Jump to content

Stargazers Lounge Uses Cookies

Like most websites, SGL uses cookies in order to deliver a secure, personalised service, to provide social media functions and to analyse our traffic. Continued use of SGL indicates your acceptance of our cookie policy.

sgl_imaging_challenge_banner_galaxies_winners.thumb.jpg.92ab73c3699031e815897184054b8f7c.jpg

Uranium235

ZWO ASI178MM Cool - DSO imaging review & experiments

Recommended Posts

 ZWO ASI178MM Cool Review:

This experiment/review came about after some discussion about whether images taken with small pixels and short focal lengths really do work out to be similar (in the resolution of detail) to images taken with larger pixels and a longer focal length - given that the aperture used was capable of the required sampling rate. So, with that question in mind - the ASI178MM Cool piqued my interest as to whether it could compete with my Atik 383L+ in terms of a viable imaging platform - not just for random pleasure imaging, but also for serious competition standard images that will stand up to the most demanding pixel peepers.

With a huge thank-you to Grant, I've been able to put it to the test and hopefully answer a few questions for those who might be considering whether tiny pixels really can produce a passable astro image. This is my first ever review, so I hope what I cant get across in words, would more than made up for by the images.


So, a few days later the camera arrived - and here is what I found in the box:


ASI178MM Cool (obviously!) in a nice protective soft case
1x USB3 cable (rather sturdy looking)
2x USB ribbon cables (short) for connection from ASI USB hub to guide camera (nice touch!)
1x Nosepiece
1x paper instruction manual
1x mini DVD with drivers, sofware and manual

To my surprise the existing nosepiece of the camera (that terminates in a female M42 thread) can be removed to reveal a short M42 male thread (that can just go straight into any 1.25" filter wheel). This means the backfocus distance to the CMOS chip is a mere 2.5mm according to the mechanical drawing. This leaves any potential user with tons of options for using filter wheels or an OAG etc.. not that the setup needs an OAG due to the short nature of the exposures we are going to be using.


Driver and software installation:

Fairly straightforward, just click on the installers found on the disk and windows will pick it up without issue. For reference, the imaging notebook I was using for the test runs on Windows XP. Sharpcap is on the disk, I installed it - though I didnt quite take an instant liking to the user interface (being a user of Artemis for 8 years). Therefore I went to the more familiar MaximDL for camera control and capture, setting the gain/offset function for the camera via the ASCOM control panel.

Being as this was a test camera, no firmware or driver updates were applied. Running straight out of the box.


Camera Spec:

Pixel size: 2.4 microns
Resolution: 3072x2048 (6.2 MP)
FWC: 15k
Bit depth: 14 bit
Peak QE: 84%
Weight: 400g

Fairly tempting figures there, 15k FWC isnt disasterous as its roughly 2/3rds of the 383L+, and enough pixels to make decent image size when viewed at 100% (in theory). Another thing of note is that while the sensor size is quite small, that can be offset to some degree by selecting optics with a short enough focal length. The advantage of the small sensor size is that it will not overly test the corrected field of your optics, which is good news for those who like to build mosaics.

The first test had to be with the Star71 as I was still awaiting delivery of the new f2 lens. However, the camera isnt a good match for the Star71 since its apeture and focal length meant that the camera would be oversampling what resolution the optics were capable of by a factor of 0.3" p/p. So with that in mind, any potential buyer will need to check whether their short FL instument has enough apeture to give a low enough dawes/rayleigh figure.

 

Test #1: ASI178MM + Star71

    Imaging resolution: 1.41" p/p
    Conditions: Poor
    Settings: Max dynamic range
    Subject: Rosette (Ha)
    Exposure: 4x300s

Not the best night to be out, with quite a lot of cloud dodging so only a few of the captured subs were useable. So, I concentrated on matching 3x300 from the ASI with one 1x900 from the Atik. One minor issue I had at this stage was difficulty calibrating out the starburst-type amp glow you get with this chip (more on that later).

Whether its a result of the oversampling, or the small pixels, or the conditions - but the stacked image looked a little "softer" than the single 383 sub. But given the state of the sky, it wasnt really a fair run to be honest, as nothing ever goes smoothly first time! So, the results were inconclusive.


Test 2 ASI178MM + Samyang 135mm @f2 (narrowband):

Now with the Samyang 135mm firmly in my posession, I constructed a compact and lightweight imaging rig - hilariously undermounted on the NEQ6.

20180317_163913.thumb.jpg.d224edab3be4a30d7fb3a1142ae089f1.jpg

I set it up this way - using the ZWO tripod collar on the lens (padded out with some felt), and the guidescope ring - rather than have the lens dangling off the camera. That is because the lens is at least twice the weight of the camera, and that would put too much strain on the bayonet connection. Remember, we are at f2 - so any compression of the weakest link (the bayonet) would be punished severely in regard to field flatness. Ideally, the camera needs to be supported as well, so there is no strain anywhere along the imaging train.

At 135mm f2, the resolution achieved was now in the much more familiar territory of 3.6" p/p - only slightly (theoretically) lower than what I achieve with the 383 + Star71. As a bonus, the sky was particulary good that night.

Conditions: Good
Settings: Unity gain
Subject: Rosette (Ha)
Exposure: 36x180s

A quick inspection of the data revealed  it was good enough for stacking, so the following night I took some calibration frames (dark, flat, bias, dark flat) that I could apply to the data using my usual method of DSS. However, as I found again - the starburst was still not calibrating out. Eventually, I settled on Maxim again to bail me out with its calibration routine - which to my relief calibrated out the starburst  and any residual bias noise perfectly! There is probably a way to make DSS calibrate correctly with that data, but time was of the essence for this test so I needed something that just "works".

From there I could stack as normal, then pop it into Photoshop for development.

The data responds slightly differently on inital processing, taking at least two more curves to bring out the initial detail (when compared to CCD data). What became apparent quite quickly was the amount of detail that had been revealed by the camera and lens combination - which was quite surprising, and better than some telescope based images Ive seen over the years.

It responded well to high pass and local contrast (as well as it could for just 90min!). The noise in the raw data is different to that of a CCD, not as consistent - but easily dealt with using a light touch of Ps NR (preserve details = 85%).

Rosette_DL_proc_80_web.jpg

As you can see, not bad for a small setup and short exposures!

Now lets compare that with 90min of data from the 383 + Star71 (6x900) - a setup costing almost three times more:

Group1_383L_proc.jpg

Now thats interesting, the ASI178 and 135mm while not delivering tighter stars - seems to a have caught almost just as much of the outlying nebulosity - despite the short exposures. Both images have noise, but it is different. As explained earlier, the CCD noise looks more distrubted, predicatable and repeatable - while the CMOS noise is slightly more random, especially between each calibration sub... which takes me to the next point.

Once the ASI data is stretched a little more to reach into the background, what became apparent is that there was a small issue with the dark frame calibration. What seems to have happened is that pixels have been rejected that were good (ie: not hot pixels), which left a smattering of black dots across the image. Not noticeable at first glance, but they become more apparent once the image is pixel peeped.

Nomally, with a CCD I would just ignore darks and stack the hot pixels out. But with CMOS technology, darks are an absolute must in order to remove the amp glow - so its a bit of a catch-22 situation. On further reading of topics regarding this, it seems that this (and some similar) CMOS sensors have variable dark frame noise from one session to the next due the the ambient temperatue - and the indirect way that particular sensor temperature is measured. Which may account for why the master dark didnt match the hot pixels present in the L frames - as my darks were taken indoors.

This effect could be reduced or eliminated with a combination of dithering, a greater weight of subs (both light and dark), and dark frames taken in the same session as the lights. Though that last point could prove to be the most inconvenient in you dont have an observatory - especially if you have to pack up in a hurry!

Why not just remove the hot pixels from the darks? Well, there is a good reason for that - if you attempt to remove the hot pixels, it drastically changes the data and therefore it is no longer a well matched dark and renders it useless.

However, this issue was largely countered in the next test! 

 

Test #3: M81 + M82, M106

After some discussion with a fellow SGL member in regard to calibration files, I decided to set the camera up with a UV/IR filer and see if this tiny rig could bag any galaxies. Now, in broadand imaging - this camera is completely different being as we are not having to separate a weak target signal from the camera noise (as would be the case in narrowband). So therefore any camera noise is quite quickly swamped, but that does not exempt you from full calibration! If you want to get into faint background signals - you need to calibrate.... no excuses!

This time I took a set of darks at the end of the session (more on that in a moment), then followed it up with flats and bias. The reason why I left it until the end is becuase the ambient temperature of the sensor and electronics will be at its lowest point - and therefore less hot pixels - with only the most stubborn, persistent ones remaining.

The logic behind this is that if the camera is ran as cold as possible, in a cool environment - then only "repeatable" hot pixels remain. Once that master dark is subtracted from the image data, the only things that should remain are the more "random" hot pixel, and hots that have been generated by a higher ambient temperature (at the start of the session). This way, any hot pixels that remain will be stacked out, and the "black dot" issue will be much reduced (or eliminated).

The following two images are comprised of 40x120 exposures, fully calibrated:

27726365788_ad060c44bb_k.jpg&key=46cc669

40704190485_cc2d7417a2_k.jpg

For just 1hr 20min each, thats actually quite good! Though I would think it would require at least double that amount of data to take more noise out of the image (I like clean data!).


But what was more surprising, was that when I heavily stretch the bodes image - traces of the IFN were clearly visible, very unexpected indeed! I do however have to add that the IFN traces were only visible because of full calibration, as without it the faint signals would be buried beneath fixed pattern noise and/or vignetting.

Group2_L_80_IFN.jpg

This has led me to the conclusion that for narrowband, this camera needs to be set up differenly (perhaps lowest read noise setting) in order to make the most of the weak signals you are looking to detect. But for broadband, just set it to max dynamic range and give it some fast optics.

 

Test #4: M101 + M51

Now, time to see what happens when you give this camera a telescope with a larger aperture (and hence more resolving power). The dawes limit for 130mm is 0.89" p/p, but with the 130 running a 0.9x corrector we will be oversampling at roughly 0.85" p/p.

No real time to test the collimation of the 130pds, and quickly set up using the bayonet connectors I used for the Samyang 135 so the spacing was roughly 2.5mm too short. However, I was quite surprised at the results.


36x120s (UV/IR filter only):

28108433298_276ae516d8_k.jpg&key=9377ad0

With more time on target it would improve sufficiently in order to reduce noise, and apply better sharpening and contrast enhancement.

 

M51 - 20x120s (UV/IR filter only)

Group2_L_80_web.thumb.jpg.a7fa8aee0d7195c2f28bb9487e281128.jpg

 

Quite a surprising result given the short exposure run, with some of the fainter outlying dust starting to show. However, the "black dots" issue returned in this image - but being as it was in the background (and not on the target) I was able cosmetically correct it by using photoshop to sample, then paste in the correct background level (with added fake noise) using blend mode lighten. This technique can be learnt here:

http://bf-astro.com/backgndRepair.htm

So, as you can see - tiny pixels can turn a rather modest telescope like the 130pds into a fairly good galaxy hoover!

 

Test #6 - Needle Galaxy

Now pushing it a step further to see what happens when I use 100+ subs on a target. Well, so unexpected was the result that it made Flickr "Explore" - which doesnt happen all that often! So its no mean feat.

Also, a slight change in the imaging train as I decided to use the Baader MkIII corrector for better stars.


120 x 120s

41872415692_f00dd79a06_k.jpg

 

Also, its picked up a fair few background galaxies to boot. I'd go as far as to say its my best ever result on this particular target. I think perhaps its a case of "the more you use the camera, the better you get!".


Conclusion:

So, could this CMOS based camera do what CCD has done for me for the last 8 years? Well... its a close one to call, but at the moment - CCD still has the edge in narrowband imaging (my most used imaging mode) - mostly becuase the noise enitrely predictable and slow changing,  the exposure length is unavoidably long, and I like big sensors. However, this review is purely based on the ability of the camera to image deep-sky objects. So when you take into account its flexibility as a planetary or solar imaging camera - then you have something quite rare... a jack-of-all-trades camera which wont break the bank.

Indeed, if I were to point my 383L+ at the Moon or try Solar imaging - it would turn to me as if to say "What on Earth are you doing?!, Dont be daft!"

Therefore, while not quite being a direct replacement of CCD technology for the (very) serious imager - it is however an excellent introduction to mono imaging - a place that (in my opinion) used to belong to the Atik 314L+. And made all the more tempting by the fact that this camera is half the (new) price of what would be considered an entry level CCD.

So, if you're thinking of stepping up from DSLR, but cant afford or justify the best part of 2k for a CCD camera setup, then this camera is a good way forward as long as its paired with short focal length optics that suit the pixel size (making sure to check the dawes/rayleigh limit for your optics first). A good match for this camera would be a fast lens between 135-200mm in focal length (eg: Samyang 135 or Canon L series 200mm) for widefield, or for sub arc-second imaging - the Skywatcher 130pds for galaxies (barring M31 & M33).

https://astronomy.tools/calculators/telescope_capabilities

Pros:     

    Makes for a lightweight, compact imaging system
    Sensitive
    Places fewer demands on the mount & guiding
    Quick to cool
    Fairly clean data (when properly calibrated)
    Built in USB hub
    Easy adjustment of camera settings
    At a price which is easily scalable (twin or triple shooter)
    Multiple applications (DSO, Solar and planetary)
    Ideal for the mobile imager that requires a lightweight, compact setup.

Cons:  (though minor)

    Requires suitable focal length and/or aperture
    Tricky dark frame calibration (since resolved - see test #3)
    High HDD storage & network (on data transfer) useage
    Paper instructions hard to read (even with glasses on)
 
CMOS technology is definitely heading in the right direction - and its very, very close to catching CCD. Ideally, the next step for this technology should be the implementation of sCMOS, where a lot of the electronics have been taken off-sensor (amps etc..) - that would remove all amp glow and would (going by whats on paper) spell the end of the road for CCD. However, these sensors are still very new, and very expensive. Perhaps in 5-10 years time, one of them will end up inside an affordable astro imaging camera.

As to whether the issue of the detail given by small pixels and short focal lengths being the same as a camera with larger pixels - on a telescope/lens with more focal length? Well... its close - the setup with a longer focal length and larger pixels still has the upper hand. But perhaps given a similar level of exposure to what you would put in with a CCD, that would provide a depth of data good enough to get in there quite agressively with whatever processing tools you have to hand.

Lastly, while it might not be so fair as to compare one setup to another costing three times more - its definitely worth consideration for those on a tight budget or who already have the camera for planetary/solar, and fancy a go at some DSO work. 

 

 

  • Like 5
  • Thanks 1

Share this post


Link to post
Share on other sites

A very interesting read Rob, great job. I'm happy to see that CMOS technology is improving rapidly for the field of astrophotography and hope that the perfect CMOS camera will be available by the time I am forced to replace my beloved QSI 683 CCD. As for the images - I just can't help but think that the Rosette shot with the Star 71 and 383 is the best of the lot. I'm probably just biased, but it has a clarity that the CMOS can't match, yet. Thank goodness, at that price!

  • Like 1

Share this post


Link to post
Share on other sites
25 minutes ago, PhotoGav said:

A very interesting read Rob, great job. I'm happy to see that CMOS technology is improving rapidly for the field of astrophotography and hope that the perfect CMOS camera will be available by the time I am forced to replace my beloved QSI 683 CCD. As for the images - I just can't help but think that the Rosette shot with the Star 71 and 383 is the best of the lot. I'm probably just biased, but it has a clarity that the CMOS can't match, yet. Thank goodness, at that price!

Thanks :) I think the S71/383L image had the edge due to the focal length and larger pixels. Though if given a 2nd stab at the rosette I could probably improve upon the results being as that image was pretty much the first image I got out of it. 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.

×

Important Information

By using this site, you agree to our Terms of Use.