Table of mono cameras for EAA

[Martin Meredith]

I've been looking into the state of mono cameras for EAA under 2.5 megapixels and thought it might be useful to share what I've found. Let me know if you have any of the missing info, anything needs correcting, or if I've not included cameras that I should have, and I'll update the table. 

[updated 8th March 2019]

 

Notes

(1) These are all uncooled except for Atik 414EX

(2) All except the two Atik models have a 1.25" profile so can slide into a 1.25" eyepiece barrel to reach focus.

(3) The two FOV and resolution (res) columns are meant to be illustrative for two types of scopes often used in EAA (that I happen to have at hand). "Quat" is an 8" f/4 Quattro Newtonian reflector. "Borg" is a 77mm f4.3 refractor.

(4) QE is given along with peak wavelength

(5) Prices as of 7th March 2019

(6) mp = megapixels

(7) read noise in e- is a 'typical' value as obtained from manufacturers specs in the main; check the specs to see how it varies with gain etc. Likewise FWC = full well capacity varies with gain. 

Courtesy of astronomy.tools, here are indications of the FOV for most of these sensor/scope combinations (note that the same sensor is used in the Ultrastar and the two Atiks so only the Ultrastar is shown)

8" f/4 Quattro:

 

77mm f/4.3 Borg:

 

[Edit] I should probably add that much of this info is pulled off FLO's website!

[Chris]

Nicely put together!

[festoon]

Hi @Martin Meredith thanks for posting!!

The 414Ex has 1391 pixels x 1039 pixels and a gain of 0.28 e-/ADU, FWC of 18,000 e-, and read noise of 4e-

Only difference compared to the infinity (apart from cooling) is a gain on the infinity of 0.37 e-/ADU, FWC of 24,000 e- and read noise of 6e-

Atik cameras have T thread to 1.25" nosepiece adaptor included

If you can also put ASI385MC and Revolution R2 imager also onto this would be great

And just a point, if the camera has variable gain, then FWC and read noise very much depends on the gain setting used e.g.

[Martin Meredith]

Thanks Lockie & Festoon. I've updated the two Atik models with the data provided and clarified what I meant by the 1.25" profile (important for reaching focus on Newts).

The ASI385MC is colour I think? I'll look into the R2 but is that also colour? The reason for sticking to monochrome and sub 2.5m is to identify which cameras will work reasonably responsively with the first version of the EAA tool I'm hoping to release shortly. 

Martin

[festoon]

Sorry you are right ASI385MC is colour

R2 has both B&W mode and colour mode (its an analogue ccd)

Not sure I understand your point on 1.25" profile. Apologies if I am misunderstanding

e.g. ASI290 has 1.25" nose piece

And Atik has 1.25" nose piece

So both Atik's have 1.25" nose piece that can slide into 1.25" nose piece barrel.

[Martin Meredith]

My point is that the others can slide all the way down the eyepiece tube bringing the sensor to practically any focal plane whereas that wouldn't be possible with the Atiks. Its not a big deal but for Newts in particular it can sometimes be an issue. 

Ah, I see what you mean about the ASI 290. Actually, I was referring to the MM Mini versions which all have a 1.25" profile, at least for the bulk of their length. Like I say, its not a big deal so I'm happy to remove the note if it is confusing or contentious ?

 

[JamesF]

Based on the sensors in the listed cameras (ASI120 is 4.8mm x 3.6mm, btw, as far as I'm aware), you might also add:

Altair GPCAM1 and GPCAM2 mono AR0130-based, GPCAM2 and GPCAM3 290-based, Hypercam 174M

QHY 5-III 174M, QHY 5-III 290M, QHY5-II.  Perhaps also the QHY5L-II?

Touptek GPCMOS02000KMA, GPCMOS01200KMB, AG3CMOS02300KMC

One or two of the Mallincam models might fit the bill too, though they're probably not so easy to get in the UK.

James

[JamesF]

ZWO also give the dimensions of the 290MM sensor as 11.3x7.1mm and the 290MM sensor as 5.6x3.2mm.

James

[Martin Meredith]

Thanks James. I will do all of your suggestions that are relevant as a batch when I next get a little time in the coming days.

Martin

[JamesF]

That'll give me time to think of a few more

James

[JamesF]

Possibly cameras based on the ICX445 mono sensor could reasonably be in the list?  Atik GP, Point Grey Chameleon, Celestron Skyris 445M.  Maybe the Skyris 274M as well?

I think that's enough to be going on with

James

[umasscrew39]

Hi Martin

It has been a long time since we've chatted.  Seems even longer from the frequent and fun CN chats and challenges.  Anyway, nice chart you put together.  I would add another column on available software for live viewing and indicating the operating system(s) available for its use.  This has always been a critical and hot topic even when I was first learning.   The options are not great so they play a critical role in one's choice.

Bruce

[Howie_Oz]

Out of all those specs Martin, how does one work out the sensitivity for one vs another camera? It's often stated to use a "more sensitive" camera if on Alt Az for instance, due the shorter exposure times due field rotation effects. So sure with a less sensitive camera/sensor you can just stack more and eventually get nice bright(er) image. But you have to wait longer for the stack as you are doing more. But if doing outreach, or 'true' VA (not really caring too much for quality and just want to see lots of stuff during the night) then you want something more sensitive (so you don't have to stack much to 'see'). So, can you work out 'sensitivity' for those cameras from the info in the table?

[Martin Meredith]

Thanks Bruce. I agree re software/OSs. I don't think I could easily produce a complete list for these cams but I could make a start and let others chip in. My entire experience in EAA has been with SLL on MacOS.

Howie, what you've reminded me of is some code I wrote (based on an article) a while back concerning the ability to see (i.e. at a specified SNR) a point source object of a given magnitude, given a bunch of parameters, most of which are specified in this table. The only thing missing is local skyglow I think (and maybe dark current). I will dig into that and see if I can produce some illustrative values for comparison purposes, at least for the two scopes I have in the table. Once I've added the other suggested cameras I'll make the .csv available in case people want to add their own analyses!

BTW My purpose is wasn't really to do much in the way of companions but rather, as I mentioned above, simply trying to identify which cameras could work well with my new software. To this end, if anyone has any FITs from any of the listed cameras or the ones James mentioned, I'd be very happy to see them (don't bother with the Lodestar as I have 1000s...). In particularly I'd like to ensure that the live stacking algorithm works well for the different resolutions we see in the table.

Martin

[Martin Meredith]

So here Howie is a first stab at an answer. I've implemented the method in Herbert Raab's paper "Detecting and measuring faint sources with a CCD". (available http://www.astrometrica.at/Papers/PointSources.pdf ). This will churn out the peak SNR for a given sensor/scope setting based on the magnitude of the target point source, the sky magnitude, and a few other things. For example, asking the question "what SNR will I reach when looking for a mag 20 target using my 8" f/4 scope in mag18 skies with FWHM = 3" seeing using 10 x 30s subs", I get the following SNRs for the cameras listed above:

[tables updated 12th April 2019]

This shows the value of the larger pixels on the Lodestar (of course, there is a downside!).

These are the results if I use 30 x 10s subs (differences due to read noise):

If I improve my sky background by 1 Mag (i.e. SQM meter reading of 19) I can cut down the number of subs from 10 to 4 to achieve the same SNR (as you would expect):

I'm using the read noise values in the table above, which are kind of typical but can be lowered for the CMOS cameras at higher gains. If anyone wants to play around with (and improve) the script (Python), and optimise for the CMOS cameras, here it is:  

 peaksnr.py

It would be interesting to test out the predictions with some real data... For the Lodestar it seems about right to me in the sense that I can spot mag20 objects in well under 10 x 30s subs from my site, perhaps because I'm fortunate enough to get SQM 19.6-20 some of the time.

If I were upgrading I must admit the ASI 174 comes out very well here. Its a shame there is amp glow reported. I rarely feel the need for a wider FOV but on occasions being able to frame a larger OC or Abell cluster or dark nebula is appealing. But I do like the simplicity of the Lodestar and the lack of need to worry about gain settings.

Martin

 

 

 

 

[Howie_Oz]

Thank you Martin for all your work doing that. Interesting read too that PointSources pdf.

[festoon]

Hi @Martin Meredith

Thank you so much for posting the python code.

I wonder am I doing something wrong. Below are a couple of screen shots and I am trying to recreate your lodetar result for "what SNR will I reach when looking for a mag 20 target using my 8" f/4 scope in mag18 skies with FWHM = 3" seeing using 10 x 30s subs"

I've probably inputted something incorrectly!!!! But I cannot see what it is From your data I'd expect to see an output of 3.6432

[Martin Meredith]

Hi Festoon

Pixel refers to pixel scale in Raab's paper (at least that's my take on it), so should be 2.11 for the Lodestar at this focal length.

cheers

Martin

[festoon]

thanks @Martin Meredith that makes sense. I'm now getting numbers as you do.

I understand its not exactly important, but just for my understanding....For the obstruction - you have input 0.02 - does that correspond to 20mm? Would the f/4 quattro not be 68mm (or 0.068m)?

[Martin Meredith]

Ah yes, thanks for picking that up! Not sure where it came from (wishful thinking). It lowers all the SNRs a little bit (by about 0.2).

(I have a vague recollection that at the time I was exploring the impact of using the Lodestar instead of a secondary, but that is more like 0.03 rather than 0.02 in cross-section)

Martin

[Thalestris24]

Don't forget QHY versions! Also the 20MP QHY183M which can be binned and/or use a ROI and has a high QE - maybe 85%

Louise

[Martin Meredith]

Thanks Louise. I was going to include all these other brands but the info is very similar (not surprisingly) for the same sensors, with what appear to be minor tweaks. When I get time I will reorganise around sensors, removing manufacturer info but maybe mentioning manufacturers models grouped somehow.... 

I didn't include the 20MP (pixel?) versions as I'm restricting it to smaller pixel count cameras.

But 85% sounds wonderful in any universe (do you have a link for that figure?)

Martin

[Thalestris24]

2 minutes ago, Martin Meredith said:

Thanks Louise. I was going to include all these other brands but the info is very similar (not surprisingly) for the same sensors, with what appear to be minor tweaks. When I get time I will reorganise around sensors, removing manufacturer info but maybe mentioning manufacturers models grouped somehow.... 

I didn't include the 20MP (pixel?) versions as I'm restricting it to smaller pixel count cameras.

But 85% sounds wonderful in any universe (do you have a link for that figure?)

Martin

Oops - it's only 84% actually: https://www.qhyccd.com/index.php?m=content&c=index&a=show&catid=94&id=16

As I say, you can choose to bin 2x2  and/or to select a small region of interest which makes it quite versatile imho . So you can choose to just acquire 2MP if you like  It's not a cheap camera though. ZWO do a slightly cheaper, uncooled version https://astronomy-imaging-camera.com/product/asi183mm-mono

Louise

 

[festoon]

Hi @Martin Meredith...just been looking at the code again.

There is one discrepancy between the code, and what Robin has published when discussing optimum exposure times for imaging

In the python code you have the read noise of the stack as noise_read = n_subs * read_noise

If we look here (taken from https://forums.sharpcap.co.uk/viewtopic.php?t=456)

The contribution of the read noise is noise_read= sqrt(n_subs) * read noise

I'm not saying one is right and one is not. I'm not sure. What do you think?

[Martin Meredith]

A quick look suggests it is certainly possible that my formulation is wrong. The main effect of this change in terms of the numbers above will be to boost SNRs across the board in the N=30 subs case (with more boosting as N gets larger), which will favour the high read noise sensors (so the numbers for the Costar will be increased the most, and the ASA 290 the least). Does this agree with your findings?

I'll look in more detail tonight and update the tables.

cheers

Martin