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

QHY461 first light and review

  • Please log in to reply
39 replies to this topic

#1 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 28 November 2020 - 03:35 PM

this last week i machined an adapter for the creatively named QHY461(uses the sony IMX461chip) and was able to do some initial test images.

 

here's the camera mounted to an FLI filter wheel w/ 65mm chroma filters.  the adapter (made from some scrap 6061 aluminum i had laying around) hasn't been painted yet, but i'm not really sure i was able to see a problem with that in the end results, but i should probably paint it so it doesn't look weird.

 

the telescope is a CDK1000 w/ a reducer which gives it a focal length of 4250mm.

 

eAowGbv.jpg

 

you can see that the camera has liquid cooling ports (yellow capped) and two power inputs (black capped) and some gigabit and USB3 ports.

 

my initial impression is that the base cooling isn't super great.  i took it down to -20C (but the ambient temperature was -12C) and it looked like it was taking a fair amount of power to achieve that.  i'm guessing the unit really is meant to be liquid cooled.

 

the camera only supports 1x and 2x binning and w/ a pixel size of 3.7um i end up with extremely oversampled images even at 2x bin, giving me 0.36 arcsec per pixel and very large files (~50MB) that balloon up to ~100MB during processing in pixinsight.

 

here's a screenshot of a bias frame:

Hjc0EQq.png

 

superbias frame (96 frames):

 

BDYB1Kj.png

 

single dark frame (180s b2):

 

wgKbB8T.png

 

 

running cosmetic correction seems to not be very useful - doing a realtime preview i wasn't able to see any noticeable differences while tweaking the hot and cold pixel settings. i ran it anyway, but that was noticeably different than my experience with other cameras.

 

here is the processed image.  74 frames, 180s each w/ Ha filter.  average seeing for my location (~2.5 FWHM)

 

processed w/ pixinsight (curve adjustments and that's it) and a touch of photoshop (histogram adjustments and very light sharpening.)

 

fCi1L9E.png

 

so i don't know if this is just psychology or not, but i'm really impressed by what this camera was able to do.  i feel like i can say this is the best camera i've ever used.

 

anyway, this concludes my QHY461 first light and review post. smile.gif


  • Gregory, Mert, X-Lurker and 15 others like this

#2 Tapio

Tapio

    Soyuz

  • -----
  • Posts: 3,564
  • Joined: 24 Sep 2006
  • Loc: Tampere, Finland

Posted 28 November 2020 - 03:42 PM

Nice review.

What is that thing in the back ?



#3 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 28 November 2020 - 04:09 PM

Nice review.

What is that thing in the back ?

i'm not sure what you are referring to.  maybe that orange thing?  that's a couch.  or do you mean that cylinder attached to the back of the camera?  i actually have no idea what that is, but it's part of the camera.  even though the images on the QHY website don't show it, it came attached to my unit.



#4 dghent

dghent

    Viking 1

  • *****
  • Vendors
  • Posts: 909
  • Joined: 10 Jun 2007

Posted 28 November 2020 - 10:23 PM

This looks stunning. I believe the IMX411 and IMX461 share the same technology stack as the IMX455. If that's the case, you basically the biggest version of the best mono cmos commercially available out there.

 

You can try IntegerResample in PI to do binning to a desired level after the fact. This won't help you with file sizes as the data lands on your storage, but it'll address the pixel scale - plus you can experiment with the same 1x1 data set that way to see what works for you. NINA can directly write out compressed XISF files if it's a real concern for you, and it does the compression in the background so it doesn't block the following exposure from starting.


  • leviathan and happylimpet like this

#5 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 28 November 2020 - 10:37 PM

the file sizes don't bother me too much, but maybe i'll try to play around with PI and/or NINA to help out there until QHY can get 4x binning functional (i'm hoping it's something they can update via software.)

 

btw, i didn't mention that the QHY461 seems to be very "plug and play" compared with the QHY50GX or the FLI kepler 6060 which were both pretty awful to use when it came to software compatibility and ease of use.

 

btw, i tried using NINA with it and it worked great.  i'm trying to wean myself off of maximDL to switch to NINA but old habits die hard and planewave mount controller software requires it to make pointing models.

 

one more first light pick - this is a one hour integration i took last night, 180s frames: (aperture plus high QE is fun!)

 

50657577443_7f2a43766b_c.jpg


  • Mert, dghent, R Botero and 3 others like this

#6 dghent

dghent

    Viking 1

  • *****
  • Vendors
  • Posts: 909
  • Joined: 10 Jun 2007

Posted 28 November 2020 - 11:26 PM

That's super nice, but the darks seem a little too dark, just based on what I see here in this jpeg. Did you up the gain for these shots without also upping the offset perchance?



#7 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 11,473
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 29 November 2020 - 12:49 AM

Now ZWO needs to come up with a consumer version of IMX461; matter of time. Regards

 



#8 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 29 November 2020 - 01:15 AM

That's super nice, but the darks seem a little too dark, just based on what I see here in this jpeg. Did you up the gain for these shots without also upping the offset perchance?

 

dark darks could just be how i adjusted STF in PI and curves in PS.  i left the gain/offset on the default DSO settings, which appear to be 60 and 75 respectively.

 

i'm pretty new convert from ccd to cmos so i don't quite understand gain/offset yet.



#9 dghent

dghent

    Viking 1

  • *****
  • Vendors
  • Posts: 909
  • Joined: 10 Jun 2007

Posted 29 November 2020 - 11:49 AM

dark darks could just be how i adjusted STF in PI and curves in PS.  i left the gain/offset on the default DSO settings, which appear to be 60 and 75 respectively.

 

i'm pretty new convert from ccd to cmos so i don't quite understand gain/offset yet.

 

The defaults in the ASCOM driver are "just kinda ok" and aren't geared towards any particular camera, especially with sensors that are multi conversion gain sensors such as this one. There are in all probability better settings that'll inform you where you should be with gain (and then offset), but what those settings could be are hard for me to divine because QHY hasn't posted some gain to noise/DR/well depth graphs for this sensor. Does this camera have multiple readout modes or just one?

 

I asked Dr. Qiu to post those graphs so we can nail down its basic characteristics. But we don't have to wait for those to appear. If you have a copy of sharpcap pro and are familiar with its Sensor Analysis feature, you can run that and we can look at its results. Prerequitites for using the Sensor Analysis feature are a constant light source (ie, flat panel) and the ability to block light from the sensor (as if you were taking darks). If you can do that with your setup and have SharpCap Pro, you can run the Sensor Analysis. It walks you through each step and is largely hands-off until you need to cover the sensor for the portion of the analysis where the sensor needs to be dark.


Edited by dghent, 29 November 2020 - 11:49 AM.


#10 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 29 November 2020 - 07:11 PM

okay, so i ran sensor analysis (it took about an hour or two, and ultimately sharpcap crashed, but not before i was able to copy the data.)

 

there is a strange drop off in noise between 25 and 30 gain.

 

now i'm doing a search to figure out what this all means in practical terms..

 

xfzyPud.png

 

PTkXaVi.png


  • dghent and tkottary like this

#11 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,557
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 29 November 2020 - 07:40 PM

Sounds about right. Look at the read noise chart at:

 

https://www.qhyccd.c...=94&id=55&cut=1

 

Scroll a bit down and look at the purple graph. This is for QHY600M camera but I believe it uses same sensor technology. Read noise drops at gain of 26.

 

Peter 


Edited by Peter in Reno, 29 November 2020 - 07:41 PM.

  • prefetch and tkottary like this

#12 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 29 November 2020 - 07:51 PM

thanks peter.  that's interesting that it matches pretty well with the 600.

 

so is the takeaway that if i want to avoid a big read noise hit, but still have a reasonably high dynamic range and well capacity, i should set my gain at 30?



#13 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,557
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 29 November 2020 - 08:12 PM

I have not yet imaged with my QHY600M due to personal reasons and bad weather. If it was up to me, for LRGB imaging, I plan to use low gain for higher full well and dynamic range. Read noise is pretty irrelevant for LRGB imaging because it will get quickly swamped in a short time. Low gain will allow you to increase sub-exposure times and therefore reduce total number of subs which helps to save storage space.

 

I don't plan to use NB imaging so I let others chime in who have more experience. 

 

Enjoy your new toy.

 

Peter 



#14 dghent

dghent

    Viking 1

  • *****
  • Vendors
  • Posts: 909
  • Joined: 10 Jun 2007

Posted 29 November 2020 - 08:58 PM

okay, so i ran sensor analysis (it took about an hour or two, and ultimately sharpcap crashed, but not before i was able to copy the data.)

 

there is a strange drop off in noise between 25 and 30 gain.

 

now i'm doing a search to figure out what this all means in practical terms..

OK, so QHY has programmed a gain profile into this camera that appears to mirror their other camera that have sensors in this class such as the IMX571 (QHY268) and 455 (QHY600). I'm willing to bet that the 151mpx IMX411 (QHY411) monster is no different.

 

What you found in that jump in gain from 25 to 30 is the sensor transitioning from the Low Conversion Gain (LCG) to the High Conversion Gain (HCG) realm; a characteristic of mutli-conversion gain sensors and is critical information to have in order to properly utilize these sensors in an astronomy context.

 

As you've noticed, the read noise declines as the gain rises from 0, but drops by a lot instantly upon entering the HGC realm. You will also notice that the dynamic range also recovers back to nearly the same level as gain=0. It is here that the high conversion gain directly benefits astrophotography, as it's operating the pixels in a way that is specifically designed to get a clean amplified signal for low-light levels. If you want reading material, you can start at page 42 of this dissertation.

 

So, in the context of the QHY461, it's likely safe to say that the HCG realm begins at gain=26. You can test this by capping off the camera so that it's dark, setting the fastest exposure speed you can, and take a frame at gain=25 and another at gain=26. Open these frames up in PI and run the Statistics process on each. You will need to have StdDev displayed (hit the gears and add it if you don't see it) and set the image to 16bit and turn scientific notation off. You will see the StdDev for the gain=25 image to be at one level, then the image at gain=26 will have a very noticeably lower StdDev. If you see that, you have confirmed that gain=26 is where the HCG realm begins as the standard deviation, which is a measure of read noise, has dropped.

 

As for Offset, I've written up a blog entry on how to understand what it is, why it needs to be correctly set, and how to figure out how to find what is a sufficient offset is for the gain you are operating at. I'll spoil it for you that, at gain=26 (assuming that is indeed where the HCG realm begins on the QHY461) you will likely want an offset in the 45-50 range... but read my article on it and you can learn how to identify what is a good setting regardless of the sensor.


  • R Botero likes this

#15 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,557
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 30 November 2020 - 11:53 AM

prefetch,

 

What is the file size for each downliaded unbinned 1x1 FIT format image? 

 

Peter 



#16 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 30 November 2020 - 01:12 PM

prefetch,

 

What is the file size for each downliaded unbinned 1x1 FIT format image? 

 

Peter 

203 MB.

 

yikes, right? 

 

2x2 binning gives me a 52 MB file size which is workable and i'm pretty sure i don't need the 0.18 arcsec/pixel resolution of the unbinned image.


  • R Botero likes this

#17 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,557
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 30 November 2020 - 01:57 PM

Yikes is right. For LRGB imaging, you could try gain 0 and offset 50 and expose longer per sub so your total number of subs would be less and reduce storage space.

 

Does QHY camera driver allow you to change 4 different read modes? If so, study the following graphs to determine optimal read modes for your filters and sky conditions (light pollution, seeing, transparency, etc.).

 

https://www.qhyccd.c...=94&id=55&cut=1

 

Contact QHY whether QHY600M graphs will work for your camera since they don't post graphs for your camera. 

 

Peter 



#18 dghent

dghent

    Viking 1

  • *****
  • Vendors
  • Posts: 909
  • Joined: 10 Jun 2007

Posted 30 November 2020 - 02:30 PM

I asked Dr. Qiu about readout modes last night and he said that the 411 has 8 analog channel configurations, each being represented as a unique readout mode. He said each has its own gain and full well, and none should operate in the combined LCG+HCG realms - each curve is in either the LCG or HCG realm, each with its own gain and well depth ranges.

 

prefetch: Can you pop open the QHY ASCOM driver config window, open up the readout mode selection drop-down, and provide a screenshot of what's listed there? I'm super curious based on what Dr. Q said and what you found doing the sensor analysis.



#19 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,557
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 30 November 2020 - 02:59 PM

What dghent is asking is something like this:

 

4th_read_mode.jpg

 

Peter



#20 rockstarbill

rockstarbill

    Cosmos

  • *****
  • Posts: 7,960
  • Joined: 16 Jul 2013
  • Loc: Snohomish, WA

Posted 30 November 2020 - 03:14 PM

Beastly sensor there! :) 



#21 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 30 November 2020 - 07:56 PM

peter, dghent, here are the readout modes:

 

APObgFQ.png



#22 dghent

dghent

    Viking 1

  • *****
  • Vendors
  • Posts: 909
  • Joined: 10 Jun 2007

Posted 30 November 2020 - 10:02 PM

Huh, OK. Perhaps this version of the SDK is doing something different then. But it does look like this camera is indeed being ran as if it were a bigger version of the QHY600 and its IMX455.

 

In which case, I would invite you to also explore High Gain Mode, at a gain=56 / offset=25 (assuming the characteristics of the QHY600 are the same here too, which they likely are). As far as the readout modes are concerned, it looks like you can take experiences of those with the QHY600 and use them with this camera.



#23 prefetch

prefetch

    Viking 1

  • -----
  • topic starter
  • Posts: 709
  • Joined: 24 Apr 2013

Posted 01 December 2020 - 12:43 AM

dghent, i think you're right -

 

a user named ricardo posted a sharpcap output using the different read modes, and it looks like 56 is the right gain for the 'High Gain Mode 16BIT' setting, whereas for the 'PhotoGraphic DSO 16BIT' setting the best gain would be 26.

https://www.cloudyni...ions/?p=9938875

 

so it looks like i should do another sharpcap run with the high gain readout mode and confirm that 56 will give me the best combination of lowest read noise and the highest dynamic range.

 

as far as the offset goes, i'm not sure what the optimal setting is (i read something about checking the histogram of a bias frame and making sure the left side isn't clipped) - but from what i gather it seems to be significantly less critical than the gain setting.



#24 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,557
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 01 December 2020 - 12:50 AM

prefetch,

 

Are you a LRGB or NB imager?

 

If you are LRGB imager, personally I think I would use "Photographic DSO 16 Bits" at 0 gain and offset 50 for maximum Full Well Capacity and Dynamic Range so you won't have to worry about too many saturated stars. The only disadvantage is longer sub-exposure times but fewer total number of subs.

 

For NB imaging, it's obvious that higher gain would be better but I am not an NB imager so I do not know which settings would be best so I would use dghent suggestion of High Gain Mode 16 Bits, at gain 56 and offset 25 (or 50).

 

Offset is not critical so don't go overboard about this. Probably offset of 25 or 50 for all gains should be good enough.

 

Peter


Edited by Peter in Reno, 01 December 2020 - 01:00 AM.


#25 leviathan

leviathan

    Viking 1

  • *****
  • Posts: 934
  • Joined: 29 Nov 2011
  • Loc: Azerbaijan

Posted 01 December 2020 - 06:10 AM

dghent, i think you're right -

 

a user named ricardo posted a sharpcap output using the different read modes, and it looks like 56 is the right gain for the 'High Gain Mode 16BIT' setting, whereas for the 'PhotoGraphic DSO 16BIT' setting the best gain would be 26.

https://www.cloudyni...ions/?p=9938875
 

This makes QHY461 even closer to QHY600.




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






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics