The left image is M42 from a 1 hour stack on the Canon EOS 6D + 200 mm F/2.8 lens at F/4. The right image is a 20 minute stack from the ATIK 490 + C8 HyperStar.

The obvious difference is the near total lack of red sensitivity where it counts, in the Canon 6D. But I do find it quite interesting that the level of detail is very similar in both images. The same faint stars are seen in both images, and the wispy structure is nearly the same. You can see the lack of deep red sensitivity in the Running Man nebula near the top of the images, where the Canon shows nearly black (i.e., no signal) while the ATIK shows deep red content in the man.

The 200 mm Canon lens at F/4 ought to be a 50 mm aperture. The C8 is a 203 mm aperture. So there should be a 16x difference, or 3 magnitudes in the star images. The sensitivity to extended regions, however, is dictated purely by the F-ratio of the cameras: the C8 HyperStar is F/2.1 and the Canon 200 mm lens was run at F/4, or just about 1 F-stop slower. Hence the same level of detail ought to be seen by the Canon system in about twice the exposure. We have about 3x exposure in the Canon so the images seem reasonable from that viewpoint. But the star images are a total surprise, where aperture should rule.

Things are not quite that simple, however. The pixel size on the Canon 6D is 6.5 microns, while the ATIK 490 has 3.69 micron pixels. That's an area ratio of about 3:1 in favor of the Canon for light sensitivity to extended regions. So even though its optical train was half as fast, it gains in sensitivity by having bigger pixels. It turns out that pixel size is a huge factor in what you pick up, not just a concern for sampling density.

The Canon 6D is grossly undersampled, even in the skies over my city back yard. Its pixels are 6.9 arcsec with the 200 mm Canon lens. The ATIK 490 is probably close to critical sampling with a resolution of about 1.8 arcsec / pixel on the C8 HyperStar.

Conclusions? Lack of red sensitivity is a serious concern for astrophotography with a stock Canon EOS 6D. But despite its spatial undersampling, it produces very fine images. And its pixel size makes it very sensitive to extended regions. The ATIK 490 is also a fine camera, it shows good red sensitivity, but its pixels are probably a bit too small, and its frame size corresponds to only about 1/3 of the Canon sensor.

I have measured the characteristics of both sensors using bias frames, dark frames, and some sky flats. The Canon 6D is a CMOS, unregulated in temperature. The ATIK 490 is a CCD with regulated temperature control. Both sensors show about the same full-well depth when the Canon is run at ISO 400. The gains are comparable, as is the read noise. And the dark currents in both sensors is almost unmeasurable. So they are very closely matched in terms of light sensitivity.

However, for photometric measurements, the Canon CMOS exhibits linear behavior all the way up to full-well. The ATIK CCD has a soft knee somewhere around 90% of full-well, due to its antiblooming circuitry, and so is linear only below that level.

The Canon 6D measures with a system gain of about 1.4 e-/ADU, while the ATIK 490 has a system gain of around 0.45 e-/ADU. Readout noise in the Canon is around 5-6 e-, and in the ATIK about 4 e-. Both have a full well capacity such that, in terms of readout noise steps, they both have dynamic ranges of around 30-35 dB. [... this is another way of saying that both sensors are basically 14-bit sensors, even though the ATIK uses a 16-bit ADU. And of those 14-bits, only fewer than 12 of them really count - for both sensors.]

(For some strange reason people like to double that number to say 60-70 dB. I could understand that desire if these sensors were power generators. They do produce voltages, and so the tendency is to use 20 * Log10 for dB measure. But the sensors are power accumulators and merely represent the stored energy from photons in terms of voltage. So I would argue that one should just use 10 * Log10 for dB measure. There are about 3000 steps of readout noise to full-well capacity, hence a dynamic range of 35 dB.

Maybe, if we both agree to talk in terms of stellar magnitude scales, we can agree on a dynamic range of 8.7 mag. )