5. Light frame calibration with Master Bias, Master Dark and Master Flat. If you use PixInsight, leave Optimize option unchecked under Master Dark !!! This is really critical
  6. Light frame integration into several Master Darks (Ha, OIII, R, G, B)

As Jon Rista pointed out, it is mandatory that Dark Frames are taken at the same exposure length, temperature, gain and offset of Light frames. A difference of 10°C makes all the difference of the world, I tried to do that way (with Master Dark scaling enabled) and it didn't work. When amp glow is high enough to be noticeable, dark frame scaling simply does not work but, as far as I understand, this is not a problem of this particular sensor or of QHYCCD, it is related to the non-linearity of the amp glow itself. 

 

Anyway, as per your request in the post you're mentioning, I'm sharing with you a couple of darks, the master dark and a light frame in a Dropbox folder


  • Jon Rista and RossW like this

#7 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 05 October 2017 - 09:38 AM

Thanks Pavel, I just didn't scroll down far enough on the QHY website (I didn't realize there was still real info underneath the sample images).  

 

So now my question is, what is an acceptable level of dynamic range?  12 sounds better than 11...

Joel,

in my review the graphs of QHYCCD are substantially confirmed. A high dynamic range (DR) allows you to collect more signal from faint areas of your target and, at the same time, avoid star saturation, so a higher DR is always better, particularly for LRGB. But keep in mind that you can increase the DR only at the expense of readout noise (RON), and in turn a higher RON will force you to have longer exposures to be sky limited, particularly with narrow band filters. To me, DR=11 is a very good compromise, but I have a bad sky and narrow band is almost mandatory, so I prefer a low RON to a high DR. Again, that's just me, your situation may be completely different.

 

By the way, a typical KAF-8300 camera has a DR between 11 and 12 stops, so in any case (11 or 12) you're in a good shape. 



#8 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 05 October 2017 - 01:02 PM

Typically for narrowband I set the gain to 200, which according to the graphs is about 10.5 stops of DR.  It's interesting that the read noise basically stays the same from 200-500.  But the DR drops off with higher gain.  So at gain 300 I'm down to 9.5 stops DR.  

 

Typically for LRGB I used gain 50.  I'm thinking I need to increase the gain to at least unity (120) for LRGB.  I don't lose much in DR but the read noise is improved.

 

I also wonder about increasing the gain for narrowband (from 200).   That is less clear to me.



#9 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 05 October 2017 - 03:23 PM

Joel,

I would recommend you to read this topic and particularly Jon Rista's recommendations, to me they are very sound.

 

Regarding which gain to choose, I agree with you that a low gain is better for LRGB. From my "dark" location I would saturate everything, let alone the stars, in one minute with a gain higher than 300 (BTW, my scopes are rather fast). But that is the situation where I want the highest DR, otherwise faint signal would fade away into the huge skyglow. If you wanted to try the "planetary" technique (a huge number of very short subs) you would need to do just the opposite, very high gain to get the lowest possible RON.

 

For NB it's another story. First of all, stars are less prone to saturate as you filter out most of their light. On the other hand, skyglow almost disappears. In this situation I prefer to think about exposure length, rather than DR/RON.

 

I don't have an observatory, so time on the field is precious. For this reason, I'm more comfortable with midrange exposures. I learned my lessons in guiding and I could go with very long exposures, if a wanted to. But I don't want to take the risk of loosing half an imaging session is something goes wrong with a particular sub. On the other hand, I don't want to mess with a ton of short exposures either. 

 

5 to 15 minutes is my sweet spot. Now, the optimal condition would be to have such a low RON to be sky limited, i.e. the square of RON should be negligible with respect to skyglow. A common definition of "negligible" is "less than 5% of skyglow". You calculate how much RON you can afford with an exposure of, let's say, 10 minutes and from that value you find the level of Gain you need to set into the camera. Finally, you check that with the level of Gain the DR is still appropriate (let's say at least 9 to 10).

 

Now, let's put it to practice. My sky is around 19.5 mag/arcsec2 and, if I'm doing it right, I should get around 0,08 electrons/s in H-alpha with my Astronomik 6-nm filter and with my Tak FSQ-85 @ f/5.3. In a 10 minute exposure that accounts for a total of 48 electrons. Multiply 48 electrons by 5% and take the square root and you get 1.54: that's your magical number for the RON.

 

In practice I was more conservative and I went for for a RON of 1.3 electrons, that corresponds to a DR of 11 at a gain around 180. By chance that's more or less the "DSO" preset of the ASCOM driver. All numbers look very good, so that's my starting point in narrow band. If you get something you don't like (e.g. a DR lower than 9-10) you can try to change the starting assumptions, like the exposure length, and start again.

 

That's my way of doing it, but take it with a grain of salt and don't take my calculations as gospel! In any case, you don't need to be very precise. The above condition is just the optimal case, but it's not a must.


Edited by alessio.beltrame, 05 October 2017 - 03:44 PM.


#10 AtmosFearIC

AtmosFearIC

    Apollo

  • -----
  • Posts: 1,181
  • Joined: 10 Dec 2015
  • Loc: Melbourne

Posted 05 October 2017 - 05:51 PM

I have been using a Gain of 80 as it is about as close as I can get to 12-bits of dynamic range. At 80 I am getting 1.9e- read noise which is low enough and keeps the exposures short enough for LRGB imaging. I wouldn’t want to have exposures any shorter than this.

I’ve also found that darks are required for pretty much any exposure. If I don’t dark subtract 30s exposures there is banding that cannot be removed from the background.

With my QHY22 I could do 10 minute exposures with just a bias frame and it looked perfect.

#11 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 05 October 2017 - 06:22 PM

I have been using a Gain of 80 as it is about as close as I can get to 12-bits of dynamic range. At 80 I am getting 1.9e- read noise which is low enough and keeps the exposures short enough for LRGB imaging. I wouldn’t want to have exposures any shorter than this.

I’ve also found that darks are required for pretty much any exposure. If I don’t dark subtract 30s exposures there is banding that cannot be removed from the background.

With my QHY22 I could do 10 minute exposures with just a bias frame and it looked perfect.

What are your LRGB exposures times at gain 80?  Narrowband gain and exposure times?



#12 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,336
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 06 October 2017 - 12:48 AM

If I do LRGB with the ASI1600, I try to do it at Gain 76. In my testing, that gives you 12 stops of DR, ~2e- read noise, and ~2e-/ADU gain. Gain 0 also gives 12 stops of DR (12-bit ADC clipping), but 3.5e- read noise and ~4.9e-/ADU gain!

 

Now, there is one key difference between Gain 0 and Gain 76 (and this should hold true with the counterpart gain settings on a QHY): At Gain 0 you must expose for much longer in order to A) swamp the read noise and B) fill up the pixel wells. At Gain 76 (and whatever teh counterpart is on the QHY), your exposures will be shorter, but you DO have the same 12 stops of DR. Just keep that in mind. You won't be using 10-15 minute LRGB exposures at Gain 76. DR isn't just relative to FWC, it's also relative to read noise. You can have the same ratio of FWC/RN with many different numbers for both. The amount of read noise will usually dictate how long you expose...but, if you expose properly for the read noise, then regardless of what the read noise and FWC is, your utilization of the available DR should be the same.

 

With Gain 76 exposures, I use shorter ones (which for me is fine, I like the resolution benefit, and I think that stacking at least 100 frames is ideal with these 12-bit cameras), both for L and RGB. The results are also usually better than Gain 0 with the same integration time, as Gain 0 (at least on the ASI) seems to have more banding issues than Gain 76. I don't know if the QHY is the same here, but it's the same sensor, and the readout logic is all on-die, so I imagine it is the same. 



#13 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1,734
  • Joined: 05 Mar 2009
  • Loc: Germany

Posted 06 October 2017 - 02:31 AM

alessio.beltrame: thank you! I will get the frames and have a look.

 

ad topic: I wonder WHY (what is your motivation) to mess with biases??? I do not see any reason to do so. I am, myself, paranoid about PERFECT calibration and that is why I always do it a bit different way for every different CCD/CMOS camera that I owned in the past. Some need biases and flats and no darks (typically SONY's CCDs) some need only properly matching dark frames (exact for lights and flats) like KAI chips and CMOS chips with amp glow, some CCD cameras, mainly KAF types can benefit from dark frame optimization (scaling as in MaxIm DL terminology) for cases you do not have properly long (matching long) dark for the lights you are about to calibrate and for this scaling you need to have bias and some other length dark to be able to optimize with mathematical form. There are cameras/chips where this works (also for DSLRs) and there are those where this can't be used at all (as you also point out).

 

I am, BTW not using PI for calibration, am sticking with old MaxIm DL to know what it does and keep it under my control (exactly knowing what it does, I am too old and lazy to calibrate in PI).

 

today, I finally, managed to open a new ticket on qhy support site and wonder what I will hear back from them. I expect a typical set of questions like did you have properly matching dark for gain etc. all basic things that many people mess, but I feel, in this area a bit experienced so I really know what I am doing... and when I post a comment answering all these basic questions then I do not hear back (typically) because when I do not know the answer then "nobody" :-(

 

anyways, you did a great job in your review! wishing you clear skies and good luck!

ciao

Pavel



#14 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 03:16 AM

If I do LRGB with the ASI1600, I try to do it at Gain 76. In my testing, that gives you 12 stops of DR, ~2e- read noise, and ~2e-/ADU gain. Gain 0 also gives 12 stops of DR (12-bit ADC clipping), but 3.5e- read noise and ~4.9e-/ADU gain! [...].

The results are also usually better than Gain 0 with the same integration time, as Gain 0 (at least on the ASI) seems to have more banding issues than Gain 76. I don't know if the QHY is the same here, but it's the same sensor, and the readout logic is all on-die, so I imagine it is the same. 

Jon, 

my measurements perfectly fit your numbers. By interpolating my graph of read noise vs gain and DR vs gain I get DR=12 and RON=2 e- at gain=75. After QHY changed the gain scale, comparing the QHY163M and the ASI1600 is actually much simpler. 

 

I didn't compare banding at different gain levels, that would be an interesting thing to test and perhaps I'll include it in the next  version of the review, which may come this weekend (autumn is funny in north-eastern Italy... with low atmospheric pressure you get plenty of clouds, with high pressure you get fog).

 

Anyway, I don't expect to find anything different from your ASI1600 because, as you said, the sensor is the same and it has on-die amplification and ADC sections.



#15 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 03:41 AM

Pavel,

 

ad topic: I wonder WHY (what is your motivation) to mess with biases??? 

Short answer: why not? Bias frames are very easy to collect and process. Using SGP I'm experiencing "slow" download time (if 3-4 seconds can be defined slow), but with AstroImager or Kstars/Ekos under MacOS I easily download a full frame (1x1 bin) in less than 1.5 seconds. This means that you can collect a 200 bias frames in less than 5 minutes and have a Master Bias ready in 10 minutes more with a decent computer. A since you will use the Master Bias again and again, I can see no reasons why doing it another way. Of course you will need more Master Darks, one for each combination of gain and offset. But again, I don't see any reason to use the camera with more that 3-4 gain/offset presets.

 

The long answer is just... too long for this post and very theoretical, but in the end what is really important is the end result. If bias calibration has negligible result, you can just ignore it and, yes, some cameras are better than others in this respect.

 

By the way, when I used the 7D Mark II it was a different story and I didn't use bias or dark frames. That's because dark calibration ranged from useless (in winter, with sensor near to 0°C) to impossible (in summer, with sensor temperature changing wildly from 27-28°C to 37-38° in less than one hour). Controlled cooling is a great invention.



#16 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 04:45 AM

I have been using a Gain of 80 as it is about as close as I can get to 12-bits of dynamic range. At 80 I am getting 1.9e- read noise which is low enough and keeps the exposures short enough for LRGB imaging. I wouldn’t want to have exposures any shorter than this.

I’ve also found that darks are required for pretty much any exposure. If I don’t dark subtract 30s exposures there is banding that cannot be removed from the background.

With my QHY22 I could do 10 minute exposures with just a bias frame and it looked perfect.

I can't see any banding in short LRGB exposures. This FITS file is a quick test with 5x30" subs (G=174; O=77; T=-20°C) on R, G and B channels and with minimum processing (alignment, individual channel integration, channel combination, color calibration). Please don't look at the stars, it's just a test and focus was "just in the ballpark" (the FSQ-85 is very unforgiving in this respect) with humidity very close to 100%. You don't mention your operation temperature and offset values, they are important parameters to take into account when comparing pictures and cameras.

 

Also, the QHY22 is a CCD camera, so it doesn't really compare well with a CMOS cameras. Different tools for different scenarios (just a few long exposures vs a lot of long exposures). 



#17 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1,734
  • Joined: 05 Mar 2009
  • Loc: Germany

Posted 06 October 2017 - 07:22 AM

alessio.beltrame: thanks for the dropbox data. I had a look. Your master dark is in xisf format so maybe that plays some role, but if I subtract it from a single dark then I see the remaining amp-glow in your image. This is kind of strange, because the veil seems to be calibrated well with that master dark. So to prove if all's good or strange like mine (though not that big) I'd need more single darks to create the master myself. But I won't bother you with that, have already got some darks from other 163m owners and all are good, only my camera shows the amp glow... end of OT.

 

another OT: your veil has oval stars in edges. That looks like if your back focal distance is too long. You should shorten it by 2mm (guessing) :-) to be able to focus well all corners. Just my observation made on looking on your fits.

 

ad bias:

there are only few use-cases when using bias does have a sense. When it doesn't have a sense I do not really see any reason to do something use-less or contraproductive even when it is too easy to collect them.

 

some (not all, but some) facts:

- biases may replace short exposure darks for e.g. calibration of flats

- every dark frame already contains the pattern that bias frame has so using both darks and biases is like subtracting bias twice from the light frame during calibration and that is not what I'd recommend to do...

- everything depends on the CCD/CMOS chip we talk about, every needs different approach for best calibration :D

 

and yes, I agree with many things you wrote like e.g. that one may need 2 (maybe three) gain/offset settings for all use-cases so no need to fiddle with it. I myself used only 2 settings, most optimal is the unity gain for general LRGB and lower readout noise settings (higher gain sacrifying DR) for narrow band. And with combination of longer subs I cover all I may need. But I am actually not using the 163m due to the issue that I can't calibrate it well.



#18 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 06 October 2017 - 07:41 AM

What are you guys using to measure the camera, particularly the dynamic range?  I'm trying to use Pixinsight's BasicCCDParameters script but the numbers just aren't matching up.  At low gain settings it shows DR of 8 or less.



#19 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 08:56 AM

What are you guys using to measure the camera, particularly the dynamic range?  I'm trying to use Pixinsight's BasicCCDParameters script but the numbers just aren't matching up.  At low gain settings it shows DR of 8 or less.

I'm using BasicCCDParameters too. For a more complex, but more reliable and precise method, take a look at Emil Martinec's lectures at http://theory.uchica...e/noise-p2.html



#20 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 06 October 2017 - 09:07 AM

Perhaps I'm not using the script correctly, or not understanding how to read the results I'm getting. 

 

1.  I assume that the readout depth and A/D bits should be 12.  

2.  What should the maximum ADU be?

3.  Do I need to somehow scale the resulting gain, full well capacity and dynamic range numbers to account for 12bit vs 16bit?  Like I said the numbers I'm seeing don't match at all what the specs in the graphs would indicate.

 

I should note that I don't understand what "stops" of dynamic range means.  I'm not a DSLR guy...



#21 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 09:28 AM

alessio.beltrame: thanks for the dropbox data. I had a look. Your master dark is in xisf format so maybe that plays some role, but if I subtract it from a single dark then I see the remaining amp-glow in your image. This is kind of strange, because the veil seems to be calibrated well with that master dark. So to prove if all's good or strange like mine (though not that big) I'd need more single darks to create the master myself. But I won't bother you with that, have already got some darks from other 163m owners and all are good, only my camera shows the amp glow... end of OT.

 

another OT: your veil has oval stars in edges. That looks like if your back focal distance is too long. You should shorten it by 2mm (guessing) :-) to be able to focus well all corners. Just my observation made on looking on your fits.

 

ad bias:

there are only few use-cases when using bias does have a sense. When it doesn't have a sense I do not really see any reason to do something use-less or contraproductive even when it is too easy to collect them.

 

some (not all, but some) facts:

- biases may replace short exposure darks for e.g. calibration of flats

- every dark frame already contains the pattern that bias frame has so using both darks and biases is like subtracting bias twice from the light frame during calibration and that is not what I'd recommend to do...

- everything depends on the CCD/CMOS chip we talk about, every needs different approach for best calibration laugh.gif

 

and yes, I agree with many things you wrote like e.g. that one may need 2 (maybe three) gain/offset settings for all use-cases so no need to fiddle with it. I myself used only 2 settings, most optimal is the unity gain for general LRGB and lower readout noise settings (higher gain sacrifying DR) for narrow band. And with combination of longer subs I cover all I may need. But I am actually not using the 163m due to the issue that I can't calibrate it well.

Pavel,

 

XISF is a FITS extension, so I don't believe it can cause any issue. Anyway, I uploaded in the same folder the FITS version. I also uploaded a new file, names Light-MasterDark.fit. I generated this file by simply subtracting the Master Dark from the light frame you found in the folder. That's it, a plain subtraction using a tool named PixelMath. To my eyes it calibrates well, check for yourself.

 

Sorry, I never used Maxim so I have no easy explanation for what you're experimenting.

 

Regarding star shapes, thanks for the hint but I used the bare FSQ-85, without reducer. It's a Petzval design, so spacing should not be a factor (it would be with the reducer). But it's very critical on focusing and it shows deformation along concentric lines for extra-focal error and radial lines for intra-focal error. The fact is that I still have to optimize the autofocus routine with SGP and with this picture I just wanted to check that hardware (including filter wheel and filters) and software worked smoothly. 



#22 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 09:42 AM

Perhaps I'm not using the script correctly, or not understanding how to read the results I'm getting. 

 

1.  I assume that the readout depth and A/D bits should be 12.  

2.  What should the maximum ADU be?

3.  Do I need to somehow scale the resulting gain, full well capacity and dynamic range numbers to account for 12bit vs 16bit?  Like I said the numbers I'm seeing don't match at all what the specs in the graphs would indicate.

 

I should note that I don't understand what "stops" of dynamic range means.  I'm not a DSLR guy...

Joel,

as far as I understand the script (I never had time to look into the source code) and the camera:

 

1. A/D bits are indeed 12

2. Max ADU should be 2^16 = 65536 because the A/D reading are shifted b 4 bits, anyway the max I measured is 65504

3. The number that really matter are those expressed in electrons. The gain is the exception, as it mixes electrons with ADU's, so you should multiply by 16 the output of the script in e-/ADU

 

Perhaps Jon can confirm that. By the way, The very fact that I got similar results to those publish by Jon and others, using a different camera with the same sensor, made me confident in my measurements. 

 

A stop is just a power of 2, so a DR of 4096 translates into log2(4096) = log(4096)/log(2) = 12. In the other way, DR = 11 stops means 2^11 = 2048. If you're used to dB, 12 stops of DR are equivalent to a ratio of 4096:1 between FWC and read noise, or 20*Log(4096) = 72.2 dB.



#23 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 06 October 2017 - 10:04 AM

The math makes my head hurt..

Here's my result at gain 50 offset 35.

 

Attached Thumbnails

  • QHY163_BCCDP.png


#24 alessio.beltrame

alessio.beltrame

    Sputnik

  • -----
  • topic starter
  • Posts: 29
  • Joined: 23 Dec 2016
  • Loc: Mortegliano (UD) 33050, Italy

Posted 06 October 2017 - 10:19 AM

The math makes my head hurt..

Here's my result at gain 50 offset 35.

While the physical resolution of the ADC is 12 bit, data in FITS file is 16 bit. AFAIK about this script, you need to put 16 into readout depth and A/D bits.



#25 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • Posts: 5,254
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 06 October 2017 - 10:44 AM

Obviously that's what I was doing wrong.  This makes more sense now, although multiplying the gain reading by 16 only gives me 2.72(?).  

Attached Thumbnails

  • QHY163_BCCDP2.png



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: CMOS, dslr, equipment, astrophotography, imaging



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics