Does anyone know the thermal noise characteristics of Canon EOS Ra? - Forums

Topic starter

Jul 11, 2020

Pretty much the title says it all.

The reason for the question is that, after a very helpful discussion on another Astrobin technical forum, I have purchased an Esprit 100 reflector as my step up from DSLR imaging. While I will retain my unmodded Canon 6D MkII for a while, I am thinking about the best imager for it.

I do love wide-fields, and the best sq deg/$ if I want to make an improvement to my H alpha sensitivity is the Canon Ra. The full frame of the Ra would appear to fill the 40mm corrected field of the Esprit 100 extremely well. Almost as if they had built it with the EoS Ra in mind ;-).

However, after trawling the technical pages and the mostly glowing reviews of this camera, I can't find the spec that is the limiting factor in the SNR acheivable from this device ie the thermal noise/dark current.

Here is my assumptions to date, noting I observe under Bortle class 3, for the SNR for a surface brightness 22 mag/arcsec**2 with 550mm focal length and 120sec subs. [For calculations Vega is around 3500 photons at zenith]

Canon RA/Esprit 100 : Pixel size = 2.0arcsec, QE = 0.52, Read Noise = 2.2e Dark Noise = 0.3-0.6/e/s/pix??? (for temps 6-11C, assuming Canon 6D MK II readnoise)

SNR in 1 hour: 5.8 to 7.2.

Even at the lower temperature, with this assumption I would be dominated by thermal noise (50% higher signal than sky) at the higher one it more than 3 times sky. But this is highly dependant on the dark current assumed. Hence my question.

In contrast a roughly equivalent (in $) colour astro camera gives

ASI2600MC: Pixel size =1.4arcsec. QE=0.8, Read noise=1.5e, Dark Noise =0.002 (or less) e/s/px

SNR in 1 hour : 8.9

So the ASI2600MC is effectively 2 as fast as the Canon Ra, with 2x smaller pixels but 2x smaller field.

Now, I get that the Canon Ra might be simpler to use (and have more uses) than a ZWO camera. Certainly the former is important to me, the latter less so.

For me it will be the SNR that clinches it I suspect. At the moment I am leaning towards the Canon Ra, if the dark current is as good (or even better) than my assumptions.

Sorry for the long post, but I wanted to give you the calculations behind my quandary. If anyone has experience of the Canon Ra (or it my calculations are wrong) it would be really great to hear form you.

Many thanks in advance

Brian

Arun H

Frequent contributor

Jul 12, 2020 · Edited: Jul 12, 2020

Bill Claff’s website can you give you the actual read noise for this camera. Just look up the 5D Mark IV. The sensor is basically identical. The numbers you have listed seem to be for the dark current as opposed to the noise from dark current. The noise from dark current is simply the square root of the mean dark current. The mean dark current can be subtracted from the image just like the mean sky background. What you’re concerned with is the noise from dark current. This is the number that needs to be compared to the read noise. PixInsights Basic CCD Parameters script should give you an idea of the dark current numbers.

Helpful Insightful Concise

Arun H

Frequent contributor

Jul 12, 2020

Also, your assumptions of dark current seem to be high. Roger Clark’s website has dark current numbers for some cameras. The older generation 7D Mark II has a dark current of approximately 0.03 e/px/sec at 15 C. I’d expect later generation cameras to be similar or better. So for a two minute exposure, the overall dark current would be 3.6 e/px and the noise from that would be 1.9 e/px.

Well Written Helpful Insightful Concise

Brian Boyle

Topic starter

Jul 12, 2020

Arun, Thanks for the helpful advice. Good to have the read noise for the sensor.

I made a mistake when I referred to the number above as dark noise - I should have said dark current (at least my units were dimensionally correct). I have been square rooting that number in my calculation.

Also, I think we must be looking at different Roger Clark sites! When I look at the dark current plot for various sensors, I see that the Canon 7D has a dark current of 0.6 e/pix/sec - very similar to the Canon 6D - about 20 times higher than you quote. I really hope you are right and I am getting something wrong here.

Well Written Insightful Respectful Concise

Arun H

Frequent contributor

Jul 12, 2020

This is the page I am referring to:
https://clarkvision.com/articles/dark-current-suppression-technology/
The graph on that page has both the 7D (blue line) and the 7D Mark II (orange line). The 7D Mark II has much lower dark current than the original 7D. And the 5D Mark IV is a later camera than the 7D Mark II. I use both the 7D Mark II and 5D Mark IV for Astro work.

Brian Boyle

Topic starter

Jul 12, 2020

Arun, Thanks so much for pointing me to this page. I was looking at an older version of this plot.

This is remarkable. It effectively means that sky noise is the limiting factor even for non-cooled sensors in wide-field applications.

Well Written Respectful Supportive

dkamen

Jul 13, 2020

I've been saying this constantly. The only time dark current was ever an issue for me was last summer, in a bottle 1 sky with 30 degrees at night, pointing towards M101 in Ursa Major where nebulosity and star fields are very sparse. And even that was with the Nikon D7500 which has unusually high dark current for it's generation (still nothing of practical significance when shooting closer to the Milky Way band), and mostly went away with a few dark frames. (however in the end I had no presentable image because the vast majority of my subs had tracking issues)

This is not to say cooling doesn't help but it just doesn't give you the value for money that it used to 10 or 15 years ago. Dark current suppression is the key and every modern sensor features it.

Cheers,
Dimitris

Helpful Insightful Engaging

Brian Boyle

Topic starter

Jul 13, 2020

Dimitri,
sorry for being slow on the uptake - but I had been seduced by all this talk of cooled sensors.

Living in Central Otago, New Zealand, I have natural cooling anyway. Tonight is forecast for -5C.

As I understand it, the QE for the dedicated astrocameras is a little higher - so it might be worth it for that alone.

I can also see the point in a mono camera as well, and that might be the direction I eventually go when I tire of the DSLR approach ….

Well Written Respectful Engaging

dkamen

Jul 13, 2020

Hi Brian, not sure what you are apologising for. Perhaps my tone came out slightly on the aggressive side, I assure you there was no such intention whatsoever. I got a bit carried away and spoke like I would to another Greek, I am afraid our baseline polite tone sounds very direct and aggressive to an English speaker

So back to the subject, QE is definitely a factor but in my opinion dedicated astronomy cameras are mostly worth it for ease of use. Sure it needs a computer, but if you are guiding you have a computer anyway. The astrocam will plug more precisely to your telescope, will be lighter and more compact, will allow you to control every little parameter of the capture from your computer, will send the images immediately and can do things like binning, Region of Interest etc. I don't know if Canon cameras can do all those things but Nikon cameras generally cannot. Software can do 2x2 binning and ROI of course but you still get to download the 25MB raw file which is an overkill when e.g. focusing. If you are using a planetarium, it is possible that the system will take and solve 15 pics until it centers your target, these stay in your camera's SD card by default.

So the DSLR is its own computer that mostly gets in your way, while the astrocam is more like a simple passive sensor designed to work with your computer. And like I told you in the other thread, astrocam sensors tend to be smaller because realistically DSLR sensors are not so easy to illuminate properly without aberrations. I don't think the image from my VMC110L would look very good on a crop or full frame sensor, but the tiny sensor of the 178MC works very well with it.

In terms of image quality and if we are talking about wide field, OSC and broadband, I think there is no question even a modest DSLR beats any astrocamera within the same price range (including cooled ones) hands down. I think will find that even your Canon 6D MKII has a lot to give when combined with the Esprit, especially if you follow Dr Clark's workflow here.

Cheers,
Dimitris

Helpful Insightful Respectful Engaging

Arun H

Frequent contributor

Jul 13, 2020

I can only say that I’ve been very happy with the results from my 5D Mark IV and 7D Mark II. Both unmodified. I am considering getting a modified 6D for widefield nebula work.

Well Written

Brian Boyle

Topic starter

Jul 13, 2020 · Edited: Jul 13, 2020

Dimitris,

Your tone was fine - and your message was very reassuring that I hadn't jumped to the wrong conclusion. It is alway good to have an experienced astrophotogra her like yourself validate my newbie conclusions.

Its just that I should have probably realised your point a lot earlier in my journey, rather than be distracted by the hype surrounding the importance of cooling - at least for wide-field applications. This might have been true a few years ago, but the latest generation of chips in "normal" DSLRs appear to have put paid to thermal noise being the limiting factor.

Clearly there is a place for dedicated astrocameras, the QE is substantially greater (80% v 50%), mono imaging with filters and small pixels. If I do graduate to specialised astrocameras, I think that is now the way I would go; 3um pixel, mono camera with LRGB and the Ha/OIII/SII set.

But in the short (but wide-field) term, I will stick with a DSLR - both unmodded and modded - and the best bang-for-buck in areal coverage.

Thanks again for reassuring me that I am on the right path (or, at least, not the totally wrong one).

Brian

PS In my career as a professional astronomer, I worked with a number of Greeks. I loved their directness and helpfulness. So I know exactly where you are going from.

Helpful Insightful Respectful Engaging

Brian Boyle

Topic starter

Jul 13, 2020 · Edited: Jul 13, 2020

On the subject of the data reduction flow, I am now fascinated by Dr. Clark's prescription. Having just learned the PixInsight flow (very similar to reducing CCDs in the 1980s, back when I was a postdoc), and learned what I now have about modern sensors, I can certainly see the thinking behind it.

I am not sure, if I want to leap into a whole new area of data reduction, so soon after learning one - but I am very insterested in knowing whether RawTherapee DeBayering can be used as part of a PixInsight work flow.

dkamen

Jul 13, 2020

Hi Brian,

Yes it can. Absolutely. You basically do raw file development + calibration with Rawtherapee in the manner described by Dr Clark, export 16 or 32 bit TIF files and these are what you integrate with PixInsight (or AstroPixelProcessor or Siril). Rawtherapee has excellent capabilities. It has the most advanced debayering and denoising algorithms, it can take care of hot pixels, correct vignetting, make sure stars are not blown, use tone curves that are specifically designed for your camera and as of the latest version sharpen your exposure. It can remove gradients (rather painfully) and can also use darks and flats if you absolutely can't live without them. Basically it is the most advanced preprocessor in existence.

The most painful step is aligning the zero point for R,G,B in every image using curves in Rawtherapee. Basically you need to make sure all curves (within each image and among all images) start from the same point to the left of the histogram and stay on the left 1/3rd.

He says you should avoid any kind of normalization because it destroys the colors and I do agree as far as Pixinsight is concerned. I find that AstroPixelProcessor has far superior normalization algorithms (local and global) and the result is improved significantly if you turn them on.

It is a standard part of my workflow when shooting widefield with the Nikon and the one thing I miss the most when shooting with the ZWO. In that picture I used linear processing (revision A) and then Roger Clark's workflow (revision B), I believe the difference is very pronounced.

Cheers,
Dimitris

Helpful

Brian Boyle

Topic starter

Jul 13, 2020

Thanks so much Dimitris .

I will try this for next processing and compare to my PixInsight results. The difference it made to your image is dramatic.