Local galaxies. big telescope versus small telescope
Posted: Wed Feb 14, 2024 5:51 pm
A brief gap in the clouds - but not so transparent skies - permitted me to try the VX12 (12 inchF4 Newt) and compare it's performance to that of the 4.2 inch F 5.6 refractor versuis the same galaxy target, IC342 (last week's post) -- and to see how its performance fits theory. Also being big and heavy and a considerable task to collimate and set up it was interesting to see just how well it earns its keep.
Using an ASI294 MC camera with 4.63 u pixels the OO VX12 sets an image scale of AS = 0.81 arcsec/ pixel versus the WO Zenith star refractor at 1.54 arcsec/ pixel. The relative 'etendue' of the former therefore works out about twice that of the latter (i.e, the ratio of d^2* AS^2). So the expectation is that the 12 inch will fill the pixels with signal twice as fast as the refractor and that under similar skies and for the same total time of exposure it should deliver an SNR about 1.41 X better than the refractor. So a modest but real 2X imaging time advantage to the 12 inch.
The question of resolution is much more nuanced though. Average seeing conditions of 2-3 arcsec set a limit well above the Dawes limit of either telescope (~1.1 and 0.4 arcsec) and you could argue that -under normal conditions - 1.5 arcsec/ pixel will be adequate (sampling the seeing at about 1.7x ) to find all the detail that normal seeing allows while the 0.8 arcsec/ pixel provided by the 12 inch is just empty oversampling overkill? This sort of argument is often made and many astro websites calculate ideal image scales in this way.
However routine adoption of deconvolution as an early step in processing - as part of the PI toolset but more particularly via the excellent RC BlurXt software would seem to change the normal rules. Oversampling versus the seeing is a necessity for deconvolution to work well. So with sampling up at 1.5 AS / pixel applying deconvolution will not deliver to a resolution better than 2.2 or so - which- depending on the seeing - may not be much improved over the original resolution. At 0.81 AS scale however deconvolution stands to find levels of resolution down to perhaps 1.4 or so - and significantly better than the seeing.
Of course - nothing is free - and deconvolution also only works well in regions of high SNR. So just as with dithering and drizzle there is a price to pay in terms of imaging time.
So - overall - applying the same processing steps - drizzle X1 CFA integration, BlurXt , SPCC - and imaging for similarish times (actually one hour for the refractor (under better transparency skies) and nearer 2h for the 12 inch - (poor skies low SNR leading to visible patterning noise in the background) yielded the results below. They seem to broadly fit expectations. The final deconvolved VX12 image of IC342 is a lot (~1.7X) sharper than the refractor image about equally bright and actually better coloured. So particularly for these smaller objects the VX12 earns its crust -- while the refractor does for wider skyscapes.
VX12 telescope. SW F4 aplanatic coma corrector, CEM70 mount, ASI 294MC, PDS2 guiding IC342 image ~ 178 x 40s at unity gain, NGC2403 image 34 x 70s at unity gain.
Using an ASI294 MC camera with 4.63 u pixels the OO VX12 sets an image scale of AS = 0.81 arcsec/ pixel versus the WO Zenith star refractor at 1.54 arcsec/ pixel. The relative 'etendue' of the former therefore works out about twice that of the latter (i.e, the ratio of d^2* AS^2). So the expectation is that the 12 inch will fill the pixels with signal twice as fast as the refractor and that under similar skies and for the same total time of exposure it should deliver an SNR about 1.41 X better than the refractor. So a modest but real 2X imaging time advantage to the 12 inch.
The question of resolution is much more nuanced though. Average seeing conditions of 2-3 arcsec set a limit well above the Dawes limit of either telescope (~1.1 and 0.4 arcsec) and you could argue that -under normal conditions - 1.5 arcsec/ pixel will be adequate (sampling the seeing at about 1.7x ) to find all the detail that normal seeing allows while the 0.8 arcsec/ pixel provided by the 12 inch is just empty oversampling overkill? This sort of argument is often made and many astro websites calculate ideal image scales in this way.
However routine adoption of deconvolution as an early step in processing - as part of the PI toolset but more particularly via the excellent RC BlurXt software would seem to change the normal rules. Oversampling versus the seeing is a necessity for deconvolution to work well. So with sampling up at 1.5 AS / pixel applying deconvolution will not deliver to a resolution better than 2.2 or so - which- depending on the seeing - may not be much improved over the original resolution. At 0.81 AS scale however deconvolution stands to find levels of resolution down to perhaps 1.4 or so - and significantly better than the seeing.
Of course - nothing is free - and deconvolution also only works well in regions of high SNR. So just as with dithering and drizzle there is a price to pay in terms of imaging time.
So - overall - applying the same processing steps - drizzle X1 CFA integration, BlurXt , SPCC - and imaging for similarish times (actually one hour for the refractor (under better transparency skies) and nearer 2h for the 12 inch - (poor skies low SNR leading to visible patterning noise in the background) yielded the results below. They seem to broadly fit expectations. The final deconvolved VX12 image of IC342 is a lot (~1.7X) sharper than the refractor image about equally bright and actually better coloured. So particularly for these smaller objects the VX12 earns its crust -- while the refractor does for wider skyscapes.
VX12 telescope. SW F4 aplanatic coma corrector, CEM70 mount, ASI 294MC, PDS2 guiding IC342 image ~ 178 x 40s at unity gain, NGC2403 image 34 x 70s at unity gain.
just a tad heavier than the VX12 which is already at my limit for lifting. I'm terrified of dropping it every time I hoik it onto the mount. Joking aside though it does raise the question of at what point does the benefit of increasing the aperture and decreasing the image scale start to run out? I have imaged at 0.4 AS on occasions (by using smaller pixels) but for most objects I don't think that the slight increase in attainable resolution relative to 0.8 was worth the loss of SNR. Probably would be fine on a VX24 though that would exactly make good that loss. best wishes Tim